[Refactor] Migrate Voxtral TTS XPU stage config to --deploy-config (#5467)

Signed-off-by: Joshna Medisetty <joshna.medisetty@intel.com>

.buildkite/amd/scripts/bootstrap-amd-omni.sh

@@ -6,6 +6,8 @@
 
 set -euo pipefail
 
+source .buildkite/common/scripts/resolve_skip_ci.sh
+
 if [[ -z "${RUN_ALL:-}" ]]; then
     RUN_ALL=0
 fi
@@ -22,10 +24,6 @@ if [[ -z "${AMD_MIRROR_HW:-}" ]]; then
     AMD_MIRROR_HW="amdproduction"
 fi
 
-if [[ -z "${DOCS_ONLY_DISABLE:-}" ]]; then
-    DOCS_ONLY_DISABLE=0
-fi
-
 fail_fast() {
     DISABLE_LABEL="ci-no-fail-fast"
     # If BUILDKITE_PULL_REQUEST != "false", then we check the PR labels using curl and jq
@@ -72,14 +70,14 @@ upload_pipeline() {
     fi
 
     # Use local template file for vllm-omni
-    cp .buildkite/test-template-amd-omni.j2 .buildkite/test-template.j2
+    cp .buildkite/amd/test-template-amd-omni.j2 .buildkite/amd/test-template.j2
 
 
     # (WIP) Use pipeline generator instead of jinja template
-    if [ -e ".buildkite/pipeline_generator/pipeline_generator.py" ]; then
+    if [ -e ".buildkite/amd/pipeline_generator/pipeline_generator.py" ]; then
         python -m pip install click pydantic
-        python .buildkite/pipeline_generator/pipeline_generator.py --run_all=$RUN_ALL --list_file_diff="$LIST_FILE_DIFF" --nightly="$NIGHTLY" --mirror_hw="$AMD_MIRROR_HW"
-        buildkite-agent pipeline upload .buildkite/pipeline.yaml
+        python .buildkite/amd/pipeline_generator/pipeline_generator.py --run_all=$RUN_ALL --list_file_diff="$LIST_FILE_DIFF" --nightly="$NIGHTLY" --mirror_hw="$AMD_MIRROR_HW"
+        buildkite-agent pipeline upload .buildkite/amd/pipeline.yaml
         exit 0
     fi
     echo "List file diff: $LIST_FILE_DIFF"
@@ -89,13 +87,29 @@ upload_pipeline() {
 
     FAIL_FAST=$(fail_fast)
 
-    cd .buildkite
+    cd .buildkite/amd
 
-    # Select test definition file: merge suite for main, ready suite for PRs
-    if [[ $BUILDKITE_BRANCH == "main" ]]; then
+    # Select test definition file: merge suite for main, ready suite for PRs.
+    # For debugging, DEBUG_TEST_YAML can override the selection — accepts
+    # "merge" or "ready" (case-insensitive).
+    if [[ -n "${DEBUG_TEST_YAML:-}" ]]; then
+        case "${DEBUG_TEST_YAML,,}" in
+            merge)
+                TEST_YAML="test-amd-merge.yml"
+                ;;
+            ready)
+                TEST_YAML="test-amd-ready.yml"
+                ;;
+            *)
+                echo "ERROR: DEBUG_TEST_YAML must be 'merge' or 'ready', got '$DEBUG_TEST_YAML'" >&2
+                exit 1
+                ;;
+        esac
+        echo "DEBUG_TEST_YAML override: using $TEST_YAML"
+    elif [[ $BUILDKITE_BRANCH == "main" ]]; then
         TEST_YAML="test-amd-merge.yml"
     else
-        TEST_YAML="test-amd-ready.yaml"
+        TEST_YAML="test-amd-ready.yml"
     fi
 
     (
@@ -136,40 +150,13 @@ if [[ $BUILDKITE_BRANCH == "main" ]]; then
     file_diff=$(get_diff_main)
 fi
 
-# ----------------------------------------------------------------------
-# Early exit start: skip pipeline if conditions are met
-# ----------------------------------------------------------------------
-
-# skip pipeline if all changed files are under docs/
-if [[ "${DOCS_ONLY_DISABLE}" != "1" ]]; then
-  if [[ -n "${file_diff:-}" ]]; then
-    docs_only=1
-    # Robust iteration over newline-separated file_diff
-    while IFS= read -r f; do
-      [[ -z "$f" ]] && continue
-      # **Policy:** only skip if *every* path starts with docs/
-      if [[ "$f" != docs/* ]]; then
-        docs_only=0
-        break
-      fi
-    done < <(printf '%s\n' "$file_diff" | tr ' ' '\n' | tr -d '\r')
-
-    if [[ "$docs_only" -eq 1 ]]; then
-      buildkite-agent annotate ":memo: CI skipped — docs/** only changes detected
-
-\`\`\`
-${file_diff}
-\`\`\`" --style "info" || true
-      echo "[docs-only] All changes are under docs/. Exiting before pipeline upload."
-      exit 0
-    fi
-  fi
+# Early exit: unified skip-ci (docs / skip marks) and CI-yaml-only level targeting.
+if [[ $BUILDKITE_BRANCH == "main" ]]; then
+    gate_bootstrap_ci amd l3
+else
+    gate_bootstrap_ci amd l2
 fi
 
-# ----------------------------------------------------------------------
-# Early exit end
-# ----------------------------------------------------------------------
-
 patterns=(
     "docker/Dockerfile"
     "CMakeLists.txt"

.buildkite/amd/scripts/run-amd-test.sh

@@ -11,15 +11,6 @@ set -o pipefail
 export PYTHONPATH=".."
 
 # Print ROCm version
-echo "--- Confirming Clean Initial State"
-while true; do
-        sleep 3
-        if grep -q clean /opt/amdgpu/etc/gpu_state; then
-                echo "GPUs state is \"clean\""
-                break
-        fi
-done
-
 echo "--- ROCm info"
 rocminfo
 
@@ -51,18 +42,6 @@ cleanup_docker() {
 # Call the cleanup docker function
 cleanup_docker
 
-echo "--- Resetting GPUs"
-
-echo "reset" > /opt/amdgpu/etc/gpu_state
-
-while true; do
-        sleep 3
-        if grep -q clean /opt/amdgpu/etc/gpu_state; then
-                echo "GPUs state is \"clean\""
-                break
-        fi
-done
-
 echo "--- Pulling container"
 ## Temporary change to use AMD Docker Hub to store the vllm-omni image
 # to bypass the rate limit issue with ECR Public Gallery.
@@ -77,7 +56,7 @@ container_name="rocm_${BUILDKITE_COMMIT}_$(tr -dc A-Za-z0-9 < /dev/urandom | hea
 # # Install AWS CLI to authenticate to ECR Public Gallery to get higher rate limit for pulling images
 # sudo apt-get update && sudo apt-get install -y awscli
 ##  Use safe docker login helper to prevent race conditions
-# source "$(dirname "${BASH_SOURCE[0]}")/../docker_login_ecr_public.sh"
+# source "$(dirname "${BASH_SOURCE[0]}")/../../common/scripts/docker_login_ecr_public.sh"
 # safe_docker_login_ecr_public
 
 ## Pull the container from AMD Docker Hub
@@ -85,7 +64,8 @@ container_name="rocm_${BUILDKITE_COMMIT}_$(tr -dc A-Za-z0-9 < /dev/urandom | hea
 docker pull "${image_name}"
 
 remove_docker_container() {
-   docker rm -f "${container_name}" || docker image rm -f "${image_name}" || true
+   docker ps -aq --filter "name=${container_name}" | xargs -r docker rm -f || true
+   docker image rm -f "${image_name}" || true
 }
 trap remove_docker_container EXIT
 
@@ -128,7 +108,6 @@ if [[ $commands == *"--shard-id="* ]]; then
         --network=host \
         --shm-size=16gb \
         --group-add "$render_gid" \
-        --rm \
         -e MIOPEN_DEBUG_CONV_DIRECT=0 \
         -e MIOPEN_DEBUG_CONV_GEMM=0 \
         -e VLLM_ROCM_USE_AITER=0 \
@@ -161,7 +140,6 @@ else
           --network=host \
           --shm-size=16gb \
           --group-add "$render_gid" \
-          --rm \
           -e MIOPEN_DEBUG_CONV_DIRECT=0 \
           -e MIOPEN_DEBUG_CONV_GEMM=0 \
           -e VLLM_ROCM_USE_AITER=0 \

.buildkite/amd/test-amd-merge.yml

@@ -0,0 +1,233 @@
+env:
+  VLLM_WORKER_MULTIPROC_METHOD: spawn
+  HF_HUB_DOWNLOAD_TIMEOUT: 300
+  HF_HUB_ETAG_TIMEOUT: 60
+
+steps:
+
+- group: ":card_index_dividers: Simple Test"
+  steps:
+    - label: "Simple · Diffusion & Model Executor Test (1/2)"
+      agent_pool: mi325_1
+      depends_on: amd-build
+      timeout_in_minutes: 70
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - export VLLM_ROCM_USE_AITER=0
+        # ignore test_teacache_extractors.py because it use rocm gemm kernel from vLLM
+        # that is not supported on CPU
+        # CPU tests on ROCm gpu machines are very slow, so we give them more time.
+        - "timeout 60m pytest -sv tests/diffusion -m 'core_model and cpu' --ignore=tests/diffusion/cache/test_teacache_extractors.py --cov=vllm_omni --cov-report=term-missing:skip-covered --cov-report=xml"
+
+    - label: "Simple · Diffusion & Model Executor Test (2/2)"
+      agent_pool: mi325_1
+      depends_on: amd-build
+      timeout_in_minutes: 70
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - export VLLM_ROCM_USE_AITER=0
+        # CPU tests on ROCm gpu machines are very slow, so we give them more time.
+        - "timeout 60m pytest -sv tests/model_executor -m 'core_model and cpu' --cov=vllm_omni --cov-report=term-missing:skip-covered --cov-report=xml"
+
+    - label: "Simple · Other Test"
+      agent_pool: mi325_1
+      depends_on: amd-build
+      timeout_in_minutes: 50
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - export VLLM_ROCM_USE_AITER=0
+        # ignore test_teacache_extractors.py because it use rocm gemm kernel from vLLM
+        # that is not supported on CPU
+        - "timeout 40m pytest -sv tests/diffusion tests/model_executor -m 'core_model and cpu' --ignore=tests/diffusion/cache/test_teacache_extractors.py --cov=vllm_omni --cov-report=term-missing:skip-covered --cov-report=xml"
+
+- group: ":card_index_dividers: Diffusion Test"
+  steps:
+    - label: "Diffusion · LoRA Test"
+      agent_pool: mi325_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - timeout 20m pytest -s -v tests/diffusion/lora/test_images_generations_lora.py
+
+    - label: "Diffusion Model CPU offloading Test"
+      agent_pool: mi325_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - export VLLM_LOGGING_LEVEL=DEBUG
+        - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+        - |
+          timeout 20m bash -c '
+            set +e
+            pytest -s -v tests/diffusion/offloader/test_diffusion_cpu_offload.py
+            EXIT1=\$?
+            pytest -s -v tests/diffusion/offloader/test_diffusion_layerwise_offload.py
+            EXIT2=\$?
+            exit \$((EXIT1 | EXIT2))
+          '
+
+    - label: "Diffusion Sequence Parallelism Test (Need 4 GPUs)"
+      agent_pool: mi325_4
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - export VLLM_LOGGING_LEVEL=DEBUG
+        - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+        - timeout 20m pytest -s -v tests/diffusion/distributed/test_sequence_parallel.py
+        - timeout 20m pytest -s -v tests/diffusion/distributed/test_ulysses_uaa_perf.py
+
+    # merge-only tests
+    - label: "Diffusion Tensor Parallelism Test"
+      agent_pool: mi325_2
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - pytest -s -v tests/diffusion/distributed/test_zimage_parallelism.py
+
+    - label: "Diffusion Model Test"
+      agent_pool: mi325_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - timeout 30m pytest -s -v tests/e2e/offline_inference/test_z_image.py
+
+    - label: "Diffusion GPU Worker Test"
+      agent_pool: mi325_2
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - timeout 20m pytest -s -v tests/diffusion/test_diffusion_worker.py
+
+    - label: "Diffusion · Wan22 Test"
+      agent_pool: mi325_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - |
+          timeout 40m bash -c "
+            export VLLM_IMAGE_FETCH_TIMEOUT=60
+            pytest -s -v tests/e2e/offline_inference/test_wan22_t2v.py tests/e2e/online_serving/test_wan22_t2v.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
+          "
+
+## ISSUE depends on `diffusers` package: https://github.com/huggingface/diffusers/issues/13274
+# - label: "Audio Generation Model Test"
+#   agent_pool: mi325_1
+#   depends_on: amd-build
+#   mirror_hardwares: [amdproduction]
+#   grade: Blocking
+#   commands:
+#     - export GPU_ARCHS=gfx942
+#     - export VLLM_LOGGING_LEVEL=DEBUG
+#     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+#     - timeout 20m pytest -s -v tests/e2e/offline_inference/test_stable_audio_expansion.py -m "advanced_model and diffusion and L4" --run-level advanced_model
+
+- label: "Engine Test"
+  agent_pool: mi325_1
+  depends_on: amd-build
+  mirror_hardwares: [amdproduction]
+  grade: Blocking
+  commands:
+    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+    - timeout 20m pytest -s -v tests/engine/test_async_omni_engine_abort.py
+
+# This is from test-nightly.yml.
+# This model seems to be important, so I am trying to run it here.
+# Since this is from test-nightly.yml,
+# I keep this non-blocking.
+- group: ":card_index_dividers: E2E Test"
+  steps:
+    - label: "Omni Model Test Qwen2-5-Omni"
+      agent_pool: mi325_4
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      commands:
+        - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
+        - pytest -s -v tests/e2e/*/test_qwen2_5_omni_expansion.py --run-level "core_model"
+
+    - label: "Omni Model Test Qwen3-Omni"
+      agent_pool: mi325_2
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+        - pytest -s -v tests/e2e/offline_inference/test_qwen3_omni.py tests/e2e/online_serving/test_qwen3_omni.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+
+    - label: "Qwen3-TTS CustomVoice E2E Test"
+      agent_pool: mi325_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - |
+          timeout 50m bash -c '
+            export VLLM_LOGGING_LEVEL=DEBUG
+            export VLLM_WORKER_MULTIPROC_METHOD=spawn
+            export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
+            pytest -s -v tests/e2e/online_serving/test_qwen3_tts_customvoice.py tests/e2e/offline_inference/test_qwen3_tts_customvoice.py -m "advanced_model" --run-level "advanced_model"
+          '
+
+    - label: "Qwen3-TTS Base E2E Test"
+      agent_pool: mi325_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      timeout_in_minutes: 50
+      commands:
+        - |
+          timeout 40m bash -c '
+            export VLLM_LOGGING_LEVEL=DEBUG
+            export VLLM_WORKER_MULTIPROC_METHOD=spawn
+            export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
+            pytest -s -v tests/e2e/online_serving/test_qwen3_tts_base.py tests/e2e/offline_inference/test_qwen3_tts_base.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+          '
+
+# TODO: Bagel test on ROCm is very unstable. @tjtanaa
+# Need to debug before reneable numerical changes across large PRs
+# # split Bagel Model Test with H100 (Real Weights) into three tests
+# - label: "Bagel Text2Img Model Test (1/3)"
+#   agent_pool: mi325_1
+#   depends_on: amd-build
+#   mirror_hardwares: [amdproduction]
+#   grade: Blocking
+#   commands:
+#     - export GPU_ARCHS=gfx942
+#     - export VLLM_LOGGING_LEVEL=DEBUG
+#     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+#     - export VLLM_ROCM_USE_AITER_RMSNORM=0
+#     - timeout 30m pytest -s -v tests/e2e/offline_inference/test_bagel.py -m "advanced_model" --run-level "advanced_model" -k "shared_memory" -k "rocm"
+
+# - label: "Bagel Img2Img Model Test (2/3)"
+#   agent_pool: mi325_1
+#   depends_on: amd-build
+#   mirror_hardwares: [amdproduction]
+#   grade: Blocking
+#   commands:
+#     - export GPU_ARCHS=gfx942
+#     - export VLLM_LOGGING_LEVEL=DEBUG
+#     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+#     - export VLLM_ROCM_USE_AITER_RMSNORM=0
+#     - timeout 30m pytest -s -v tests/e2e/offline_inference/test_bagel.py -m "advanced_model" --run-level "advanced_model" -k "rocm"
+
+# - label: "Bagel Online Serving Test (3/3)"
+#   agent_pool: mi325_1
+#   depends_on: amd-build
+#   mirror_hardwares: [amdproduction]
+#   grade: Blocking
+#   commands:
+#     - export GPU_ARCHS=gfx942
+#     - export VLLM_IMAGE_FETCH_TIMEOUT=60
+#     - export VLLM_LOGGING_LEVEL=DEBUG
+#     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+#     - export VLLM_ROCM_USE_AITER_RMSNORM=0
+#     - timeout 40m pytest -s -v tests/e2e/online_serving/test_bagel.py -m "advanced_model" --run-level "advanced_model" -k "rocm"

.buildkite/amd/test-amd-ready.yml

@@ -0,0 +1,223 @@
+env:
+  VLLM_WORKER_MULTIPROC_METHOD: spawn
+  HF_HUB_DOWNLOAD_TIMEOUT: 300
+  HF_HUB_ETAG_TIMEOUT: 60
+
+steps:
+
+- group: ":card_index_dividers: Simple Test"
+  steps:
+    - label: "Simple · Diffusion & Model Executor Test"
+      agent_pool: mi325_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - export VLLM_ROCM_USE_AITER=0
+        # ignore test_teacache_extractors.py because it use rocm gemm kernel from vLLM
+        # that is not supported on CPU
+        - "timeout 40m pytest -sv tests/diffusion tests/model_executor -m 'core_model and cpu' --ignore=tests/diffusion/cache/test_teacache_extractors.py --cov=vllm_omni --cov-report=term-missing:skip-covered --cov-report=xml"
+
+    - label: "Simple · Other Test"
+      agent_pool: mi325_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - export VLLM_ROCM_USE_AITER=0
+        # ignore test_teacache_extractors.py because it use rocm gemm kernel from vLLM
+        # that is not supported on CPU
+        - "timeout 40m pytest -sv tests/diffusion tests/model_executor -m 'core_model and cpu' --ignore=tests/diffusion/cache/test_teacache_extractors.py --cov=vllm_omni --cov-report=term-missing:skip-covered --cov-report=xml"
+
+- label: "Custom Pipeline Test"
+  agent_pool: mi325_1
+  depends_on: amd-build
+  mirror_hardwares: [amdproduction]
+  grade: Blocking
+  commands:
+    - timeout 20m pytest -s -v tests/e2e/offline_inference/custom_pipeline/ -m "core_model"
+
+
+- label: "Voxtral TTS CUDA Unit Test"
+  agent_pool: mi325_1
+  depends_on: amd-build
+  mirror_hardwares: [amdproduction]
+  grade: Blocking
+  commands:
+    - timeout 10m pytest -s -v tests/model_executor/models/voxtral_tts/test_cuda_graph_acoustic_transformer.py
+
+- group: ":card_index_dividers: Diffusion Test"
+  steps:
+    - label: "Diffusion · Batch Test"
+      agent_pool: mi325_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - timeout 20m pytest -sv tests/diffusion/batching -m 'core_model and cuda' --run-level "core_model"
+
+    - label: "Diffusion · Offloader Test"
+      agent_pool: mi325_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - export VLLM_LOGGING_LEVEL=DEBUG
+        - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+        - |
+          timeout 20m bash -c '
+            set +e
+            pytest -s -v tests/diffusion/offloader/test_diffusion_cpu_offload.py
+            EXIT1=\$?
+            pytest -s -v tests/diffusion/offloader/test_diffusion_layerwise_offload.py
+            EXIT2=\$?
+            exit \$((EXIT1 | EXIT2))
+          '
+
+    - label: "Diffusion · Cache Test"
+      agent_pool: mi325_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - export GPU_ARCHS=gfx942
+        - export VLLM_LOGGING_LEVEL=DEBUG
+        - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+        - timeout 15m pytest -s -v -m "core_model and cache and diffusion and not distributed_cuda and L4"
+
+    - label: "Diffusion Sequence Parallelism Test"
+      agent_pool: mi325_2
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - export VLLM_LOGGING_LEVEL=DEBUG
+        - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+        - timeout 20m pytest -s -v tests/diffusion/distributed/test_sequence_parallel.py -m core_model
+
+    - label: "Diffusion GPU Worker Test"
+      agent_pool: mi325_2
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - timeout 20m pytest -s -v tests/diffusion/test_diffusion_worker.py
+
+    - label: "Diffusion · Model Test"
+      agent_pool: mi325_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - timeout 15m pytest -sv tests/diffusion/models/ -m 'core_model and cuda' --run-level "core_model"
+
+    - label: "Diffusion · Wan22 Test"
+      agent_pool: mi325_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - |
+          timeout 40m bash -c "
+            export VLLM_TEST_CLEAN_GPU_MEMORY=1
+            export VLLM_IMAGE_FETCH_TIMEOUT=60
+            pytest -s -v tests/e2e/online_serving/test_wan22_t2v.py -m 'core_model' --run-level 'core_model'
+          "
+
+## ISSUE depends on `diffusers` package: https://github.com/huggingface/diffusers/issues/13274
+# - label: "Audio Generation Model Test"
+#   agent_pool: mi325_1
+#   depends_on: amd-build
+#   mirror_hardwares: [amdproduction]
+#   grade: Blocking
+#   commands:
+#     - export GPU_ARCHS=gfx942
+#     - export VLLM_LOGGING_LEVEL=DEBUG
+#     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+#     - timeout 20m pytest -s -v tests/e2e/offline_inference/test_stable_audio_expansion.py -m "advanced_model and diffusion and L4" --run-level advanced_model
+
+- label: "Engine Test"
+  agent_pool: mi325_1
+  depends_on: amd-build
+  mirror_hardwares: [amdproduction]
+  grade: Blocking
+  commands:
+    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+    - |
+      timeout 15m bash -c '
+              pytest -s -v tests/engine/test_async_omni_engine_abort.py
+      '
+
+# - label: "Omni Model Test Qwen3-Omni"
+#   agent_pool: mi325_2
+#   depends_on: amd-build
+#   mirror_hardwares: [amdproduction]
+#   grade: Blocking
+#   commands:
+#     - export VLLM_LOGGING_LEVEL=DEBUG
+#     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+#     - timeout 10m pytest -s -v tests/e2e/offline_inference/test_qwen3_omni.py
+#     - timeout 20m pytest -s -v tests/e2e/online_serving/test_qwen3_omni.py -m "core_model" --run-level "core_model"
+
+- group: ":card_index_dividers: E2E Test"
+  steps:
+    - label: "Qwen3-TTS E2E Test"
+      agent_pool: mi325_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - export VLLM_LOGGING_LEVEL=DEBUG
+        - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+        - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
+        - timeout 30m pytest -s -v tests/e2e/online_serving/test_qwen3_tts_customvoice.py -m "core_model" --run-level "core_model"
+
+    - label: "CosyVoice3-TTS E2E Test"
+      agent_pool: mi325_1
+      depends_on: amd-build
+      mirror_hardwares: [amdproduction]
+      grade: Blocking
+      commands:
+        - |
+          timeout 20m bash -c '
+            pytest -s -v tests/e2e/online_serving/test_cosyvoice3_tts_expansion.py -m "slow" --run-level "core_model"
+          '
+
+# TODO: Bagel test on ROCm is very unstable. @tjtanaa
+# Need to debug before reneable numerical changes across large PRs
+# - label: "Bagel Text2Img Model Test"
+#   agent_pool: mi325_1
+#   depends_on: amd-build
+#   mirror_hardwares: [amdproduction]
+#   grade: Blocking
+#   commands:
+#     - export GPU_ARCHS=gfx942
+#     - export VLLM_LOGGING_LEVEL=DEBUG
+#     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+#     - export VLLM_ROCM_USE_AITER_RMSNORM=0
+#     - timeout 30m pytest -s -v tests/e2e/offline_inference/test_bagel.py -m "core_model" --run-level "core_model" -k "rocm"
+
+# - label: "Bagel Img2Img Model Test"
+#   agent_pool: mi325_1
+#   depends_on: amd-build
+#   mirror_hardwares: [amdproduction]
+#   grade: Blocking
+#   commands:
+#     - export GPU_ARCHS=gfx942
+#     - export VLLM_LOGGING_LEVEL=DEBUG
+#     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+#     - export VLLM_ROCM_USE_AITER_RMSNORM=0
+#     - timeout 30m pytest -s -v tests/e2e/offline_inference/test_bagel.py -m "core_model" --run-level "core_model" -k "rocm"
+
+# - label: "Bagel Online Serving Test"
+#   agent_pool: mi325_1
+#   depends_on: amd-build
+#   mirror_hardwares: [amdproduction]
+#   grade: Blocking
+#   commands:
+#     - export GPU_ARCHS=gfx942
+#     - export VLLM_IMAGE_FETCH_TIMEOUT=60
+#     - export VLLM_LOGGING_LEVEL=DEBUG
+#     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+#     - export VLLM_ROCM_USE_AITER_RMSNORM=0
+#     - timeout 40m pytest -s -v tests/e2e/online_serving/test_bagel.py -m "core_model" --run-level "core_model" -k "rocm"

.buildkite/amd/test-template-amd-omni.j2

@@ -0,0 +1,141 @@
+{# vllm-omni customized version
+   Based on: https://github.com/vllm-project/ci-infra/blob/main/buildkite/test-template-amd.j2
+   Last synced: 2025-12-15
+   Modifications: Removed unused CUDA/NVIDIA logic, keeping only AMD tests
+#}
+{% set docker_image_amd = "rocm/vllm-omni:$BUILDKITE_COMMIT" %}
+{% set default_working_dir = "/app/vllm-omni" %}
+
+{% if env %}
+env:
+{% for name in env %}
+  {{ name }}: "{{ env[name] }}"
+{% endfor %}
+{% endif %}
+steps:
+  - group: "AMD Tests"
+    depends_on: ~
+    steps:
+      - label: "AMD: :docker: build image"
+        depends_on: ~
+        soft_fail: false
+        commands:
+          - "docker build -f docker/Dockerfile.rocm -t {{ docker_image_amd }} --target test --build-arg ARG_PYTORCH_ROCM_ARCH=gfx942 --progress plain ."
+          - "docker push {{ docker_image_amd }}"
+        key: "amd-build"
+        env:
+          DOCKER_BUILDKIT: "1"
+        retry:
+          automatic:
+            - exit_status: -1  # Agent was lost
+              limit: 2
+            - exit_status: -10  # Agent was lost
+              limit: 2
+            - exit_status: 128  # Git connectivity issues
+              limit: 2
+            - exit_status: 1  # Machine occasionally fail
+              limit: 1
+        agents:
+          queue: amd-cpu
+
+{% for step_group in steps %}
+{% if step_group.group %}
+  - group: "{{ step_group.group }}"
+{% if step_group.depends_on %}
+    depends_on: {{ step_group.depends_on }}
+{% endif %}
+    steps:
+{% for step in step_group.steps %}
+{% if step.mirror_hardwares and mirror_hw in step.mirror_hardwares %}
+      - label: "{{ step.agent_pool }}: {{ step.label }}"
+        depends_on: amd-build
+        agents:
+          {% if step.agent_pool %}
+          queue: amd_{{ step.agent_pool }}
+          {% else %}
+          queue: amd_mi325_1
+          {% endif %}
+{% set cmd_body = (step.command or (step.commands | join("\n"))) | trim %}
+{% set indented_cmd = cmd_body | replace("\n", "\n            ") %}
+        command: bash .buildkite/amd/scripts/run-amd-test.sh
+        env:
+          DOCKER_BUILDKIT: "1"
+{% if step.env %}
+{% for name in step.env %}
+          {{ name }}: "{{ step.env[name] }}"
+{% endfor %}
+{% endif %}
+          TEST_COMMAND: |-
+            (command rocm-smi || true) && cd {{ (step.working_dir or default_working_dir) | safe }}
+{% if env %}
+{% for name in env %}
+            export {{ name }}="{{ env[name] }}"
+{% endfor %}
+{% endif %}
+{% if step.env %}
+{% for name in step.env %}
+            export {{ name }}="{{ step.env[name] }}"
+{% endfor %}
+{% endif %}
+{% if "mi250" in step.agent_pool %}
+            python3 -m pip uninstall -y amd-aiter
+{% endif %}
+            {{ indented_cmd | safe }}
+        priority: 100
+        {% if step.timeout_in_minutes %}
+        timeout_in_minutes: {{ step.timeout_in_minutes }}
+        {% endif %}
+        {% if step.grade and step.grade == "Blocking" %}
+        soft_fail: false
+        {% else %}
+        soft_fail: true
+        {% endif%}
+{% endif %}
+{% endfor %}
+{% elif step_group.mirror_hardwares and mirror_hw in step_group.mirror_hardwares %}
+{% set step = step_group %}
+  - label: "{{ step.agent_pool }}: {{ step.label }}"
+    depends_on: amd-build
+    agents:
+      {% if step.agent_pool %}
+      queue: amd_{{ step.agent_pool }}
+      {% else %}
+      queue: amd_mi325_1
+      {% endif %}
+{% set cmd_body = (step.command or (step.commands | join("\n"))) | trim %}
+{% set indented_cmd = cmd_body | replace("\n", "\n        ") %}
+    command: bash .buildkite/amd/scripts/run-amd-test.sh
+    env:
+      DOCKER_BUILDKIT: "1"
+{% if step.env %}
+{% for name in step.env %}
+      {{ name }}: "{{ step.env[name] }}"
+{% endfor %}
+{% endif %}
+      TEST_COMMAND: |-
+        (command rocm-smi || true) && cd {{ (step.working_dir or default_working_dir) | safe }}
+{% if env %}
+{% for name in env %}
+        export {{ name }}="{{ env[name] }}"
+{% endfor %}
+{% endif %}
+{% if step.env %}
+{% for name in step.env %}
+        export {{ name }}="{{ step.env[name] }}"
+{% endfor %}
+{% endif %}
+{% if "mi250" in step.agent_pool %}
+        python3 -m pip uninstall -y amd-aiter
+{% endif %}
+        {{ indented_cmd | safe }}
+    priority: 100
+    {% if step.timeout_in_minutes %}
+    timeout_in_minutes: {{ step.timeout_in_minutes }}
+    {% endif %}
+    {% if step.grade and step.grade == "Blocking" %}
+    soft_fail: false
+    {% else %}
+    soft_fail: true
+    {% endif%}
+{% endif %}
+{% endfor %}

.buildkite/bootstrap-intel-omni.sh

@@ -1,51 +0,0 @@
-#!/bin/bash
-# vllm-omni Intel bootstrap
-# Uses static pipeline-intel.yaml for Intel XPU tests
-# Based on: bootstrap-amd-omni.sh
-
-set -euo pipefail
-
-if [[ -z "${DOCS_ONLY_DISABLE:-}" ]]; then
-    DOCS_ONLY_DISABLE=0
-fi
-
-upload_pipeline() {
-    echo "--- 🛠 Preparing Intel pipeline"
-    buildkite-agent pipeline upload .buildkite/pipeline-intel.yaml
-}
-
-get_diff() {
-    $(git add .)
-    echo $(git diff --name-only --diff-filter=ACMDR $(git merge-base origin/main HEAD))
-}
-
-get_diff_main() {
-    $(git add .)
-    echo $(git diff --name-only --diff-filter=ACMDR HEAD~1)
-}
-
-file_diff=$(get_diff)
-if [[ $BUILDKITE_BRANCH == "main" ]]; then
-    file_diff=$(get_diff_main)
-fi
-
-# ----------------------------------------------------------------------
-# Early exit: skip pipeline if all changed files are under docs/
-# ----------------------------------------------------------------------
-if [[ "${DOCS_ONLY_DISABLE}" != "1" ]] && [[ -n "${file_diff:-}" ]]; then
-    docs_only=1
-    while IFS= read -r f; do
-        [[ -z "$f" ]] && continue
-        if [[ "$f" != docs/* ]]; then
-            docs_only=0
-            break
-        fi
-    done < <(printf '%s\n' "$file_diff" | tr ' ' '\n' | tr -d '\r')
-
-    if [[ "$docs_only" -eq 1 ]]; then
-        buildkite-agent annotate ":memo: CI skipped — docs only" --style "info"
-        exit 0
-    fi
-fi
-
-upload_pipeline

.buildkite/common/ci_mirror_hardwares.yml

@@ -0,0 +1,239 @@
+# Uploader-only hardware presets for CUDA / NPU test YAML.
+# Referenced as ``mirror_hardwares: l4_1`` on a step; ``upload_pipeline.py`` expands to
+# ``agents`` (+ optional top-level ``image`` for NPU) + ``plugins``, then strips
+# ``mirror_hardwares`` before Buildkite upload.
+#
+# CUDA naming: {chip}_{count}
+#   l4_1, l4_4, h100_1 … h100_4
+# NPU naming: {queue}_{resource_class}
+#   a2b3_npu_1  — ascend-a2b3, resource_class npu-1 (A2/B3 CI image)
+#   a2b3_npu_4  — ascend-a2b3, resource_class npu-4
+#   a3_npu_2    — ascend-a3, resource_class npu-2 (A3 CI image tag …:$BUILDKITE_COMMIT-a3)
+
+x-npu-k8s-plugins: &npu_k8s_plugins
+  - kubernetes:
+      podSpecPatch:
+        imagePullSecrets:
+          - name: swr-secret
+
+
+mirror_hardwares:
+  l4_1:
+    agents:
+      queue: gpu_1_queue
+    plugins:
+      - docker#v5.2.0:
+          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
+          always-pull: true
+          propagate-environment: true
+          shm-size: "8gb"
+          environment:
+            - HF_HOME=/fsx/hf_cache
+            - HF_TOKEN
+          volumes:
+            - /fsx/hf_cache:/fsx/hf_cache
+
+  l4_4:
+    agents:
+      queue: gpu_4_queue
+    plugins:
+      - docker#v5.2.0:
+          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
+          always-pull: true
+          propagate-environment: true
+          shm-size: "8gb"
+          environment:
+            - HF_HOME=/fsx/hf_cache
+            - HF_TOKEN
+          volumes:
+            - /fsx/hf_cache:/fsx/hf_cache
+
+  h100_1:
+    agents:
+      queue: mithril-h100-pool
+    plugins:
+      - kubernetes:
+          podSpec:
+            containers:
+              - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
+                resources:
+                  limits:
+                    nvidia.com/gpu: 1
+                volumeMounts:
+                  - name: devshm
+                    mountPath: /dev/shm
+                  - name: hf-cache
+                    mountPath: /root/.cache/huggingface
+                env:
+                  - name: HF_HOME
+                    value: /root/.cache/huggingface
+                  - name: HF_TOKEN
+                    valueFrom:
+                      secretKeyRef:
+                        name: hf-token-secret
+                        key: token
+            nodeSelector:
+              node.kubernetes.io/instance-type: gpu-h100-sxm
+            volumes:
+              - name: devshm
+                emptyDir:
+                  medium: Memory
+              - name: hf-cache
+                hostPath:
+                  path: /mnt/hf-cache
+                  type: DirectoryOrCreate
+
+  h100_2:
+    agents:
+      queue: mithril-h100-pool
+    plugins:
+      - kubernetes:
+          podSpec:
+            containers:
+              - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
+                resources:
+                  limits:
+                    nvidia.com/gpu: 2
+                volumeMounts:
+                  - name: devshm
+                    mountPath: /dev/shm
+                  - name: hf-cache
+                    mountPath: /root/.cache/huggingface
+                env:
+                  - name: HF_HOME
+                    value: /root/.cache/huggingface
+                  - name: HF_TOKEN
+                    valueFrom:
+                      secretKeyRef:
+                        name: hf-token-secret
+                        key: token
+            nodeSelector:
+              node.kubernetes.io/instance-type: gpu-h100-sxm
+            volumes:
+              - name: devshm
+                emptyDir:
+                  medium: Memory
+              - name: hf-cache
+                hostPath:
+                  path: /mnt/hf-cache
+                  type: DirectoryOrCreate
+
+  h100_3:
+    agents:
+      queue: mithril-h100-pool
+    plugins:
+      - kubernetes:
+          podSpec:
+            containers:
+              - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
+                resources:
+                  limits:
+                    nvidia.com/gpu: 3
+                volumeMounts:
+                  - name: devshm
+                    mountPath: /dev/shm
+                  - name: hf-cache
+                    mountPath: /root/.cache/huggingface
+                env:
+                  - name: HF_HOME
+                    value: /root/.cache/huggingface
+                  - name: HF_TOKEN
+                    valueFrom:
+                      secretKeyRef:
+                        name: hf-token-secret
+                        key: token
+            nodeSelector:
+              node.kubernetes.io/instance-type: gpu-h100-sxm
+            volumes:
+              - name: devshm
+                emptyDir:
+                  medium: Memory
+              - name: hf-cache
+                hostPath:
+                  path: /mnt/hf-cache
+                  type: DirectoryOrCreate
+
+  h100_4:
+    agents:
+      queue: mithril-h100-pool
+    plugins:
+      - kubernetes:
+          podSpec:
+            containers:
+              - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
+                resources:
+                  limits:
+                    nvidia.com/gpu: 4
+                volumeMounts:
+                  - name: devshm
+                    mountPath: /dev/shm
+                  - name: hf-cache
+                    mountPath: /root/.cache/huggingface
+                env:
+                  - name: HF_HOME
+                    value: /root/.cache/huggingface
+                  - name: HF_TOKEN
+                    valueFrom:
+                      secretKeyRef:
+                        name: hf-token-secret
+                        key: token
+            nodeSelector:
+              node.kubernetes.io/instance-type: gpu-h100-sxm
+            volumes:
+              - name: devshm
+                emptyDir:
+                  medium: Memory
+              - name: hf-cache
+                hostPath:
+                  path: /mnt/hf-cache
+                  type: DirectoryOrCreate
+
+  a2b3_npu_1:
+    agents:
+      queue: "ascend-a2b3"
+      resource_class: "npu-1"
+    image: "swr.cn-southwest-2.myhuaweicloud.com/modelfoundry/vllm-omni-ci-npu:${BUILDKITE_COMMIT}"
+    plugins: *npu_k8s_plugins
+
+  a2b3_npu_4:
+    agents:
+      queue: "ascend-a2b3"
+      resource_class: "npu-4"
+    image: "swr.cn-southwest-2.myhuaweicloud.com/modelfoundry/vllm-omni-ci-npu:${BUILDKITE_COMMIT}"
+    plugins: *npu_k8s_plugins
+
+
+  a2b3_npu_8:
+    agents:
+      queue: "ascend-a2b3"
+      resource_class: "npu-8"
+    image: "swr.cn-southwest-2.myhuaweicloud.com/modelfoundry/vllm-omni-ci-npu:${BUILDKITE_COMMIT}"
+    plugins: *npu_k8s_plugins
+
+  a3_npu_2:
+    agents:
+      queue: "ascend-a3"
+      resource_class: "npu-2"
+    image: "swr.cn-southwest-2.myhuaweicloud.com/modelfoundry/vllm-omni-ci-npu:${BUILDKITE_COMMIT}-a3"
+    plugins: *npu_k8s_plugins
+
+  a3_npu_4:
+    agents:
+      queue: "ascend-a3"
+      resource_class: "npu-4"
+    image: "swr.cn-southwest-2.myhuaweicloud.com/modelfoundry/vllm-omni-ci-npu:${BUILDKITE_COMMIT}-a3"
+    plugins: *npu_k8s_plugins
+
+  a3_npu_8:
+    agents:
+      queue: "ascend-a3"
+      resource_class: "npu-8"
+    image: "swr.cn-southwest-2.myhuaweicloud.com/modelfoundry/vllm-omni-ci-npu:${BUILDKITE_COMMIT}-a3"
+    plugins: *npu_k8s_plugins
+
+  a3_npu_16:
+    agents:
+      queue: "ascend-a3"
+      resource_class: "npu-16"
+    image: "swr.cn-southwest-2.myhuaweicloud.com/modelfoundry/vllm-omni-ci-npu:${BUILDKITE_COMMIT}-a3"
+    plugins: *npu_k8s_plugins

.buildkite/common/scripts/resolve_skip_ci.sh

@@ -0,0 +1,23 @@
+#!/usr/bin/env bash
+# Shared Buildkite skip-ci helpers for AMD / Intel bootstrap scripts.
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+SKIP_CI_PY="${SCRIPT_DIR}/skip_ci.py"
+
+# Resolve git diff once. Exit 0 when this bootstrap should stop uploading.
+# Decision details are logged by skip_ci.py; this helper only annotates skip-all.
+# Usage: gate_bootstrap_ci <platform> <l2|l3>
+gate_bootstrap_ci() {
+    local platform="$1"
+    local level="$2"
+    local kind
+    if kind="$(python3 "${SKIP_CI_PY}" gate "${platform}" "${level}")"; then
+        if [[ "${kind}" == "skip-all" ]]; then
+            buildkite-agent annotate \
+                ":memo: CI skipped — docs or pytest skip-mark changes only" \
+                --style "info" 2>/dev/null || true
+        fi
+        exit 0
+    fi
+}

.buildkite/common/scripts/skip_ci.py

@@ -0,0 +1,617 @@
+#!/usr/bin/env python3
+# SPDX-License-Identifier: Apache-2.0
+"""Shared Buildkite skip-ci and CI-YAML level targeting logic.
+
+``resolve_ci_decision(changed_files, ...)`` buckets explicit file lists (tests, dry-run).
+``resolve_ci_context_from_git()`` resolves git diff once and returns decision + changed files
+for Buildkite bootstrap / upload scripts.
+
+CLI (for bootstrap scripts; exit code is the signal):
+  python3 skip_ci.py gate <platform> <level>
+      Exit 0 when bootstrap should stop (skip-all, or that L2/L3 target is off).
+      Prints ``skip-all`` or ``skip-l23`` on stdout; logs the decision once.
+  python3 skip_ci.py check-skip-all
+      Exit 0 when only docs/skip-mark changes → bootstrap skips entire CI upload.
+  python3 skip_ci.py check-skip-l2-l3 <platform> <level>
+      Exit 0 when that L2/L3 pipeline should be skipped (CI config YAML-only diffs).
+  python3 skip_ci.py print-annotate
+      Print decision message for Buildkite annotation.
+"""
+
+from __future__ import annotations
+
+import argparse
+import os
+import re
+import subprocess
+import sys
+from dataclasses import dataclass, field, replace
+from pathlib import Path
+
+LOG = "[skip-ci]"
+ROOT = Path(__file__).resolve().parent.parent.parent.parent
+
+PLATFORMS = ("cuda", "amd", "intel", "npu")
+L23_LEVELS = frozenset({"l2", "l3"})
+
+L2_YAML_FILES: dict[str, str] = {
+    ".buildkite/cuda/test-ready.yml": "cuda",
+    ".buildkite/npu/test-npu-ready.yml": "npu",
+    ".buildkite/amd/test-amd-ready.yml": "amd",
+    ".buildkite/intel/pipeline-intel.yml": "intel",
+}
+
+L3_YAML_FILES: dict[str, str] = {
+    ".buildkite/cuda/test-merge.yml": "cuda",
+    ".buildkite/amd/test-amd-merge.yml": "amd",
+}
+
+L45_YAML_FILES: dict[str, str] = {
+    ".buildkite/cuda/test-nightly.yml": "cuda",
+    ".buildkite/cuda/test-weekly.yml": "cuda",
+    ".buildkite/npu/test-npu-nightly.yml": "npu",
+}
+
+
+def _empty_platform_buckets() -> dict[str, list[str]]:
+    return {platform: [] for platform in PLATFORMS}
+
+
+@dataclass(frozen=True)
+class CiDevice:
+    """Per-platform / per-level skip flags for yaml-gated bootstrap (``skip_l2_l3``)."""
+
+    skip_cuda_l2: bool = False
+    skip_cuda_l3: bool = False
+    skip_amd_l2: bool = False
+    skip_amd_l3: bool = False
+    skip_intel_l2: bool = False
+    skip_intel_l3: bool = False
+    skip_npu_l2: bool = False
+    skip_npu_l3: bool = False
+
+    def is_skip(self, platform: str, level: str) -> bool:
+        return bool(getattr(self, f"skip_{platform}_{level}", True))
+
+
+@dataclass
+class DiffBuckets:
+    """Changed files grouped by doc / skip-mark / CI yaml / other."""
+
+    docs: list[str] = field(default_factory=list)
+    skip_tests: list[str] = field(default_factory=list)
+    other: list[str] = field(default_factory=list)
+    l2: dict[str, list[str]] = field(default_factory=_empty_platform_buckets)
+    l3: dict[str, list[str]] = field(default_factory=_empty_platform_buckets)
+    l45: dict[str, list[str]] = field(default_factory=_empty_platform_buckets)
+
+    @property
+    def has_doc_changes(self) -> bool:
+        return bool(self.docs)
+
+    @property
+    def has_skip_test_changes(self) -> bool:
+        return bool(self.skip_tests)
+
+    @property
+    def has_other_changes(self) -> bool:
+        """Product code or paths outside skip-ci scope."""
+        return bool(self.other)
+
+    @property
+    def has_l23_yaml_changes(self) -> bool:
+        """At least one whitelisted L2/L3 CI yaml file changed."""
+        return any(self.l2[platform] or self.l3[platform] for platform in PLATFORMS)
+
+    @property
+    def has_l45_yaml_changes(self) -> bool:
+        """At least one whitelisted L4/L5 (nightly/weekly) CI yaml file changed."""
+        return any(self.l45[platform] for platform in PLATFORMS)
+
+    @property
+    def has_skip_ci_scope_changes(self) -> bool:
+        """At least one doc / skip-mark / whitelisted CI yaml change."""
+        return (
+            self.has_doc_changes or self.has_skip_test_changes or self.has_l23_yaml_changes or self.has_l45_yaml_changes
+        )
+
+    def l2_platforms(self) -> frozenset[str]:
+        return frozenset(platform for platform in PLATFORMS if self.l2[platform])
+
+    def l3_platforms(self) -> frozenset[str]:
+        return frozenset(platform for platform in PLATFORMS if self.l3[platform])
+
+
+@dataclass(frozen=True)
+class SkipL2L3Basis:
+    skip_all_l2: bool
+    skip_all_l3: bool
+
+    @classmethod
+    def from_buckets(cls, buckets: DiffBuckets) -> SkipL2L3Basis | None:
+        """L4/L5 yaml → skip both L2 and L3 (L2/L3 buckets may rescue).
+
+        Docs / skip-mark changes do not block this path; they are ignored for
+        L2/L3 targeting when CI YAML is also present.
+        """
+        if buckets.has_other_changes:
+            return None
+        if not buckets.has_l45_yaml_changes:
+            return None
+        return cls(
+            skip_all_l2=not buckets.l2_platforms(),
+            skip_all_l3=not buckets.l3_platforms(),
+        )
+
+
+@dataclass(frozen=True)
+class SkipL3Basis:
+    enable_l2_platforms: frozenset[str]
+
+    @classmethod
+    def from_buckets(cls, buckets: DiffBuckets) -> SkipL3Basis | None:
+        """L2 yaml changed → skip L3; keep L2 on touched platforms."""
+        platforms = buckets.l2_platforms()
+        if not platforms:
+            return None
+        return cls(enable_l2_platforms=platforms)
+
+
+@dataclass(frozen=True)
+class SkipL2Basis:
+    enable_l3_platforms: frozenset[str]
+
+    @classmethod
+    def from_buckets(cls, buckets: DiffBuckets) -> SkipL2Basis | None:
+        """L3 yaml changed → skip L2; keep L3 on touched platforms."""
+        platforms = buckets.l3_platforms()
+        if not platforms:
+            return None
+        return cls(enable_l3_platforms=platforms)
+
+
+@dataclass(frozen=True)
+class CiDecision:
+    """Final CI skip decision for one diff."""
+
+    skip_all: bool = False
+    skip_l2_l3: bool = False
+    device: CiDevice = field(default_factory=CiDevice)
+    message: str = ""
+
+    @classmethod
+    def skip_all_basis(cls, buckets: DiffBuckets, *, diff_range: str | None) -> bool:
+        """Docs / pytest skip-mark only → suppress entire default CI."""
+        if buckets.has_other_changes:
+            return False
+        if (buckets.has_l23_yaml_changes or buckets.has_l45_yaml_changes) and (
+            buckets.has_doc_changes or buckets.has_skip_test_changes
+        ):
+            return False
+        if not buckets.has_doc_changes and not buckets.has_skip_test_changes:
+            return False
+        if diff_range is None and not buckets.has_doc_changes:
+            return False
+        return True
+
+    @classmethod
+    def from_yaml_basis(
+        cls,
+        l2_l3_basis: SkipL2L3Basis | None,
+        l3_basis: SkipL3Basis | None,
+        l2_basis: SkipL2Basis | None,
+    ) -> CiDecision:
+        """Build a yaml-gated decision by merging L2/L3 skip bases.
+
+        Bases are applied as a union of enables: L2 yaml keeps L2 on touched
+        platforms, L3 yaml keeps L3 on touched platforms. Cross-platform mixes
+        (e.g. CUDA ready + AMD merge) must preserve both enables. The resulting
+        per-platform / per-level matrix is stored on ``device`` as ``skip_*``.
+        """
+        run = {platform: {"l2": True, "l3": True} for platform in PLATFORMS}
+
+        if l2_l3_basis is not None:
+            if l2_l3_basis.skip_all_l2:
+                for platform in PLATFORMS:
+                    run[platform]["l2"] = False
+            if l2_l3_basis.skip_all_l3:
+                for platform in PLATFORMS:
+                    run[platform]["l3"] = False
+
+        if l3_basis is not None:
+            for platform in PLATFORMS:
+                run[platform]["l3"] = False
+                run[platform]["l2"] = platform in l3_basis.enable_l2_platforms
+
+        if l2_basis is not None:
+            for platform in PLATFORMS:
+                if platform in l2_basis.enable_l3_platforms:
+                    run[platform]["l3"] = True
+                    if l3_basis is None or platform not in l3_basis.enable_l2_platforms:
+                        run[platform]["l2"] = False
+                elif l3_basis is None:
+                    # L3-yaml-only: disable platforms that did not change L3.
+                    run[platform]["l2"] = False
+                    run[platform]["l3"] = False
+                # else: leave L2 enables from SkipL3Basis; keep this platform's L3 off.
+
+        return cls(
+            skip_l2_l3=True,
+            device=cls._device_from_run(run),
+        )
+
+    @staticmethod
+    def _device_from_run(run: dict[str, dict[str, bool]]) -> CiDevice:
+        return CiDevice(
+            **{f"skip_{platform}_{level}": not run[platform][level] for platform in PLATFORMS for level in ("l2", "l3")}
+        )
+
+    def is_run(self, platform: str, level: str) -> bool:
+        if self.skip_all:
+            return False
+        if not self.skip_l2_l3:
+            return True
+        if level not in L23_LEVELS:
+            return True
+        return not self.device.is_skip(platform, level)
+
+
+@dataclass(frozen=True)
+class CiContext:
+    """Skip-ci decision plus git diff inputs used to compute it."""
+
+    decision: CiDecision
+    changed_files: list[str] | None
+    diff_range: str | None
+
+
+def _log(message: str) -> None:
+    print(f"{LOG}: {message}", file=sys.stderr)
+
+
+def _git(*args: str) -> subprocess.CompletedProcess[str]:
+    return subprocess.run(
+        ["git", *args],
+        check=False,
+        capture_output=True,
+        text=True,
+    )
+
+
+# --- Diff helpers ---
+
+
+def _resolve_diff_range() -> str | None:
+    """Return a git diff range for PR or main builds, or None when unavailable."""
+    is_pr = os.environ.get("BUILDKITE_PULL_REQUEST", "false") != "false" and os.environ.get(
+        "BUILDKITE_PULL_REQUEST", ""
+    )
+    commit = os.environ.get("BUILDKITE_COMMIT", "")
+
+    if is_pr:
+        base_branch = os.environ.get("BUILDKITE_PULL_REQUEST_BASE_BRANCH", "main")
+        base_ref = f"origin/{base_branch}"
+        if _git("rev-parse", "--verify", base_ref).returncode != 0:
+            _log(f"origin/{base_branch} not found locally; trying fetch")
+            _git("fetch", "--depth=200", "origin", base_branch)
+        if _git("rev-parse", "--verify", base_ref).returncode != 0:
+            if _git("rev-parse", "--verify", base_branch).returncode == 0:
+                base_ref = base_branch
+            else:
+                _log(f"cannot resolve PR base {base_branch}; using safe defaults")
+                return None
+        return f"{base_ref}...{commit}"
+
+    if os.environ.get("BUILDKITE_BRANCH", "") == "main":
+        if _git("rev-parse", "--verify", f"{commit}^").returncode != 0:
+            _log("main commit has no parent; using safe defaults")
+            return None
+        return f"{commit}^..{commit}"
+
+    _log("not PR/main build; using safe defaults")
+    return None
+
+
+def _changed_files_for_diff_range(diff_range: str | None) -> list[str] | None:
+    """Return changed file paths for *diff_range*, or None when unavailable."""
+    if diff_range is None:
+        return None
+
+    result = _git("diff", "--name-only", diff_range)
+    if result.returncode != 0:
+        _log(f"git diff failed ({diff_range}); using safe defaults")
+        return None
+
+    files = [line for line in result.stdout.splitlines() if line.strip()]
+    _log(f"{len(files)} changed file(s)")
+    return files
+
+
+def resolve_ci_context_from_git() -> CiContext:
+    """Resolve git diff once and return skip-ci decision plus changed files."""
+    diff_range = _resolve_diff_range()
+    changed_files = _changed_files_for_diff_range(diff_range)
+    decision = resolve_ci_decision(changed_files, diff_range=diff_range)
+    return CiContext(decision=decision, changed_files=changed_files, diff_range=diff_range)
+
+
+# --- Doc / skip-mark helpers ---
+
+
+_SKIP_MARK_RE = re.compile(r"pytest\.mark\.skip(?:if)?\b|pytest\.skip\s*\(")
+_PYTESTMARK_SKIP_RE = re.compile(r"pytest\.mark\.skip\b")
+_PYTEST_MARK_ONLY_RE = re.compile(r"pytest\.mark\.\w+")
+
+
+def _diff_only_contains_skip_mark_changes(diff_text: str) -> bool:
+    """Return True when a unified diff only *adds* pytest skip marks.
+
+    Deleting a skip mark re-enables the test and must not suppress CI. Editing an
+    existing skip (``-`` old / ``+`` new) still qualifies when at least one skip
+    mark is added.
+    """
+
+    def paren_balance(line: str) -> int:
+        return line.count("(") - line.count(")")
+
+    def is_blank_or_comment(line: str) -> bool:
+        stripped = line.strip()
+        return not stripped or stripped.startswith("#")
+
+    def is_skip_mark_line(line: str) -> bool:
+        return _SKIP_MARK_RE.search(line.strip()) is not None
+
+    def is_pytestmark_adjacent_line(line: str) -> bool:
+        stripped = line.strip().rstrip(",")
+        if not stripped or stripped in {"[", "]"} or stripped.startswith("#"):
+            return True
+        if stripped.startswith("pytestmark"):
+            return True
+        return _PYTEST_MARK_ONLY_RE.search(stripped) is not None
+
+    pending_depth = 0
+    saw_change = False
+    has_added_skip_mark = False
+    for raw_line in diff_text.splitlines():
+        if raw_line.startswith("@@"):
+            pending_depth = 0
+            continue
+        if not (raw_line.startswith("+") or raw_line.startswith("-")):
+            continue
+        if raw_line.startswith("+++") or raw_line.startswith("---"):
+            continue
+
+        saw_change = True
+        is_add = raw_line.startswith("+")
+        content = raw_line[1:]
+
+        if pending_depth > 0:
+            pending_depth += paren_balance(content)
+            if pending_depth < 0:
+                pending_depth = 0
+            continue
+
+        if is_blank_or_comment(content):
+            continue
+
+        if is_skip_mark_line(content):
+            if is_add:
+                has_added_skip_mark = True
+            pending_depth = max(0, paren_balance(content))
+            continue
+
+        if is_pytestmark_adjacent_line(content):
+            continue
+
+        return False
+
+    return saw_change and has_added_skip_mark
+
+
+def _is_skip_testcase_change(file_path: str, *, diff_range: str) -> bool:
+    """Return True when a test file diff is skip-mark-only (including new-file cases)."""
+    diff_result = _git("diff", diff_range, "--", file_path)
+    if diff_result.returncode != 0:
+        _log(f"skip-mark-only: git diff failed for {file_path}")
+        return False
+
+    diff_text = diff_result.stdout
+    if not diff_text.strip():
+        _log(f"skip-mark-only: empty diff for {file_path}")
+        return False
+
+    if _diff_only_contains_skip_mark_changes(diff_text):
+        return True
+
+    if "new file mode" not in diff_text and "--- /dev/null" not in diff_text:
+        _log(f"skip-mark-only: non-skip changes in {file_path}")
+        return False
+
+    commit = os.environ.get("BUILDKITE_COMMIT", "")
+    if not commit:
+        return False
+
+    content_result = _git("show", f"{commit}:{file_path}")
+    if content_result.returncode != 0:
+        _log(f"skip-mark-only: cannot read {file_path} at {commit}")
+        return False
+
+    content = content_result.stdout
+    match = re.search(r"^pytestmark\s*=\s*\[(.*?)^\s*\]", content, re.MULTILINE | re.DOTALL)
+    if match is not None and _PYTESTMARK_SKIP_RE.search(match.group(1)) is not None:
+        _log(f"skip-mark-only: new test file with module-level pytest.mark.skip: {file_path}")
+        return True
+    if re.search(r"^pytestmark\s*=\s*pytest\.mark\.skip\b", content, re.MULTILINE):
+        _log(f"skip-mark-only: new test file with module-level pytest.mark.skip: {file_path}")
+        return True
+
+    _log(f"skip-mark-only: new test file without module-level pytest.mark.skip: {file_path}")
+    return False
+
+
+def _classify_changed_files_into_buckets(changed_files: list[str], *, diff_range: str | None) -> DiffBuckets:
+    """Classify each changed file path into a ``DiffBuckets`` category."""
+    buckets = DiffBuckets()
+    for file_path in changed_files:
+        if not file_path:
+            continue
+        if file_path.startswith("docs/") or file_path.endswith(".md") or file_path == "mkdocs.yml":
+            buckets.docs.append(file_path)
+            continue
+        path = Path(file_path)
+        if path.suffix == ".py" and path.parts and path.parts[0] == "tests":
+            if diff_range is not None and _is_skip_testcase_change(file_path, diff_range=diff_range):
+                buckets.skip_tests.append(file_path)
+            else:
+                buckets.other.append(file_path)
+            continue
+        if file_path in L2_YAML_FILES:
+            platform = L2_YAML_FILES[file_path]
+            buckets.l2[platform].append(file_path)
+        elif file_path in L3_YAML_FILES:
+            platform = L3_YAML_FILES[file_path]
+            buckets.l3[platform].append(file_path)
+        elif file_path in L45_YAML_FILES:
+            platform = L45_YAML_FILES[file_path]
+            buckets.l45[platform].append(file_path)
+        else:
+            buckets.other.append(file_path)
+    return buckets
+
+
+# --- Decision ---
+
+
+def _finish(decision: CiDecision, message: str) -> CiDecision:
+    finished = replace(decision, message=message)
+    _log(message)
+    return finished
+
+
+def _format_yaml_gated_message(buckets: DiffBuckets, decision: CiDecision) -> str:
+    """Human-readable summary of which CI YAML changed and which L2/L3 targets run."""
+    changed_parts: list[str] = []
+    for level_name, by_platform in (("L2", buckets.l2), ("L3", buckets.l3), ("L4/L5", buckets.l45)):
+        files = [path for platform in PLATFORMS for path in by_platform[platform]]
+        if files:
+            changed_parts.append(f"{level_name}=[{', '.join(files)}]")
+
+    run_targets = [
+        f"{platform}/{level}" for platform in PLATFORMS for level in ("l2", "l3") if decision.is_run(platform, level)
+    ]
+    skip_targets = [
+        f"{platform}/{level}"
+        for platform in PLATFORMS
+        for level in ("l2", "l3")
+        if not decision.is_run(platform, level)
+    ]
+
+    changed = "; ".join(changed_parts) if changed_parts else "none"
+    run = ", ".join(run_targets) if run_targets else "none"
+    skip = ", ".join(skip_targets) if skip_targets else "none"
+    if buckets.has_doc_changes or buckets.has_skip_test_changes:
+        return (
+            "CI config YAML + docs/skip-mark changed — follow CI YAML gating; "
+            f"changed: {changed}; run: {run}; skip: {skip}"
+        )
+    return f"only CI config YAML changed — changed: {changed}; run: {run}; skip: {skip}"
+
+
+def resolve_ci_decision(
+    changed_files: list[str] | None = None,
+    *,
+    diff_range: str | None = None,
+) -> CiDecision:
+    if changed_files is None:
+        return _finish(CiDecision(), "could not resolve changed files; run normal CI")
+
+    buckets = _classify_changed_files_into_buckets(changed_files, diff_range=diff_range)
+    if buckets.has_other_changes or not buckets.has_skip_ci_scope_changes:
+        return _finish(CiDecision(), "product or non-whitelisted changes present; run normal CI")
+
+    if CiDecision.skip_all_basis(buckets, diff_range=diff_range):
+        return _finish(
+            CiDecision(skip_all=True),
+            "CI skipped — docs or pytest skip-mark changes only",
+        )
+
+    # Docs / skip-mark mixed with whitelisted CI YAML → same yaml-gated path
+    # as CI-YAML-only (docs/skip-mark do not widen to normal CI).
+    if not buckets.has_l23_yaml_changes and not buckets.has_l45_yaml_changes:
+        if buckets.has_skip_test_changes and diff_range is None:
+            message = (
+                "only pytest skip-mark test changes detected, but no git diff range "
+                "is available to verify them (skip-all requires diff_range); run normal CI"
+            )
+        elif buckets.has_skip_test_changes:
+            message = "pytest skip-mark test changes did not qualify for skip-all and no ci yaml changed; run normal CI"
+        else:
+            message = "doc/skip-mark changes did not qualify for skip-all; run normal CI"
+        return _finish(CiDecision(), message)
+
+    decision = CiDecision.from_yaml_basis(
+        SkipL2L3Basis.from_buckets(buckets),
+        SkipL3Basis.from_buckets(buckets),
+        SkipL2Basis.from_buckets(buckets),
+    )
+    return _finish(decision, _format_yaml_gated_message(buckets, decision))
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser(
+        description=__doc__,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    subparsers = parser.add_subparsers(dest="command", required=True)
+
+    gate_parser = subparsers.add_parser(
+        "gate",
+        help="exit 0 when bootstrap should stop (skip-all or PLATFORM/LEVEL disabled); logs once",
+    )
+    gate_parser.add_argument("platform", help="cuda, amd, intel, or npu")
+    gate_parser.add_argument("level", choices=("l2", "l3"), help="test level")
+
+    subparsers.add_parser(
+        "check-skip-all",
+        help="exit 0 when docs/skip-mark-only diff should skip entire CI upload",
+    )
+
+    skip_parser = subparsers.add_parser(
+        "check-skip-l2-l3",
+        help="exit 0 when PLATFORM l2/l3 pipeline should be skipped (CI config YAML-only diffs)",
+    )
+    skip_parser.add_argument("platform", help="cuda, amd, intel, or npu")
+    skip_parser.add_argument("level", choices=("l2", "l3"), help="test level")
+
+    subparsers.add_parser(
+        "print-annotate",
+        help="print human-readable skip reason for Buildkite annotation",
+    )
+
+    args = parser.parse_args()
+    decision = resolve_ci_context_from_git().decision
+
+    if args.command == "gate":
+        if decision.skip_all:
+            print("skip-all")
+            return 0
+        if not decision.is_run(args.platform, args.level):
+            print("skip-l23")
+            return 0
+        return 1
+
+    if args.command == "check-skip-all":
+        return 0 if decision.skip_all else 1
+
+    if args.command == "check-skip-l2-l3":
+        return 0 if not decision.is_run(args.platform, args.level) else 1
+
+    if args.command == "print-annotate":
+        print(decision.message or "CI will run normally")
+        return 0
+
+    return 1
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

.buildkite/common/scripts/upload_pipeline.py

@@ -0,0 +1,449 @@
+#!/usr/bin/env python3
+# SPDX-License-Identifier: Apache-2.0
+"""Render and optionally upload Buildkite pipeline YAML with diff-aware logic.
+
+Bootstrap mode (``bootstrap-upload-steps.yml``):
+  - Hook uploads the entry YAML (``pipeline.yml`` / ``pipeline-npu.yml``) with one step that runs
+    ``upload_pipeline.py --upload <platform>/bootstrap-upload-steps.yml``.
+  - Injects ``if`` by step ``key`` from skip-ci and uploads child steps (image build, L2–L5 upload).
+  - Detect docs-only, pytest skip-mark-only, or combined skip-ci from git diff.
+  - When only CI level YAML changes, enable **L2/L3** upload steps for affected levels only.
+
+Test pipeline mode (e.g. test-merge.yml):
+  - Drop steps whose ``source_file_dependencies`` do not match changed files.
+  - Expand uploader-only ``mirror_hardwares: l4_1`` into ``agents`` (+ optional ``image``
+    for NPU) + ``plugins`` (see ci_mirror_hardwares.yml).
+
+Usage:
+  python3 upload_pipeline.py [--upload] [--all | --e2e] <pipeline.yml>
+
+Requires PyYAML (``pip install pyyaml``); installs it automatically when missing.
+"""
+
+from __future__ import annotations
+
+import argparse
+import copy
+import subprocess
+import sys
+from functools import lru_cache
+from pathlib import Path
+from typing import Any
+
+try:
+    import yaml
+except ModuleNotFoundError:
+    subprocess.run(
+        [sys.executable, "-m", "pip", "install", "-q", "pyyaml"],
+        check=True,
+    )
+    import yaml
+
+from skip_ci import (
+    ROOT,
+    resolve_ci_context_from_git,
+)
+
+# --- Constants ---
+
+LOG = "upload_pipeline"
+BOOTSTRAP_STEPS_FILENAME = "bootstrap-upload-steps.yml"
+BOOTSTRAP_IMAGE_BUILD_KEYS = frozenset({"image-build", "image-build-a2", "image-build-a3"})
+BOOTSTRAP_UPLOAD_IF_KEYS = {
+    "upload-ready-pipeline": "ready",
+    "upload-merge-pipeline": "merge",
+    "upload-nightly-pipeline": "nightly",
+    "upload-weekly-pipeline": "weekly",
+}
+E2E_GROUP_MARKER = "E2E Test"
+CI_MIRROR_HARDWARES_PATH = ROOT / ".buildkite/common/ci_mirror_hardwares.yml"
+
+CUDA_NIGHTLY_ONLY = (
+    '(build.pull_request.labels includes "nightly-test") || (build.branch == "main" && build.env("NIGHTLY") == "1")'
+)
+NPU_NIGHTLY_ONLY = (
+    '(build.branch == "main" && build.env("NIGHTLY") == "1") || '
+    '(build.branch != "main" && ('
+    'build.pull_request.labels includes "nightly-test" || '
+    'build.pull_request.labels includes "omni-test" || '
+    'build.pull_request.labels includes "tts-test" || '
+    'build.pull_request.labels includes "diffusion-x2iat-test" || '
+    'build.pull_request.labels includes "diffusion-x2v-test"'
+    "))"
+)
+
+
+# --- Logging ---
+
+
+def _log(message: str) -> None:
+    print(f"{LOG}: {message}", file=sys.stderr)
+
+
+# --- Bootstrap pipeline (bootstrap-upload-steps.yml) ---
+
+
+def _get_bootstrap_platform(path: Path) -> str:
+    parts = path.as_posix().split("/")
+    return "npu" if "npu" in parts else "cuda"
+
+
+def _load_bootstrap_steps(path: Path) -> str:
+    if path.name != BOOTSTRAP_STEPS_FILENAME:
+        raise ValueError(f"expected {BOOTSTRAP_STEPS_FILENAME}, got {path.name}")
+    return path.read_text(encoding="utf-8")
+
+
+def _format_bootstrap_if(expr: str) -> str:
+    """Return a Buildkite ``if`` string. Buildkite rejects YAML bool ``if`` values."""
+    if expr in ("true", "false"):
+        return expr
+    return f"({expr})"
+
+
+def _compute_bootstrap_if_exprs(*, decision, platform: str) -> dict[str, str]:
+    disabled = "false"
+    nightly_main = 'build.branch == "main" && build.env("NIGHTLY") == "1"'
+    nightly_only = NPU_NIGHTLY_ONLY if platform == "npu" else CUDA_NIGHTLY_ONLY
+
+    if platform == "npu":
+        ready_pr = (
+            'build.branch != "main" && ('
+            'build.pull_request.labels includes "npu-test" || '
+            'build.pull_request.labels includes "ready"'
+            ")"
+        )
+        nightly_label_if = nightly_only
+        weekly_label_if = disabled
+        merge_base = disabled
+    else:
+        ready_pr = 'build.branch != "main" && build.pull_request.labels includes "ready"'
+        merge_main = 'build.branch == "main" && build.env("NIGHTLY") != "1" && build.env("WEEKLY") != "1"'
+        merge_pr = 'build.branch != "main" && build.pull_request.labels includes "merge-test"'
+        nightly_label_if = (
+            '(build.branch == "main" && build.env("NIGHTLY") == "1") || '
+            '(build.branch != "main" && ('
+            'build.pull_request.labels includes "nightly-test" || '
+            'build.pull_request.labels includes "omni-test" || '
+            'build.pull_request.labels includes "tts-test" || '
+            'build.pull_request.labels includes "diffusion-x2iat-test" || '
+            'build.pull_request.labels includes "diffusion-x2v-test"'
+            "))"
+        )
+        weekly_label_if = (
+            '(build.branch == "main" && build.env("WEEKLY") == "1") || '
+            '(build.branch != "main" && build.pull_request.labels includes "weekly-test")'
+        )
+        merge_base = f"({nightly_main}) || (({merge_main}) || ({merge_pr}))"
+
+    ready_base = f"({nightly_main}) || ({ready_pr})"
+    weekly_main = 'build.branch == "main" && build.env("WEEKLY") == "1"'
+
+    if decision.skip_all:
+        # Docs / skip-mark only: no PR-label escape hatch. Main scheduled
+        # NIGHTLY=1 still runs L4 + L2/L3 (--e2e); WEEKLY=1 still runs L5.
+        image_expr = f"({nightly_main}) || ({weekly_main})" if platform == "cuda" else nightly_main
+        ready_expr = nightly_main
+        merge_expr = nightly_main if platform == "cuda" else disabled
+        nightly_expr = nightly_main
+        weekly_expr = weekly_main if platform == "cuda" else disabled
+    elif decision.skip_l2_l3:
+        l2_enabled = decision.is_run("npu", "l2") if platform == "npu" else decision.is_run("cuda", "l2")
+        l3_enabled = platform == "cuda" and decision.is_run("cuda", "l3")
+
+        ready_expr = ready_base if l2_enabled else disabled
+        merge_expr = merge_base if l3_enabled else disabled
+        nightly_expr = nightly_label_if
+        weekly_expr = weekly_label_if if platform == "cuda" else disabled
+
+        image_parts = [f"({nightly_label_if})"]
+        if platform == "cuda":
+            image_parts.append(f"({weekly_label_if})")
+        if l2_enabled:
+            image_parts.insert(0, f"({ready_base})")
+        if l3_enabled:
+            image_parts.insert(1 if l2_enabled else 0, f"({merge_base})")
+        image_expr = " || ".join(image_parts)
+    else:
+        image_expr = "true"
+        ready_expr = ready_base
+        merge_expr = merge_base if platform == "cuda" else disabled
+        nightly_expr = nightly_label_if
+        weekly_expr = weekly_label_if if platform == "cuda" else disabled
+
+    return {
+        "image": _format_bootstrap_if(image_expr),
+        "ready": _format_bootstrap_if(ready_expr),
+        "merge": _format_bootstrap_if(merge_expr),
+        "nightly": _format_bootstrap_if(nightly_expr),
+        "weekly": _format_bootstrap_if(weekly_expr),
+    }
+
+
+def _apply_bootstrap_if(steps: list[Any], if_exprs: dict[str, str]) -> list[Any]:
+    """Inject ``if`` by step key; drop steps that are unconditionally disabled."""
+    kept: list[Any] = []
+    for step in steps:
+        if not isinstance(step, dict):
+            kept.append(step)
+            continue
+        nested = step.get("steps")
+        if isinstance(nested, list):
+            nested_kept = _apply_bootstrap_if(nested, if_exprs)
+            if nested_kept:
+                kept.append({**step, "steps": nested_kept})
+            continue
+        key = step.get("key")
+        if key in BOOTSTRAP_IMAGE_BUILD_KEYS:
+            step["if"] = if_exprs["image"]
+        elif key in BOOTSTRAP_UPLOAD_IF_KEYS:
+            step["if"] = if_exprs[BOOTSTRAP_UPLOAD_IF_KEYS[key]]
+        if_expr = step.get("if")
+        if if_expr == "false":
+            _log(f"omit disabled bootstrap step {key!r}")
+            continue
+        if if_expr == "true":
+            # Unconditional step: omit ``if`` (Buildkite requires string ``if``, not YAML bool).
+            step.pop("if", None)
+        kept.append(step)
+    return kept
+
+
+def _render_bootstrap_pipeline(
+    steps_yaml: str,
+    *,
+    decision,
+    path: Path,
+) -> str:
+    """Load bootstrap steps YAML and inject ``if`` expressions from skip-ci decision."""
+    doc = yaml.safe_load(steps_yaml)
+    if not isinstance(doc, dict):
+        raise ValueError(f"invalid bootstrap steps YAML for {path}")
+    steps = doc.get("steps")
+    if not isinstance(steps, list):
+        raise ValueError(f"bootstrap steps YAML must contain steps: list in {path}")
+
+    platform = _get_bootstrap_platform(path)
+    if_exprs = _compute_bootstrap_if_exprs(decision=decision, platform=platform)
+    doc["steps"] = _apply_bootstrap_if(steps, if_exprs)
+    return yaml.safe_dump(doc, sort_keys=False)
+
+
+# --- Test pipeline (test-ready.yml, test-merge.yml) ---
+
+
+@lru_cache(maxsize=1)
+def _load_mirror_hardwares() -> dict[str, dict[str, Any]]:
+    if not CI_MIRROR_HARDWARES_PATH.is_file():
+        raise FileNotFoundError(f"missing CI mirror_hardwares registry: {CI_MIRROR_HARDWARES_PATH}")
+    doc = yaml.safe_load(CI_MIRROR_HARDWARES_PATH.read_text(encoding="utf-8"))
+    if not isinstance(doc, dict):
+        raise ValueError(f"invalid CI mirror_hardwares registry: {CI_MIRROR_HARDWARES_PATH}")
+    presets = doc.get("mirror_hardwares")
+    if not isinstance(presets, dict):
+        raise ValueError(f"mirror_hardwares must be a mapping in {CI_MIRROR_HARDWARES_PATH}")
+    return presets
+
+
+def _expand_mirror_hardwares(step: dict[str, Any]) -> dict[str, Any]:
+    """Replace uploader-only ``mirror_hardwares`` with preset fields from ci_mirror_hardwares.yml."""
+    hardware = step.get("mirror_hardwares")
+    if hardware is None:
+        return step
+
+    if not isinstance(hardware, str) or not hardware.strip():
+        raise ValueError(
+            f"mirror_hardwares must be a non-empty string in step {_get_step_label(step)!r}",
+        )
+
+    preset = _load_mirror_hardwares().get(hardware)
+    if preset is None:
+        known = ", ".join(sorted(_load_mirror_hardwares()))
+        raise ValueError(
+            f"unknown mirror_hardwares {hardware!r} in step {_get_step_label(step)!r}; known: {known}",
+        )
+
+    if step.get("agents") is not None or step.get("plugins") is not None or step.get("image") is not None:
+        raise ValueError(
+            f"step {_get_step_label(step)!r} sets mirror_hardwares together with agents/plugins/image; "
+            "use mirror_hardwares only",
+        )
+
+    expanded = copy.deepcopy(preset)
+    return {key: value for key, value in step.items() if key != "mirror_hardwares"} | expanded
+
+
+def _match_source_file(changed_files: list[str], prefixes: list[str]) -> bool:
+    for path in changed_files:
+        for prefix in prefixes:
+            normalized = prefix.rstrip("/")
+            if path == normalized or path.startswith(f"{normalized}/"):
+                return True
+    return False
+
+
+def _get_step_label(step: dict[str, Any]) -> str:
+    return str(step.get("group") or step.get("label") or "<step>")
+
+
+def _process_test_steps(
+    steps: list[Any],
+    changed_files: list[str] | None,
+) -> list[Any]:
+    """Drop steps by ``source_file_dependencies`` when *changed_files* is set; always strip that field."""
+    processed: list[Any] = []
+    for step in steps:
+        if not isinstance(step, dict):
+            processed.append(step)
+            continue
+
+        deps = step.get("source_file_dependencies")
+        if deps is not None and not isinstance(deps, list):
+            raise ValueError(
+                f"source_file_dependencies must be a list in step {_get_step_label(step)!r}",
+            )
+        if changed_files is not None and deps is not None and not _match_source_file(changed_files, deps):
+            _log(f"skip {_get_step_label(step)!r} (no changes under {deps})")
+            continue
+
+        nested = step.get("steps")
+        if nested is not None:
+            kept_nested = _process_test_steps(nested, changed_files)
+            if changed_files is not None and not kept_nested:
+                _log(f"omit empty group {_get_step_label(step)!r}")
+                continue
+            new_step = {key: value for key, value in step.items() if key != "source_file_dependencies"}
+            new_step["steps"] = kept_nested
+            processed.append(new_step)
+            continue
+
+        if deps is not None:
+            processed.append(
+                _expand_mirror_hardwares(
+                    {key: value for key, value in step.items() if key != "source_file_dependencies"},
+                ),
+            )
+        else:
+            processed.append(_expand_mirror_hardwares(step))
+
+    return processed
+
+
+def _select_e2e_group_steps(steps: list[Any]) -> list[Any]:
+    """Keep only top-level groups whose name contains ``E2E_GROUP_MARKER``."""
+    selected = [
+        step
+        for step in steps
+        if isinstance(step, dict) and isinstance(step.get("group"), str) and E2E_GROUP_MARKER in step["group"]
+    ]
+    if not selected:
+        _log(f"no group matching {E2E_GROUP_MARKER!r} found")
+    else:
+        _log(f"keep {len(selected)} group(s) matching {E2E_GROUP_MARKER!r}")
+    return selected
+
+
+def _render_test_pipeline(
+    doc: dict[str, Any],
+    changed_files: list[str] | None,
+    *,
+    e2e_only: bool = False,
+) -> dict[str, Any]:
+    """Filter steps by PR diff and strip uploader-only ``source_file_dependencies`` metadata."""
+    steps = doc.get("steps")
+    if not isinstance(steps, list):
+        return doc
+    if e2e_only:
+        steps = _select_e2e_group_steps(steps)
+    steps = _process_test_steps(steps, changed_files)
+    return {**doc, "steps": steps}
+
+
+# --- Entry (read file → bootstrap or test render → YAML string) ---
+
+
+def _render_pipeline(
+    path: Path,
+    *,
+    force_all: bool = False,
+    e2e_only: bool = False,
+) -> str:
+    if path.name == BOOTSTRAP_STEPS_FILENAME:
+        ctx = resolve_ci_context_from_git()
+        continuation = _load_bootstrap_steps(path)
+        return _render_bootstrap_pipeline(
+            continuation,
+            decision=ctx.decision,
+            path=path,
+        )
+
+    text = path.read_text(encoding="utf-8")
+    ctx = resolve_ci_context_from_git()
+    if force_all or e2e_only:
+        changed_files = None
+    else:
+        changed_files = ctx.changed_files
+
+    doc = yaml.safe_load(text)
+    if not isinstance(doc, dict):
+        raise ValueError(f"invalid pipeline YAML: {path}")
+
+    doc = _render_test_pipeline(doc, changed_files, e2e_only=e2e_only)
+    return yaml.safe_dump(doc, sort_keys=False)
+
+
+def _upload_to_buildkite(content: str) -> None:
+    subprocess.run(
+        ["buildkite-agent", "pipeline", "upload"],
+        input=content,
+        text=True,
+        check=True,
+    )
+
+
+# --- CLI ---
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument(
+        "pipeline",
+        nargs="?",
+        default=".buildkite/cuda/bootstrap-upload-steps.yml",
+        help="Pipeline YAML path (default: .buildkite/cuda/bootstrap-upload-steps.yml)",
+    )
+    parser.add_argument(
+        "--upload",
+        action="store_true",
+        help="Pipe rendered YAML to buildkite-agent pipeline upload",
+    )
+    mode = parser.add_mutually_exclusive_group()
+    mode.add_argument(
+        "--all",
+        action="store_true",
+        help="Keep all steps (disable diff-aware skipping)",
+    )
+    mode.add_argument(
+        "--e2e",
+        action="store_true",
+        help="Keep only the E2E Test group",
+    )
+    args = parser.parse_args()
+
+    path = Path(args.pipeline)
+    if not path.is_absolute():
+        path = ROOT / path
+    if not path.is_file():
+        _log(f"missing pipeline file: {path}")
+        return 1
+
+    rendered = _render_pipeline(path, force_all=args.all, e2e_only=args.e2e)
+    if args.upload:
+        _upload_to_buildkite(rendered)
+    else:
+        sys.stdout.write(rendered)
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

.buildkite/cuda/bootstrap-upload-steps.yml

@@ -0,0 +1,55 @@
+# Uploaded by upload_pipeline.py --upload (not the Buildkite hook).
+# ``if`` is injected by step ``key`` from skip-ci at upload time.
+steps:
+  - label: ":docker: Build image"
+    key: image-build
+    commands:
+      - "bash .buildkite/cuda/scripts/build-ci-image.sh"
+    agents:
+      queue: "cpu_queue_premerge"
+
+  # L2 Test — PR ready label (diff-aware) or main+NIGHTLY=1 (--e2e, full E2E group)
+  - label: "Upload Ready Pipeline"
+    depends_on: image-build
+    key: upload-ready-pipeline
+    commands:
+      - |
+        EXTRA=""
+        if [ "${NIGHTLY:-}" = "1" ] && [ "${BUILDKITE_BRANCH}" = "main" ]; then
+          EXTRA="--e2e"
+        fi
+        python3 .buildkite/common/scripts/upload_pipeline.py --upload $$EXTRA .buildkite/cuda/test-ready.yml
+    agents:
+      queue: "cpu_queue_premerge"
+
+  # L3 Test — PR merge-test label (diff-aware) or main+NIGHTLY=1 (--e2e, full E2E group)
+  - label: "Upload Merge Pipeline"
+    depends_on: image-build
+    key: upload-merge-pipeline
+    commands:
+      - |
+        EXTRA=""
+        if [ "${NIGHTLY:-}" = "1" ] && [ "${BUILDKITE_BRANCH}" = "main" ]; then
+          EXTRA="--e2e"
+        fi
+        python3 .buildkite/common/scripts/upload_pipeline.py --upload $$EXTRA .buildkite/cuda/test-merge.yml
+    agents:
+      queue: "cpu_queue_premerge"
+
+  # L4 Test — main+NIGHTLY=1 (scheduled), or PR with specific label (e.g. add label then Rebuild)
+  - label: "Upload Nightly Pipeline"
+    depends_on: image-build
+    key: upload-nightly-pipeline
+    commands:
+      - python3 .buildkite/common/scripts/upload_pipeline.py --upload .buildkite/cuda/test-nightly.yml
+    agents:
+      queue: "cpu_queue_premerge"
+
+  # L5 Test — main+WEEKLY=1 (scheduled), or PR with weekly-test label
+  - label: "Upload Weekly Pipeline"
+    depends_on: image-build
+    key: upload-weekly-pipeline
+    commands:
+      - python3 .buildkite/common/scripts/upload_pipeline.py --upload .buildkite/cuda/test-weekly.yml
+    agents:
+      queue: "cpu_queue_premerge"

.buildkite/cuda/pipeline.yml

@@ -0,0 +1,9 @@
+# Hook uploads this entry YAML; upload_pipeline.py --upload reads bootstrap-upload-steps.yml
+# and injects ``if`` by step ``key`` from skip-ci.
+steps:
+  - label: ":github: Resolve skip-ci & upload pipeline"
+    key: upload-ci-pipeline
+    commands:
+      - "python3 .buildkite/common/scripts/upload_pipeline.py --upload .buildkite/cuda/bootstrap-upload-steps.yml"
+    agents:
+      queue: "cpu_queue_premerge"

.buildkite/cuda/rebase-pipeline.yml

@@ -2,10 +2,7 @@ steps:
   - label: ":docker: Build image"
     key: image-build
     commands:
-      - "aws ecr-public get-login-password --region us-east-1 | docker login --username AWS --password-stdin public.ecr.aws/q9t5s3a7"
-      - "docker build --progress=plain --file docker/Dockerfile.ci -t vllm-omni-ci ."
-      - "docker tag vllm-omni-ci public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT"
-      - "docker push public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT"
+      - "bash .buildkite/cuda/scripts/build-ci-image.sh"
     agents:
       queue: "cpu_queue_premerge"
 
@@ -14,7 +11,7 @@ steps:
     depends_on: image-build
     key: upload-ready-pipeline
     commands:
-      - buildkite-agent pipeline upload .buildkite/test-ready.yml
+      - python3 .buildkite/common/scripts/upload_pipeline.py --upload --all .buildkite/cuda/test-ready.yml
     agents:
       queue: "cpu_queue_premerge"
 
@@ -23,7 +20,7 @@ steps:
     depends_on: image-build
     key: upload-merge-pipeline
     commands:
-      - buildkite-agent pipeline upload .buildkite/test-merge.yml
+      - python3 .buildkite/common/scripts/upload_pipeline.py --upload --all .buildkite/cuda/test-merge.yml
     agents:
       queue: "cpu_queue_premerge"
 
@@ -33,6 +30,6 @@ steps:
     key: upload-nightly-pipeline
     if: build.env("NIGHTLY") == "1"
     commands:
-      - buildkite-agent pipeline upload .buildkite/test-nightly.yml
+      - python3 .buildkite/common/scripts/upload_pipeline.py --upload --all .buildkite/cuda/test-nightly.yml
     agents:
       queue: "cpu_queue_premerge"

.buildkite/cuda/scripts/build-ci-image.sh

@@ -0,0 +1,52 @@
+#!/bin/bash
+set -euo pipefail
+
+# Build and push the CI Docker image with registry-based layer caching.
+#
+# Usage: build-ci-image.sh
+#
+# Expects the following environment variables (set by Buildkite):
+#   BUILDKITE_COMMIT          - commit SHA used as the image tag
+ECR_NAMESPACE="public.ecr.aws/q9t5s3a7"
+REGISTRY="${ECR_NAMESPACE}/vllm-ci-test-repo"
+REGION="us-east-1"
+DOCKERFILE="docker/Dockerfile.ci"
+BUILDER_NAME="vllm-omni-builder"
+
+# Ensure that the env vars are actually set, otherwise exit early
+if [ -z "${BUILDKITE_COMMIT:-}" ]; then
+    echo "ERROR: BUILDKITE_COMMIT is not set"
+    exit 1
+fi
+echo "BUILDKITE_COMMIT: ${BUILDKITE_COMMIT}"
+
+# Authenticate to ECR Public
+aws ecr-public get-login-password --region "$REGION" \
+    | docker login --username AWS --password-stdin "$ECR_NAMESPACE"
+
+# Compute the cache tag for the dependencies image as the hash of the contents.
+# This is because dependencies don't change often, so most PRs will have the same hash.
+#
+# The Dockerfile is also included to avoid churn in the cached image based on the vLLM version.
+DEP_FILES="pyproject.toml setup.py requirements/*.txt docker/Dockerfile.ci"
+CACHE_KEY=$(cat ${DEP_FILES} | sha256sum | cut -c1-16)
+CACHE_TAG="deps-cache-${CACHE_KEY}"
+echo "Cache key: ${CACHE_TAG}"
+
+# Set up buildx with docker-container driver; we need to do this
+# since cache export is not supported for the default docker driver
+# that is running in the CI.
+docker buildx inspect $BUILDER_NAME >/dev/null 2>&1 \
+|| docker buildx create --name $BUILDER_NAME --driver docker-container
+docker buildx use $BUILDER_NAME
+
+echo "Building image tag: ${REGISTRY}:${BUILDKITE_COMMIT}"
+
+# Build + push the image to the registry, using the hash from the dependencies.
+# Note that the caching here is layerwise, so we're just making sure we always
+# can cache hit on everything but the last copy at the moment.
+docker buildx build --push --progress=plain \
+    --cache-from "type=registry,ref=${REGISTRY}:${CACHE_TAG}" \
+    --cache-to "type=registry,ref=${REGISTRY}:${CACHE_TAG},mode=max,compression=zstd" \
+    --file "$DOCKERFILE" \
+    -t "${REGISTRY}:${BUILDKITE_COMMIT}" .

.buildkite/cuda/test-merge.yml

@@ -0,0 +1,278 @@
+env:
+  VLLM_WORKER_MULTIPROC_METHOD: spawn
+  VLLM_USE_FLASHINFER_MOE_FP16: 0
+  HF_HUB_DOWNLOAD_TIMEOUT: 300
+  HF_HUB_ETAG_TIMEOUT: 60
+
+steps:
+  - group: ":card_index_dividers: Simple Test"
+    depends_on: upload-merge-pipeline
+    steps:
+      - label: "Simple · Diffusion & Model Executor Test"
+        commands:
+          - "timeout 20m pytest -sv tests/diffusion tests/model_executor -m 'core_model and cpu' --cov=vllm_omni --cov-report=term-missing:skip-covered --cov-report=xml"
+        mirror_hardwares: l4_1
+
+      - label: "Simple · Other Test"
+        commands:
+          - "timeout 20m pytest -sv -m 'core_model and cpu' --ignore=tests/diffusion --ignore=tests/model_executor --cov=vllm_omni --cov-report=term-missing:skip-covered --cov-report=xml"
+        mirror_hardwares: l4_1
+
+  - group: ":card_index_dividers: Diffusion Test"
+    depends_on: upload-merge-pipeline
+    steps:
+      - label: "Diffusion · LoRA Test"
+        timeout_in_minutes: 30
+        commands:
+          - pytest -sv tests/diffusion/lora/ -m 'advanced_model and cuda' --run-level "advanced_model"
+        mirror_hardwares: l4_1
+
+      - label: "Diffusion · Offloader Test"
+        timeout_in_minutes: 40
+        commands:
+          - timeout 40m pytest -s -v tests/diffusion/offloader -m "advanced_model and cuda"
+        mirror_hardwares: l4_1
+
+  - label: "Distributed Test"
+    depends_on: upload-merge-pipeline
+    timeout_in_minutes: 60
+    commands:
+      - |
+        timeout 55m bash -c "
+          set -e
+          export VLLM_IMAGE_FETCH_TIMEOUT=60
+          pytest -s -v tests/distributed -m 'advanced_model and cuda and H100' --run-level 'advanced_model'
+        "
+    mirror_hardwares: h100_1
+
+  - group: ":card_index_dividers: Entrypoints Test"
+    depends_on: upload-merge-pipeline
+    steps:
+      - label: "Entrypoint Test with H100"
+        # TEMPORARY 60 -> 90: engine startup/sleep-wake is ~13-15 min slower per
+        # run on the vLLM 0.26 image (model loading ~2x, init ~2x; see builds
+        # 2915 vs 2916), which pushed this suite past 60 min. Revert once the
+        # 0.26 startup regression is understood and fixed.
+        timeout_in_minutes: 90
+        commands:
+          - pytest -s -v tests/entrypoints/ -m "advanced_model and cuda and H100" --run-level "advanced_model"
+        mirror_hardwares: h100_2
+
+      - label: "Entrypoint Test with L4"
+        timeout_in_minutes: 60
+        commands:
+          - pytest -s -v tests/entrypoints/ -m "advanced_model and cuda and L4" --run-level "advanced_model"
+        mirror_hardwares: l4_1
+
+  - group: ":card_index_dividers: E2E Test"
+    depends_on: upload-merge-pipeline
+    steps:
+      - label: "TTS · Qwen3-TTS CustomVoice Test"
+        source_file_dependencies:
+          - tests/e2e/online_serving/test_qwen3_tts_customvoice.py
+          - tests/e2e/offline_inference/test_qwen3_tts_customvoice.py
+          - vllm_omni/model_executor/models/qwen3_tts/
+          - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
+          - vllm_omni/deploy/qwen3_tts.yaml
+        commands:
+          - |
+            timeout 40m bash -c "
+              export VLLM_LOGGING_LEVEL=DEBUG
+              export VLLM_ALLOW_LONG_MAX_MODEL_LEN=1
+              pytest -s -v tests/e2e/online_serving/test_qwen3_tts_customvoice.py tests/e2e/offline_inference/test_qwen3_tts_customvoice.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+            "
+        mirror_hardwares: l4_1
+
+      - label: "TTS · Qwen3-TTS Base Test"
+        source_file_dependencies:
+          - tests/e2e/online_serving/test_qwen3_tts_base.py
+          - tests/e2e/offline_inference/test_qwen3_tts_base.py
+          - vllm_omni/model_executor/models/qwen3_tts/
+          - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
+          - vllm_omni/deploy/qwen3_tts.yaml
+        commands:
+          - |
+            timeout 40m bash -c "
+              export VLLM_LOGGING_LEVEL=DEBUG
+              export VLLM_ALLOW_LONG_MAX_MODEL_LEN=1
+              pytest -s -v tests/e2e/online_serving/test_qwen3_tts_base.py tests/e2e/offline_inference/test_qwen3_tts_base.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+            "
+        mirror_hardwares: l4_1
+
+      - label: "TTS · Higgs-Audio-V3 Test"
+        source_file_dependencies:
+          - tests/e2e/online_serving/test_higgs_audio_v3.py
+          - vllm_omni/model_executor/models/higgs_audio_v3/
+          - vllm_omni/model_executor/stage_input_processors/higgs_audio_v3.py
+          - vllm_omni/deploy/higgs_multimodal_qwen3.yaml
+        commands:
+          - |
+            timeout 20m bash -c "
+              export VLLM_WORKER_MULTIPROC_METHOD=spawn
+              export VLLM_USE_DEEP_GEMM=0
+              export VLLM_MOE_USE_DEEP_GEMM=0
+              pytest -s -v tests/e2e/online_serving/test_higgs_audio_v3.py -m 'advanced_model' --run-level 'advanced_model'
+            "
+        mirror_hardwares: l4_1
+
+      - label: "Omni · Qwen3-Omni Test"
+        source_file_dependencies:
+          - tests/e2e/offline_inference/test_qwen3_omni.py
+          - tests/e2e/online_serving/test_qwen3_omni.py
+          - vllm_omni/model_executor/models/qwen3_omni/
+          - vllm_omni/model_executor/stage_input_processors/qwen3_omni.py
+          - vllm_omni/deploy/qwen3_omni_moe.yaml
+        timeout_in_minutes: 50
+        commands:
+          - pytest -s -v tests/e2e/offline_inference/test_qwen3_omni.py tests/e2e/online_serving/test_qwen3_omni.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+        mirror_hardwares: h100_2
+
+      - label: "Omni · MiniCPM-o 4.5 Test"
+        source_file_dependencies:
+          - tests/e2e/offline_inference/test_minicpmo_4_5.py
+          - tests/e2e/online_serving/test_minicpmo_4_5.py
+          - vllm_omni/deploy/minicpmo_4_5.yaml
+        timeout_in_minutes: 50
+        commands:
+          - pytest -s -v tests/e2e/offline_inference/test_minicpmo_4_5.py tests/e2e/online_serving/test_minicpmo_4_5.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+        agents:
+          queue: "mithril-h100-pool"
+        plugins:
+          - kubernetes:
+              podSpec:
+                containers:
+                  - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
+                    resources:
+                      limits:
+                        nvidia.com/gpu: 2
+                    volumeMounts:
+                      - name: devshm
+                        mountPath: /dev/shm
+                      - name: hf-cache
+                        mountPath: /root/.cache/huggingface
+                    env:
+                      - name: HF_HOME
+                        value: /root/.cache/huggingface
+                      - name: HF_TOKEN
+                        valueFrom:
+                          secretKeyRef:
+                            name: hf-token-secret
+                            key: token
+                nodeSelector:
+                  node.kubernetes.io/instance-type: gpu-h100-sxm
+                volumes:
+                  - name: devshm
+                    emptyDir:
+                      medium: Memory
+                  - name: hf-cache
+                    hostPath:
+                      path: /mnt/hf-cache
+                      type: DirectoryOrCreate
+
+      - label: "TTS · VoxCPM2 Test"
+        source_file_dependencies:
+          - tests/e2e/offline_inference/test_voxcpm2_tts.py
+          - tests/e2e/online_serving/test_voxcpm2_tts.py
+          - vllm_omni/model_executor/models/voxcpm2/
+          - vllm_omni/deploy/voxcpm2.yaml
+        timeout_in_minutes: 30
+        commands:
+          - |
+            timeout 30m bash -c "
+              export VLLM_LOGGING_LEVEL=DEBUG
+              export VLLM_WORKER_MULTIPROC_METHOD=spawn
+              pytest -s -v tests/e2e/offline_inference/test_voxcpm2_tts.py tests/e2e/online_serving/test_voxcpm2_tts.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+            "
+        mirror_hardwares: l4_1
+
+      - label: "TTS · SoulX-Singer Test"
+        source_file_dependencies:
+          - tests/e2e/offline_inference/test_soulxsinger.py
+          - vllm_omni/diffusion/models/soulx_singer/
+          - vllm_omni/deploy/soulxsinger_svc.yaml
+          - vllm_omni/deploy/soulxsinger_svs.yaml
+        timeout_in_minutes: 30
+        commands:
+          - |
+            timeout 30m bash -c "
+              export VLLM_LOGGING_LEVEL=DEBUG
+              export VLLM_WORKER_MULTIPROC_METHOD=spawn
+              pytest -s -v tests/e2e/offline_inference/test_soulxsinger.py -k 'svc or precomputed' -m 'advanced_model and cuda' --run-level 'advanced_model'
+            "
+        mirror_hardwares: l4_1
+
+      - label: "Diffusion · Qwen Image Edit Test"
+        source_file_dependencies:
+          - tests/e2e/online_serving/test_qwen_image_edit.py
+          - vllm_omni/diffusion/models/qwen_image/
+        timeout_in_minutes: 20
+        commands:
+          - pytest -s -v tests/e2e/online_serving/test_qwen_image_edit.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+        mirror_hardwares: h100_1
+      - label: "Diffusion · Bagel Test"
+        source_file_dependencies:
+          - tests/e2e/offline_inference/test_bagel.py
+          - tests/e2e/online_serving/test_bagel.py
+          - vllm_omni/model_executor/models/bagel/
+          - vllm_omni/diffusion/models/bagel/
+          - vllm_omni/model_executor/stage_input_processors/bagel.py
+          - vllm_omni/deploy/bagel.yaml
+        timeout_in_minutes: 20
+        commands:
+          - pytest -s -v tests/e2e/offline_inference/test_bagel.py tests/e2e/online_serving/test_bagel.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · Qwen Image Test"
+        source_file_dependencies:
+          - tests/e2e/online_serving/test_qwen_image.py
+          - vllm_omni/diffusion/models/qwen_image/
+        commands:
+          - |
+            timeout 40m bash -c '
+              export VLLM_IMAGE_FETCH_TIMEOUT=60
+              pytest -s -v tests/e2e/online_serving/test_qwen_image.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+            '
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · Qwen Image Layered Test"
+        source_file_dependencies:
+          - tests/e2e/online_serving/test_qwen_image_layered.py
+          - vllm_omni/diffusion/models/qwen_image/
+        commands:
+          - |
+            timeout 40m bash -c "
+              export VLLM_IMAGE_FETCH_TIMEOUT=60
+              pytest -s -v tests/e2e/online_serving/test_qwen_image_layered.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+            "
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · Wan22 Test"
+        source_file_dependencies:
+          - tests/e2e/offline_inference/test_wan22_t2v.py
+          - tests/e2e/online_serving/test_wan22_t2v.py
+          - vllm_omni/diffusion/models/wan2_2/
+        commands:
+          - |
+            timeout 40m bash -c "
+              export VLLM_IMAGE_FETCH_TIMEOUT=60
+              pytest -s -v tests/e2e/offline_inference/test_wan22_t2v.py tests/e2e/online_serving/test_wan22_t2v.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
+            "
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · Z Image Test"
+        source_file_dependencies:
+          - tests/e2e/offline_inference/test_z_image.py
+          - tests/e2e/online_serving/test_z_image.py
+          - vllm_omni/diffusion/models/z_image/
+        timeout_in_minutes: 30
+        commands:
+          - pytest -s -v tests/e2e/offline_inference/test_z_image.py tests/e2e/online_serving/test_z_image.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
+        mirror_hardwares: l4_1
+      - label: "Diffusion · FLUX.2 Klein Test"
+        source_file_dependencies:
+          - tests/e2e/offline_inference/test_flux2_klein.py
+          - vllm_omni/diffusion/models/flux2_klein/
+        timeout_in_minutes: 30
+        commands:
+          - pytest -s -v tests/e2e/offline_inference/test_flux2_klein.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
+        mirror_hardwares: l4_1

.buildkite/cuda/test-nightly.yml

@@ -0,0 +1,534 @@
+env:
+  VLLM_WORKER_MULTIPROC_METHOD: spawn
+  VLLM_USE_FLASHINFER_MOE_FP16: 0
+  HF_HUB_DOWNLOAD_TIMEOUT: 300
+  HF_HUB_ETAG_TIMEOUT: 60
+
+steps:
+  # Group: collapses under one heading in the Buildkite UI; child steps still run in parallel.
+  - group: ":card_index_dividers: Omni Model Test"
+    key: nightly-omni-test-group
+    depends_on: upload-nightly-pipeline
+    if: build.env("NIGHTLY") == "1" || build.pull_request.labels includes "nightly-test" || build.pull_request.labels includes "omni-test"
+    steps:
+
+      - label: ":full_moon: Omni · Function Test with H100"
+        timeout_in_minutes: 120
+        commands:
+          - pytest -sv tests/e2e --run-level "full_model" -m "full_model and H100 and omni" --ignore=tests/e2e/accuracy --ignore=tests/e2e/online_serving/test_qwen3_omni_multi_replicas.py
+        mirror_hardwares: h100_2
+
+      - label: ":full_moon: Omni · Doc Test with L4"
+        timeout_in_minutes: 90
+        commands:
+          - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
+          - pytest -s -v tests/examples/ -m "full_model and omni and L4" --run-level "full_model"
+        mirror_hardwares: l4_4
+
+      - label: ":full_moon: Omni · Doc Test with H100"
+        timeout_in_minutes: 90
+        commands:
+          - pytest -s -v tests/examples/ -m "full_model and omni and H100" --run-level "full_model"
+        mirror_hardwares: h100_2
+
+      - label: ":full_moon: Omni · Accuracy Test"
+        timeout_in_minutes: 180
+        commands:
+          - export SEED_TTS_WER_EVAL=1
+          - export SEED_TTS_EVAL_DEVICE=cuda:1
+          - |
+            set +e
+            pytest -s -v tests/e2e/accuracy/qwen3_omni/test_qwen3_omni.py -m "full_model" --run-level full_model
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/e2e/accuracy/qwen3_omni/results/qwen_omni_acc/*.json"
+            exit $$EXIT
+        mirror_hardwares: h100_2
+
+      - label: ":full_moon: Omni · Perf Test · No Async Chunk"
+        key: nightly-omni-performance-no-async-chunk
+        timeout_in_minutes: 300
+        commands:
+          - export BENCHMARK_DIR=tests/dfx/perf/results
+          - |
+            set +e
+            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen3_omni_no_async_chunk.json
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
+            exit $$EXIT
+        mirror_hardwares: h100_2
+
+      - label: ":full_moon: Omni · Perf Test · Async Chunk"
+        key: nightly-omni-performance-async-chunk
+        timeout_in_minutes: 180
+        commands:
+          - export BENCHMARK_DIR=tests/dfx/perf/results
+          - |
+            set +e
+            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen3_omni_async_chunk.json
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
+            exit $$EXIT
+        mirror_hardwares: h100_2
+
+      - label: ":full_moon: Omni · Perf Test · vLLM Text"
+        key: nightly-omni-performance-vllm-text
+        timeout_in_minutes: 180
+        commands:
+          - export BENCHMARK_DIR=tests/dfx/perf/results
+          - |
+            set +e
+            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen3_omni_vllm_text.json
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
+            exit $$EXIT
+        mirror_hardwares: h100_2
+
+      - label: ":full_moon: Omni · Perf Test · Multi-Replica"
+        key: nightly-omni-performance-multi-replicas
+        timeout_in_minutes: 180
+        commands:
+          - export BENCHMARK_DIR=tests/dfx/perf/results
+          - |
+            set +e
+            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen3_omni_multi_replicas.json
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
+            exit $$EXIT
+        mirror_hardwares: h100_3
+
+      - label: ":full_moon: Omni · Multi-Replica Startup Test with 4x H100"
+        timeout_in_minutes: 45
+        commands:
+          - pytest -s -v tests/e2e/online_serving/test_qwen3_omni_multi_replicas.py -m "full_model" --run-level "core_model"
+        mirror_hardwares: h100_4
+
+  - group: ":card_index_dividers: TTS Model Test"
+    key: nightly-tts-test-group
+    depends_on: upload-nightly-pipeline
+    if: build.env("NIGHTLY") == "1" || build.pull_request.labels includes "nightly-test" || build.pull_request.labels includes "tts-test"
+    steps:
+
+      - label: ":full_moon: TTS · Function Test with L4"
+        timeout_in_minutes: 120
+        commands:
+          - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
+          - export VLLM_USE_DEEP_GEMM="0"
+          - export VLLM_MOE_USE_DEEP_GEMM="0"
+          - pytest -s -v tests/e2e/ -m "full_model and L4 and tts" --run-level "full_model" --ignore=tests/e2e/accuracy --ignore=tests/e2e/online_serving/test_voxcpm2_tts_expansion.py
+        mirror_hardwares: l4_1
+
+      - label: ":full_moon: TTS · Accuracy Test · VoxCPM2"
+        timeout_in_minutes: 30
+        commands:
+          - |
+            timeout 30m bash -c "
+              export VLLM_LOGGING_LEVEL=DEBUG
+              export VLLM_WORKER_MULTIPROC_METHOD=spawn
+              export VLLM_ALLOW_LONG_MAX_MODEL_LEN=1
+              pytest -s -v tests/e2e/online_serving/test_voxcpm2_tts_expansion.py -m 'full_model and cuda and tts' --run-level 'full_model'
+            "
+        mirror_hardwares: l4_1
+
+      - label: ":full_moon: TTS · Perf Test"
+        key: nightly-tts-performance
+        timeout_in_minutes: 180
+        commands:
+          - export BENCHMARK_DIR=tests/dfx/perf/results
+          - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
+          - |
+            set +e
+            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_tts.json
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
+            exit $$EXIT
+        mirror_hardwares: h100_2
+
+      - label: ":full_moon: TTS · VoxCPM2 · Perf Test"
+        key: nightly-tts-performance-voxcpm2
+        timeout_in_minutes: 180
+        commands:
+          - export BENCHMARK_DIR=tests/dfx/perf/results
+          - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
+          - |
+            set +e
+            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_voxcpm2.json
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
+            exit $$EXIT
+        mirror_hardwares: h100_1
+
+      - label: ":full_moon: TTS · Higgs-Audio-V3 · Perf Test"
+        key: nightly-tts-performance-higgs-audio-v3
+        timeout_in_minutes: 180
+        commands:
+          - export BENCHMARK_DIR=tests/dfx/perf/results
+          - export VLLM_USE_DEEP_GEMM="0"
+          - export VLLM_MOE_USE_DEEP_GEMM="0"
+          - |
+            set +e
+            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_higgs_audio_v3.json
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
+            exit $$EXIT
+        mirror_hardwares: h100_1
+
+  - group: ":card_index_dividers: Tiny Model Tests [Multi-GPU]"
+    key: nightly-tiny-model-test-group
+    depends_on: upload-nightly-pipeline
+    if: >-
+      build.env("NIGHTLY") == "1" ||
+      build.pull_request.labels includes "nightly-test"
+    steps:
+
+      - label: ":full_moon: Tiny Model Tests · Diffusion (Multi-GPU accelerations)"
+        timeout_in_minutes: 30
+        # NOTE: Single-GPU (core_model) tiny model tests run on the ready label,
+        # so nightly only adds the multi-GPU parallelism configurations.
+        #
+        # This is a bit confusing, but `full_model` is used to mark the multigpu tests, and
+        # --run-level `core_model` is used to indicate that we should use tiny model weights
+        # for the CI, since these tests should pass irrespective of whether or not we use
+        # the full weights or not.
+        commands:
+          - timeout 25m pytest -sv tests/model_tests/diffusion -m 'full_model and cuda' --run-level "core_model"
+        mirror_hardwares: l4_4
+
+  - group: ":card_index_dividers: Diffusion X2I(&A&T) Model Test"
+    key: nightly-diffusion-x2iat-group
+    depends_on: upload-nightly-pipeline
+    if: >-
+      build.env("NIGHTLY") == "1" ||
+      build.pull_request.labels includes "nightly-test" ||
+      build.pull_request.labels includes "diffusion-x2iat-test"
+    steps:
+      # Use explicit file shards so single-card cases do not reserve 2+ GPUs,
+      # and Bagel feature coverage is not duplicated via a broad tests/e2e sweep.
+
+      - label: ":full_moon: Diffusion X2I(&A&T) · Function Test with H100 · Single-GPU"
+        timeout_in_minutes: 120
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_qwen_image_expansion.py
+            tests/e2e/online_serving/test_qwen_image_edit_expansion.py
+            tests/e2e/online_serving/test_qwen_image_layered_expansion.py
+            tests/e2e/online_serving/test_boogu_image.py
+            tests/e2e/online_serving/test_boogu_image_edit.py
+            tests/e2e/online_serving/test_boogu_image_expansion.py
+            tests/e2e/online_serving/test_boogu_image_edit_expansion.py
+            tests/e2e/online_serving/test_bagel_expansion.py
+            tests/e2e/online_serving/test_krea2_expansion.py
+            tests/e2e/offline_inference/test_krea2_expansion.py
+            -m "full_model and diffusion and H100 and not distributed_cuda"
+            --run-level "full_model"
+        mirror_hardwares: h100_1
+
+      - label: ":full_moon: Diffusion X2I(&A&T) · Function Test with H100 · Multi-GPU Qwen-Image"
+        timeout_in_minutes: 120
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_qwen_image_expansion.py
+            tests/e2e/online_serving/test_qwen_image_edit_expansion.py
+            tests/e2e/offline_inference/test_krea2_expansion.py
+            -m "full_model and diffusion and H100 and distributed_cuda"
+            --run-level "full_model"
+        mirror_hardwares: h100_2
+
+      - label: ":full_moon: Diffusion X2I(&A&T) · Function Test with H100 · Multi-GPU Layered/LongCat/FLUX"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_qwen_image_layered_expansion.py
+            -m "full_model and diffusion and H100 and distributed_cuda"
+            --run-level "full_model"
+        mirror_hardwares: h100_2
+
+      - label: ":full_moon: Diffusion X2I(&A&T) · Function Test with H100 · Multi-GPU Bagel"
+        timeout_in_minutes: 120
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_bagel_expansion.py
+            -m "full_model and diffusion and H100 and distributed_cuda"
+            --run-level "full_model"
+        mirror_hardwares: h100_2
+
+      - label: ":full_moon: Diffusion X2I(&A&T) · GLM-Image Function Test with H100"
+        timeout_in_minutes: 120
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/online_serving/test_glm_image_expansion.py
+            tests/e2e/offline_inference/test_glm_image_autoround_w4a16_expansion.py
+            -m "full_model and diffusion and H100"
+            --run-level "full_model"
+        mirror_hardwares: h100_2
+
+      - label: ":full_moon: Diffusion X2I(&A&T) · Function Test with L4"
+        timeout_in_minutes: 120
+        commands:
+          - pytest -sv tests/e2e/ -k "not test_wan and not test_bagel_expansion and not hunyuan" -m "full_model and diffusion and L4" --run-level "full_model" --ignore=tests/e2e/accuracy
+        mirror_hardwares: l4_4
+
+      - label: ":full_moon: Diffusion X2I(&A&T) · Doc Test"
+        timeout_in_minutes: 60
+        commands:
+          - pytest -s -v tests/examples/*/test_text_to_image.py -m "full_model and example and H100" --run-level "full_model"
+        mirror_hardwares: h100_2
+
+      - label: ":full_moon: Diffusion X2I(&A&T) · Accuracy Test"
+        timeout_in_minutes: 180
+        commands:
+          - export VLLM_HTTP_TIMEOUT_KEEP_ALIVE=120
+          - pytest -s -v tests/e2e/accuracy/test_qwen_image*.py --run-level full_model
+          - pytest -s -v tests/e2e/accuracy/test_diffusers_backend_similarity.py -k '2i_matches_diffusers' -m full_model --run-level full_model
+        mirror_hardwares: h100_1
+
+      - label: ":full_moon: HunyuanImage3-DIT · Accuracy Test"
+        timeout_in_minutes: 180
+        commands:
+          - export HUNYUAN_IMAGE3_MODEL="tencent/HunyuanImage-3.0-Instruct"
+          - export HUNYUAN_IMAGE3_DEVICES="0,1,2,3"
+          - pytest -s -v tests/e2e/accuracy/test_hunyuan_image3_pixel_accuracy.py -m full_model --run-level full_model
+        mirror_hardwares: h100_4
+
+      - label: ":full_moon: Diffusion X2I(&A&T) · Perf Test · Qwen-Image"
+        key: nightly-diffusion-x2iat-performance-qwen-image
+        timeout_in_minutes: 180
+        commands:
+          - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
+          - export DIFFUSION_ATTENTION_BACKEND=FLASH_ATTN
+          - export CACHE_DIT_VERSION=1.3.0
+          - |
+            set +e
+            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen_image_vllm_omni.json
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/dfx/perf/results/diffusion_result_*.json"
+            buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
+            exit $$EXIT
+        mirror_hardwares: h100_4
+
+      - label: ":full_moon: Diffusion X2I(&A&T) · Perf Test · Qwen-Image-Edit"
+        key: nightly-diffusion-x2iat-performance-qwen-image-edit
+        timeout_in_minutes: 180
+        commands:
+          - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
+          - export DIFFUSION_ATTENTION_BACKEND=FLASH_ATTN
+          - export CACHE_DIT_VERSION=1.3.0
+          - |
+            set +e
+            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen_image_edit_vllm_omni.json
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/dfx/perf/results/diffusion_result_*.json"
+            buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
+            exit $$EXIT
+        mirror_hardwares: h100_4
+
+      - label: ":full_moon: Diffusion X2I(&A&T) · Perf Test · Qwen-Image-Edit-2511"
+        key: nightly-diffusion-x2iat-performance-qwen-image-edit-2511
+        timeout_in_minutes: 180
+        commands:
+          - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
+          - export DIFFUSION_ATTENTION_BACKEND=FLASH_ATTN
+          - export CACHE_DIT_VERSION=1.3.0
+          - |
+            set +e
+            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen_image_edit_2511_vllm_omni.json
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/dfx/perf/results/diffusion_result_*.json"
+            buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
+            exit $$EXIT
+        mirror_hardwares: h100_4
+
+      - label: ":full_moon: Diffusion X2I(&A&T) · Perf Test · Qwen-Image-Layered"
+        key: nightly-diffusion-x2iat-performance-qwen-image-layered
+        timeout_in_minutes: 120
+        commands:
+          - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
+          - export DIFFUSION_ATTENTION_BACKEND=FLASH_ATTN
+          - export CACHE_DIT_VERSION=1.3.0
+          - |
+            set +e
+            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen_image_layered_vllm_omni.json
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/dfx/perf/results/diffusion_result_*.json"
+            buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
+            exit $$EXIT
+        mirror_hardwares: h100_1
+
+      - label: ":full_moon: Diffusion X2I(&A&T) · Perf Test · BAGEL"
+        key: nightly-diffusion-x2iat-performance-bagel
+        timeout_in_minutes: 180
+        commands:
+          - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
+          - export DIFFUSION_ATTENTION_BACKEND=FLASH_ATTN
+          - export CACHE_DIT_VERSION=1.3.0
+          - |
+            set +e
+            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_bagel_vllm_omni.json
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/dfx/perf/results/diffusion_result_*.json"
+            buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
+            exit $$EXIT
+        mirror_hardwares: h100_1
+
+  - group: ":card_index_dividers: Diffusion X2V Model Test"
+    key: nightly-diffusion-x2v-group
+    depends_on: upload-nightly-pipeline
+    if: >-
+      build.env("NIGHTLY") == "1" ||
+      build.pull_request.labels includes "nightly-test" ||
+      build.pull_request.labels includes "diffusion-x2v-test"
+    steps:
+
+      - label: ":full_moon: Diffusion X2V · Wan2.2 T2V Function Test"
+        timeout_in_minutes: 90
+        commands:
+          - pytest -s -v tests/e2e/online_serving/test_wan22_expansion.py -k "t2v" -m "full_model and cuda" --run-level "full_model"
+        mirror_hardwares: h100_2
+
+      - label: ":full_moon: Diffusion X2V · Wan2.2 I2V/TI2V Function Test"
+        timeout_in_minutes: 90
+        commands:
+          - pytest -s -v tests/e2e/online_serving/test_wan22_expansion.py -k "i2v" -m "full_model and cuda" --run-level "full_model"
+        mirror_hardwares: h100_2
+
+      - label: ":full_moon: Diffusion X2V · LingBot Video MoE Function Test"
+        timeout_in_minutes: 90
+        commands:
+          - pytest -s -v tests/e2e/online_serving/test_lingbot_video_moe.py -m "full_model and cuda" --run-level "full_model"
+        mirror_hardwares: h100_1
+
+      - label: ":full_moon: Diffusion X2V · Doc Test"
+        timeout_in_minutes: 60
+        commands:
+          - pytest -s -v tests/examples/offline_inference/test_image_to_video.py -m "full_model and example and H100" --run-level "full_model"
+        mirror_hardwares: h100_1
+
+      - label: ":full_moon: Diffusion X2V · Other Function Test"
+        timeout_in_minutes: 90
+        commands:
+          - pytest -s -v tests/e2e/offline_inference/test_wan22_autoround_w4a16_expansion.py -m "full_model and cuda" --run-level "full_model"
+          - pytest -s -v tests/e2e/online_serving/test_wan_2_1_vace_expansion.py tests/e2e/online_serving/test_ltx2_expansion.py -m "full_model and cuda" --run-level "full_model"
+        mirror_hardwares: h100_2
+
+      - label: ":full_moon: Diffusion X2V · Wan2.2 Accuracy Test"
+        timeout_in_minutes: 180
+        commands:
+          - pytest -s -v tests/e2e/accuracy/wan22_i2v/test_wan22_i2v_video_similarity.py -m full_model --run-level full_model
+          - pytest -s -v tests/e2e/accuracy/test_diffusers_backend_similarity.py -k '2v_matches_diffusers' -m full_model --run-level full_model
+        mirror_hardwares: h100_2
+
+      - label: ":full_moon: Diffusion X2V · Other Accuracy Test"
+        timeout_in_minutes: 180
+        commands:
+          - pytest -s -v tests/e2e/accuracy/hunyuanvideo15_t2v/test_hunyuanvideo15_t2v_video_similarity.py -m full_model --run-level full_model
+          - pytest -s -v tests/e2e/accuracy/hunyuanvideo15_i2v/test_hunyuanvideo15_i2v_video_similarity.py -m full_model --run-level full_model
+          - pytest -s -v tests/e2e/accuracy/test_ltx2_3_video_similarity.py -m full_model --run-level full_model
+          - pytest -s -v tests/e2e/accuracy/test_video_streaming_output_similarity.py -m full_model --run-level full_model
+        mirror_hardwares: h100_2
+
+      - label: ":full_moon: Diffusion X2V · Perf Test"
+        key: nightly-diffusion-x2v-performance
+        timeout_in_minutes: 180
+        commands:
+          - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
+          - export DIFFUSION_ATTENTION_BACKEND=FLASH_ATTN
+          - |
+            set +e
+            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_wan22_i2v_vllm_omni.json
+            EXIT1=$$?
+            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_cosmos3_vllm_omni.json
+            EXIT2=$$?
+            buildkite-agent artifact upload "tests/dfx/perf/results/diffusion_result_*.json"
+            buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
+            if [ $$EXIT1 -ne 0 ]; then exit $$EXIT1; fi
+            exit $$EXIT2
+        mirror_hardwares: h100_2
+
+  - group: ":card_index_dividers: Diffusion Test"
+    depends_on: upload-nightly-pipeline
+    if: build.env("NIGHTLY") == "1" || build.pull_request.labels includes "nightly-test" || build.pull_request.labels includes "diffusion-x2iat-test"
+    steps:
+
+      - label: ":full_moon: Diffusion · Offloader Test"
+        timeout_in_minutes: 40
+        commands:
+          - timeout 40m pytest -s -v tests/diffusion/offloader -m "full_model and cuda"
+        mirror_hardwares: l4_1
+
+      - label: ":full_moon: Diffusion · Quantization Test with H100"
+        commands:
+          - timeout 60m pytest -sv tests/diffusion/quantization -m 'full_model and cuda and H100' --run-level "full_model"
+        mirror_hardwares: h100_1
+
+      - label: ":full_moon: Diffusion · Quantization Test with L4"
+        timeout_in_minutes: 60
+        commands:
+          - pytest -sv tests/diffusion/quantization -m 'full_model and cuda and L4' --run-level "full_model"
+        mirror_hardwares: l4_1
+
+      - label: ":full_moon: Diffusion · Wan VAE Spatial Shard Correctness Test"
+        timeout_in_minutes: 60
+        commands:
+          - >-
+            pytest -sv tests/diffusion/distributed/test_wan_spatial_shard.py
+            -m "full_model and diffusion and H100 and distributed_cuda"
+            --run-level "full_model"
+        mirror_hardwares: h100_2
+
+  - label: ":bar_chart: Testcase Statistics"
+    key: nightly-testcase-statistics
+    timeout_in_minutes: 120
+    depends_on: upload-nightly-pipeline
+    if: build.env("NIGHTLY") == "1"
+    commands:
+      - python tools/nightly/buildkite_testcase_statistics.py -o tests/dfx/perf/results/buildkite_testcase_statistics.html
+      - buildkite-agent artifact upload "tests/dfx/perf/results/*.html"
+    mirror_hardwares: h100_1
+
+  # NOTE: ":email: Nightly Collection & Email" is deprecated and will be removed soon.
+  # Do not add new depends_on entries or extend this step.
+  - label: ":email: Nightly Collection & Email"
+    key: nightly-perf-distribution
+    depends_on:
+      - nightly-omni-performance-no-async-chunk
+      - nightly-omni-performance-async-chunk
+      - nightly-omni-performance-vllm-text
+      - nightly-omni-performance-multi-replicas
+      - nightly-tts-performance
+      - nightly-tts-performance-higgs-audio-v3
+      - nightly-diffusion-x2iat-performance-qwen-image
+      - nightly-diffusion-x2iat-performance-qwen-image-edit
+      - nightly-diffusion-x2iat-performance-qwen-image-edit-2511
+      - nightly-diffusion-x2iat-performance-qwen-image-layered
+      - nightly-diffusion-x2iat-performance-bagel
+      - nightly-diffusion-x2v-performance
+      - nightly-testcase-statistics
+    if: build.env("NIGHTLY") == "1" && build.env("EMAIL_DISTRIBUTION") == "1"
+    commands:
+      - pip install openpyxl
+      - export DEFAULT_INPUT_DIR=tests/dfx/perf/results
+      - export DEFAULT_OUTPUT_DIR=tests/dfx/perf/results
+      - buildkite-agent artifact download "tests/dfx/perf/results/*.json" . --step nightly-tts-performance
+      - buildkite-agent artifact download "tests/dfx/perf/results/*.json" . --step nightly-tts-performance-higgs-audio-v3
+      - buildkite-agent artifact download "tests/dfx/perf/results/*.json" . --step nightly-omni-performance-no-async-chunk
+      - buildkite-agent artifact download "tests/dfx/perf/results/*.json" . --step nightly-omni-performance-async-chunk
+      - buildkite-agent artifact download "tests/dfx/perf/results/*.json" . --step nightly-omni-performance-vllm-text
+      - buildkite-agent artifact download "tests/dfx/perf/results/*.json" . --step nightly-omni-performance-multi-replicas
+      - buildkite-agent artifact download "tests/dfx/perf/results/*.json" . --step nightly-diffusion-x2iat-performance-qwen-image
+      - buildkite-agent artifact download "tests/dfx/perf/results/*.json" . --step nightly-diffusion-x2iat-performance-qwen-image-edit
+      - buildkite-agent artifact download "tests/dfx/perf/results/*.json" . --step nightly-diffusion-x2iat-performance-qwen-image-edit-2511
+      - buildkite-agent artifact download "tests/dfx/perf/results/*.json" . --step nightly-diffusion-x2iat-performance-qwen-image-layered
+      - buildkite-agent artifact download "tests/dfx/perf/results/*.json" . --step nightly-diffusion-x2iat-performance-bagel
+      - buildkite-agent artifact download "tests/dfx/perf/results/*.json" . --step nightly-diffusion-x2v-performance
+      - buildkite-agent artifact download "tests/dfx/perf/results/*.html" . --step nightly-testcase-statistics
+      - python tools/nightly/generate_nightly_perf_excel.py
+      - python tools/nightly/generate_nightly_perf_html.py
+      - python tools/nightly/send_nightly_email.py --report-file "tests/dfx/perf/results/*.xlsx, tests/dfx/perf/results/*.html"
+      - buildkite-agent artifact upload "tests/dfx/perf/results/*.xlsx"
+      - buildkite-agent artifact upload "tests/dfx/perf/results/*.html"
+    agents:
+      queue: "cpu_queue_premerge"

.buildkite/cuda/test-ready.yml

@@ -0,0 +1,356 @@
+env:
+  VLLM_WORKER_MULTIPROC_METHOD: spawn
+  VLLM_USE_FLASHINFER_MOE_FP16: 0
+  HF_HUB_DOWNLOAD_TIMEOUT: 300
+  HF_HUB_ETAG_TIMEOUT: 60
+
+steps:
+  - group: ":card_index_dividers: Simple Test"
+    depends_on: upload-ready-pipeline
+    steps:
+      - label: "Simple · Diffusion & Model Executor Test"
+        commands:
+          - "timeout 20m pytest -sv tests/diffusion tests/model_executor -m 'core_model and cpu' --cov=vllm_omni --cov-report=term-missing:skip-covered --cov-report=xml"
+        mirror_hardwares: l4_1
+
+      - label: "Simple · Other Test"
+        commands:
+          - "timeout 20m pytest -sv -m 'core_model and cpu' --ignore=tests/diffusion --ignore=tests/model_executor --cov=vllm_omni --cov-report=term-missing:skip-covered --cov-report=xml"
+        mirror_hardwares: l4_1
+
+  - label: "Custom Pipeline Test"
+    depends_on: upload-ready-pipeline
+    commands:
+      - timeout 20m pytest -s -v tests/e2e/offline_inference/custom_pipeline/ -m "core_model"
+    mirror_hardwares: l4_1
+
+  - group: ":card_index_dividers: Diffusion Test"
+    depends_on: upload-ready-pipeline
+    steps:
+
+      - label: "Diffusion · Offloader Test"
+        commands:
+          - timeout 40m pytest -s -v tests/diffusion/offloader -m "core_model and cuda" --run-level "core_model"
+        mirror_hardwares: l4_1
+
+      - label: "Diffusion · Cache Test"
+        commands:
+          - timeout 15m pytest -sv tests/diffusion/cache -m 'core_model and cuda' --run-level "core_model"
+        mirror_hardwares: l4_1
+
+      - label: "Diffusion · Batch Test"
+        commands:
+          - timeout 15m pytest -sv tests/diffusion/batching -m 'core_model and cuda' --run-level "core_model"
+        mirror_hardwares: l4_1
+
+      - label: "Diffusion · Distributed Test"
+        source_file_dependencies:
+          - tests/diffusion/distributed/
+          - vllm_omni/diffusion/distributed/
+          - vllm_omni/diffusion/attention/parallel/
+          - vllm_omni/diffusion/data.py
+          - vllm_omni/diffusion/models/
+        commands:
+          - timeout 20m pytest -sv tests/diffusion/distributed -m 'core_model and cuda' --run-level "core_model"
+        mirror_hardwares: l4_4
+
+      - label: "Diffusion · Model Test"
+        commands:
+          - timeout 15m pytest -sv tests/diffusion/models/ -m 'core_model and cuda' --run-level "core_model"
+        mirror_hardwares: l4_1
+
+      - label: "Diffusion · Other Test"
+        commands:
+          - timeout 15m pytest -sv tests/diffusion/*.py -m 'core_model and cuda' --run-level "core_model"
+        mirror_hardwares: l4_1
+
+  - group: ":card_index_dividers: Tiny Model Tests [Single GPU]"
+    depends_on: upload-ready-pipeline
+    steps:
+
+      - label: "Tiny Model Tests · Diffusion (base)"
+        commands:
+          - timeout 15m pytest -sv tests/model_tests/diffusion -m 'core_model and cuda' --run-level "core_model"
+        mirror_hardwares: l4_1
+
+  - label: "Engine&Model Executor Test"
+    depends_on: upload-ready-pipeline
+    commands:
+      - |
+        timeout 15m bash -c "
+                  pytest -sv tests/engine/ tests/model_executor/ -m 'core_model and cuda' --run-level 'core_model'
+        "
+    mirror_hardwares: l4_1
+
+  - label: "Distributed Test"
+    depends_on: upload-ready-pipeline
+    commands:
+      - |
+        timeout 20m bash -c "
+          pytest -sv tests/distributed/ -m 'core_model and cuda and L4' --run-level 'core_model'
+        "
+    mirror_hardwares: l4_1
+
+  - label: "Entrypoints Test"
+    depends_on: upload-ready-pipeline
+    commands:
+      - |
+        timeout 30m bash -c "
+          export VLLM_WORKER_MULTIPROC_METHOD=spawn
+          pytest -sv tests/entrypoints/ -m 'core_model and cuda' --run-level 'core_model'
+        "
+    mirror_hardwares: l4_1
+
+  - group: ":card_index_dividers: E2E Test"
+    depends_on: upload-ready-pipeline
+    steps:
+
+      - label: "RLHF · VeRL-Omni E2E Test"
+        source_file_dependencies:
+          - tests/e2e/offline_inference/rlhf_test/test_verl_omni_e2e.py
+          - tests/e2e/offline_inference/custom_pipeline/
+          - vllm_omni/diffusion/
+          - vllm_omni/engine/
+          - vllm_omni/entrypoints/
+          - vllm_omni/inputs/
+          - vllm_omni/outputs/
+        commands:
+          - |
+            timeout 20m bash -c "
+              pytest -s -v tests/e2e/offline_inference/rlhf_test/test_verl_omni_e2e.py
+            "
+        mirror_hardwares: l4_1
+
+      - label: "Omni · Qwen3-Omni Test"
+        source_file_dependencies:
+          - tests/e2e/online_serving/test_qwen3_omni.py
+          - vllm_omni/model_executor/models/qwen3_omni/
+          - vllm_omni/model_executor/stage_input_processors/qwen3_omni.py
+          - vllm_omni/deploy/qwen3_omni_moe.yaml
+        commands:
+          - |
+            timeout 20m bash -c "
+              pytest -s -v tests/e2e/online_serving/test_qwen3_omni.py -m 'core_model' --run-level 'core_model'
+            "
+        mirror_hardwares: h100_2
+
+      - label: "Omni · MiniCPM-o 4.5 Offline Test"
+        source_file_dependencies:
+          - tests/e2e/offline_inference/test_minicpmo_4_5.py
+          - vllm_omni/deploy/minicpmo_4_5.yaml
+        commands:
+          - |
+            timeout 20m bash -c "
+              pytest -s -v tests/e2e/offline_inference/test_minicpmo_4_5.py -m 'core_model' --run-level 'core_model'
+            "
+        agents:
+          queue: "mithril-h100-pool"
+        plugins:
+          - kubernetes:
+              podSpec:
+                containers:
+                  - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
+                    resources:
+                      limits:
+                        nvidia.com/gpu: 2
+                    volumeMounts:
+                      - name: devshm
+                        mountPath: /dev/shm
+                      - name: hf-cache
+                        mountPath: /root/.cache/huggingface
+                    env:
+                      - name: HF_HOME
+                        value: /root/.cache/huggingface
+                nodeSelector:
+                  node.kubernetes.io/instance-type: gpu-h100-sxm
+                volumes:
+                  - name: devshm
+                    emptyDir:
+                      medium: Memory
+                  - name: hf-cache
+                    hostPath:
+                      path: /mnt/hf-cache
+                      type: DirectoryOrCreate
+
+      - label: "Omni · MiniCPM-o 4.5 Online Test"
+        source_file_dependencies:
+          - tests/e2e/online_serving/test_minicpmo_4_5.py
+          - vllm_omni/deploy/minicpmo_4_5.yaml
+        commands:
+          - |
+            timeout 20m bash -c "
+              pytest -s -v tests/e2e/online_serving/test_minicpmo_4_5.py -m 'core_model' --run-level 'core_model'
+            "
+        agents:
+          queue: "mithril-h100-pool"
+        plugins:
+          - kubernetes:
+              podSpec:
+                containers:
+                  - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
+                    resources:
+                      limits:
+                        nvidia.com/gpu: 2
+                    volumeMounts:
+                      - name: devshm
+                        mountPath: /dev/shm
+                      - name: hf-cache
+                        mountPath: /root/.cache/huggingface
+                    env:
+                      - name: HF_HOME
+                        value: /root/.cache/huggingface
+                nodeSelector:
+                  node.kubernetes.io/instance-type: gpu-h100-sxm
+                volumes:
+                  - name: devshm
+                    emptyDir:
+                      medium: Memory
+                  - name: hf-cache
+                    hostPath:
+                      path: /mnt/hf-cache
+                      type: DirectoryOrCreate
+
+      - label: "TTS · VoxCPM2 Test"
+        source_file_dependencies:
+          - tests/e2e/online_serving/test_voxcpm2_tts.py
+          - vllm_omni/model_executor/models/voxcpm2/
+          - vllm_omni/deploy/voxcpm2.yaml
+        timeout_in_minutes: 30
+        commands:
+          - |
+            timeout 30m bash -c "
+              export VLLM_LOGGING_LEVEL=DEBUG
+              export VLLM_WORKER_MULTIPROC_METHOD=spawn
+              pytest -s -v tests/e2e/online_serving/test_voxcpm2_tts.py -m 'core_model' --run-level 'core_model'
+            "
+        mirror_hardwares: l4_1
+
+      - label: "TTS · Qwen3-TTS CustomVoice Test"
+        source_file_dependencies:
+          - tests/e2e/online_serving/test_qwen3_tts_customvoice.py
+          - vllm_omni/model_executor/models/qwen3_tts/
+          - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
+          - vllm_omni/deploy/qwen3_tts.yaml
+        commands:
+          - |
+            timeout 20m bash -c "
+              export VLLM_ALLOW_LONG_MAX_MODEL_LEN=1
+              pytest -s -v tests/e2e/online_serving/test_qwen3_tts_customvoice.py -m 'core_model' --run-level 'core_model'
+            "
+        mirror_hardwares: l4_1
+
+      - label: "TTS · Qwen3-TTS Base Test"
+        source_file_dependencies:
+          - tests/e2e/online_serving/test_qwen3_tts_base.py
+          - vllm_omni/model_executor/models/qwen3_tts/
+          - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
+          - vllm_omni/deploy/qwen3_tts.yaml
+        commands:
+          - |
+            timeout 20m bash -c "
+              export VLLM_WORKER_MULTIPROC_METHOD=spawn
+              export VLLM_ALLOW_LONG_MAX_MODEL_LEN=1
+              export VLLM_OMNI_USE_V2_RUNNER=1
+              pytest -s -v tests/e2e/online_serving/test_qwen3_tts_base.py -m 'core_model' --run-level 'core_model'
+            "
+        mirror_hardwares: l4_1
+
+      - label: "TTS · Higgs-Audio-V3 Test"
+        source_file_dependencies:
+          - tests/e2e/online_serving/test_higgs_audio_v3.py
+          - vllm_omni/model_executor/models/higgs_audio_v3/
+          - vllm_omni/model_executor/stage_input_processors/higgs_audio_v3.py
+          - vllm_omni/deploy/higgs_multimodal_qwen3.yaml
+        commands:
+          - |
+            timeout 20m bash -c "
+              export VLLM_WORKER_MULTIPROC_METHOD=spawn
+              export VLLM_USE_DEEP_GEMM=0
+              export VLLM_MOE_USE_DEEP_GEMM=0
+              pytest -s -v tests/e2e/online_serving/test_higgs_audio_v3.py -m 'core_model' --run-level 'core_model'
+            "
+        mirror_hardwares: l4_1
+
+      - label: "Diffusion · Qwen Image Edit Test"
+        source_file_dependencies:
+          - tests/e2e/online_serving/test_qwen_image_edit.py
+          - vllm_omni/diffusion/models/qwen_image/
+        commands:
+          - |
+            timeout 20m bash -c "
+              pytest -s -v tests/e2e/online_serving/test_qwen_image_edit.py -m 'core_model' --run-level 'core_model'
+            "
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · Bagel Test"
+        source_file_dependencies:
+          - tests/e2e/online_serving/test_bagel.py
+          - vllm_omni/model_executor/models/bagel/
+          - vllm_omni/diffusion/models/bagel/
+          - vllm_omni/model_executor/stage_input_processors/bagel.py
+          - vllm_omni/deploy/bagel.yaml
+        commands:
+          - |
+            timeout 40m bash -c "
+              export VLLM_IMAGE_FETCH_TIMEOUT=60
+              pytest -s -v tests/e2e/online_serving/test_bagel.py -m 'core_model' --run-level 'core_model'
+            "
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · Wan22 Test"
+        source_file_dependencies:
+          - tests/e2e/online_serving/test_wan22_t2v.py
+          - vllm_omni/diffusion/models/wan2_2/
+        commands:
+          - |
+            timeout 40m bash -c "
+              export VLLM_IMAGE_FETCH_TIMEOUT=60
+              pytest -s -v tests/e2e/online_serving/test_wan22_t2v.py -m 'core_model' --run-level 'core_model'
+            "
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · Cosmos3 Test"
+        source_file_dependencies:
+          - tests/e2e/online_serving/test_cosmos3.py
+          - vllm_omni/diffusion/models/cosmos3/
+        commands:
+          - |
+            timeout 40m bash -c "
+              export VLLM_IMAGE_FETCH_TIMEOUT=60
+              pytest -s -v tests/e2e/online_serving/test_cosmos3.py -m 'core_model' --run-level 'core_model'
+            "
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · Qwen Image Test"
+        source_file_dependencies:
+          - tests/e2e/online_serving/test_qwen_image.py
+          - vllm_omni/diffusion/models/qwen_image/
+        commands:
+          - |
+            timeout 40m bash -c "
+              export VLLM_IMAGE_FETCH_TIMEOUT=60
+              pytest -s -v tests/e2e/online_serving/test_qwen_image.py -m 'core_model' --run-level 'core_model'
+            "
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · Qwen Image Layered Test"
+        source_file_dependencies:
+          - tests/e2e/online_serving/test_qwen_image_layered.py
+          - vllm_omni/diffusion/models/qwen_image/
+        commands:
+          - |
+            timeout 40m bash -c "
+              export VLLM_IMAGE_FETCH_TIMEOUT=60
+              pytest -s -v tests/e2e/online_serving/test_qwen_image_layered.py -m 'core_model' --run-level 'core_model'
+            "
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · GGUF Plugin Test"
+        soft_fail: true
+        source_file_dependencies:
+          - tests/plugin/test_gguf_diffusion.py
+          - vllm_omni/diffusion/model_loader
+        commands:
+          - pip install "vllm-gguf-plugin>=0.0.3"
+          - timeout 30m pytest -sv tests/plugin/test_gguf_diffusion.py
+        mirror_hardwares: l4_1

.buildkite/cuda/test-weekly.yml

@@ -0,0 +1,136 @@
+env:
+  VLLM_WORKER_MULTIPROC_METHOD: spawn
+  VLLM_USE_FLASHINFER_MOE_FP16: 0
+  HF_HUB_DOWNLOAD_TIMEOUT: 300
+  HF_HUB_ETAG_TIMEOUT: 60
+
+steps:
+
+  - label: "Reliability Test - qwen3-omni"
+    timeout_in_minutes: 180
+    depends_on: upload-weekly-pipeline
+    if: build.env("WEEKLY") == "1" || build.pull_request.labels includes "weekly-test"
+    commands:
+      - pytest -s -v tests/dfx/reliability/test_reliability_qwen3_omni.py -m "slow"
+    mirror_hardwares: h100_2
+
+  - label: "Reliability Test - wan22"
+    timeout_in_minutes: 180
+    depends_on: upload-weekly-pipeline
+    if: build.env("WEEKLY") == "1" || build.pull_request.labels includes "weekly-test"
+    commands:
+      - pytest -s -v tests/dfx/reliability/test_reliability_wan22.py -m "slow"
+    mirror_hardwares: h100_2
+
+  - label: "Reliability Test - hunyuan-image"
+    timeout_in_minutes: 180
+    depends_on: upload-weekly-pipeline
+    if: build.env("WEEKLY") == "1" || build.pull_request.labels includes "weekly-test"
+    commands:
+      - pytest -s -v tests/dfx/reliability/test_reliability_hunyuan_image.py -m "slow"
+    mirror_hardwares: h100_4
+
+  - label: "Reliability Test - voxcpm2"
+    timeout_in_minutes: 180
+    depends_on: upload-weekly-pipeline
+    if: build.env("WEEKLY") == "1" || build.pull_request.labels includes "weekly-test"
+    commands:
+      - pytest -s -v tests/dfx/reliability/test_reliability_voxcpm2.py -m "slow"
+    mirror_hardwares: l4_1
+
+  - group: "Reliability Test - Invalid parameters Test"
+    depends_on: upload-weekly-pipeline
+    if: build.env("WEEKLY") == "1" || build.pull_request.labels includes "weekly-test"
+    steps:
+
+      - label: "Invalid parameters Test · H100"
+        timeout_in_minutes: 120
+        commands:
+          - pytest -s -v tests/dfx/reliability/invalid_param_test/ -m "slow and H100"
+        mirror_hardwares: h100_2
+
+      - label: "Invalid parameters Test · L4"
+        timeout_in_minutes: 120
+        commands:
+          - pytest -s -v tests/dfx/reliability/invalid_param_test/ -m "slow and L4"
+        mirror_hardwares: l4_1
+
+  - group: ":card_index_dividers: E2E Tests"
+    depends_on: upload-weekly-pipeline
+    if: build.env("WEEKLY") == "1" || build.pull_request.labels includes "weekly-test"
+    steps:
+
+      - label: "Omni · L4"
+        timeout_in_minutes: 90
+        commands:
+          - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
+          - pytest -s -v tests/e2e/*/test_qwen2_5_omni_expansion.py --run-level "core_model"
+          - >-
+            pytest -s -v tests/e2e/
+            --ignore=tests/e2e/online_serving/test_qwen2_5_omni_expansion.py
+            --ignore=tests/e2e/offline_inference/test_qwen2_5_omni_expansion.py
+            -m "slow and L4 and omni"
+            --run-level "full_model"
+        mirror_hardwares: l4_4
+
+      - label: "Omni · H100"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/ -m "slow and H100 and omni"
+            --run-level "full_model"
+        mirror_hardwares: h100_2
+
+      - label: "TTS · L4"
+        timeout_in_minutes: 90
+        commands:
+          - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
+          - export VLLM_USE_DEEP_GEMM="0"
+          - export VLLM_MOE_USE_DEEP_GEMM="0"
+          - |
+            failed=0
+            for f in $(pytest --collect-only -q tests/e2e/ -m "slow and L4 and tts" --run-level "full_model" | grep '\.py::' | sed -E 's|::.*||' | sort -u); do
+              pytest -s -v "$$f" -m "slow and L4 and tts" --run-level "full_model" || failed=1
+            done
+            exit $$failed
+        mirror_hardwares: l4_1
+
+      - label: "TTS · H100"
+        timeout_in_minutes: 90
+        commands:
+          - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
+          - export VLLM_USE_DEEP_GEMM="0"
+          - export VLLM_MOE_USE_DEEP_GEMM="0"
+          - >-
+            pytest -sv
+            tests/e2e/ -m "slow and H100 and tts"
+            --run-level "full_model"
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · L4"
+        timeout_in_minutes: 60
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/ -m "slow and diffusion and L4"
+            --run-level "full_model"
+        mirror_hardwares: l4_4
+
+      - label: "Diffusion · H100 · Single-GPU"
+        timeout_in_minutes: 120
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/ -m "slow and diffusion and H100 and not distributed_cuda"
+            --run-level "full_model"
+        mirror_hardwares: h100_1
+
+      - label: "Diffusion · H100 · Multi-GPU"
+        timeout_in_minutes: 90
+        commands:
+          - >-
+            pytest -sv
+            tests/e2e/ -m "slow and diffusion and H100 and distributed_cuda"
+            --run-level "full_model"
+        mirror_hardwares: h100_2

.buildkite/intel/pipeline-intel.yml

@@ -1,16 +1,16 @@
-# .buildkite/pipeline-intel.yaml
+# .buildkite/intel/pipeline-intel.yml
 steps:
   - group: "INTEL Tests"
     steps:
       - label: "Intel: XPU Test"
-        command: "bash .buildkite/scripts/hardware_ci/run-xpu-test.sh"
+        command: "bash .buildkite/intel/scripts/run-xpu-test.sh"
         agents:
           queue: "intel-gpu-omni"
         env:
           DOCKER_BUILDKIT: "1"
           # Buildkite will automatically replace this with the actual commit hash
           VLLM_IMAGE_TAG: "${BUILDKITE_COMMIT}"
-          VLLM_VERSION: "v0.19.0"
+          VLLM_VERSION: "v0.25.0"
         priority: 100
         timeout_in_minutes: 60
-        soft_fail: true
+        soft_fail: false

.buildkite/intel/scripts/bootstrap-intel-omni.sh

@@ -0,0 +1,16 @@
+#!/bin/bash
+# vllm-omni Intel bootstrap
+# Uses static pipeline-intel.yml for Intel XPU tests
+
+set -euo pipefail
+
+source .buildkite/common/scripts/resolve_skip_ci.sh
+
+upload_pipeline() {
+    echo "--- 🛠 Preparing Intel pipeline"
+    buildkite-agent pipeline upload .buildkite/intel/pipeline-intel.yml
+}
+
+gate_bootstrap_ci intel l2
+
+upload_pipeline

.buildkite/intel/scripts/run-xpu-test.sh

@@ -13,12 +13,33 @@ container_name="xpu_${BUILDKITE_COMMIT}_$(
 )"
 
 cd "${omni_source_dir}"
+
+# The XPU base image is ~37GB; the default gzip layer exporter is single-threaded
+# and takes ~16min to compress it. zstd is multi-threaded (uses all cores) at a
+# similar ratio; set EXPORT_COMPRESSION=uncompressed to skip compression entirely
+# for local-only images. This requires the containerd image store (buildx docker
+# driver), which is the default here.
+EXPORT_COMPRESSION="${EXPORT_COMPRESSION:-zstd}"
+if [ "${EXPORT_COMPRESSION}" = "uncompressed" ]; then
+    export_args=(--output "type=image,name={{IMAGE}},compression=uncompressed")
+else
+    export_args=(--output "type=image,name={{IMAGE}},compression=${EXPORT_COMPRESSION},compression-level=3,force-compression=true")
+fi
+
+docker_build() {
+    # $1 = image name; remaining args passed through to docker build.
+    local image="$1"
+    shift
+    local out=("${export_args[@]/'{{IMAGE}}'/${image}}")
+    docker build "${out[@]}" "$@" -f docker/Dockerfile.xpu .
+}
+
 if [ -z "$(docker images -q "${base_image_name}")" ]; then
-    docker build --target vllm-base -t "${base_image_name}" --build-arg "VLLM_VERSION=${VLLM_VERSION}" -f docker/Dockerfile.xpu .
+    docker_build "${base_image_name}" --target vllm-base --build-arg "VLLM_VERSION=${VLLM_VERSION}"
 fi
 
 # Try building the docker image
-docker build --build-arg "VLLM_BASE=${base_image_name}" --build-arg "VLLM_VERSION=${VLLM_VERSION}" -t "${image_name}" -f docker/Dockerfile.xpu .
+docker_build "${image_name}" --build-arg "VLLM_BASE=${base_image_name}" --build-arg "VLLM_VERSION=${VLLM_VERSION}"
 
 # Setup cleanup
 remove_docker_container() {
@@ -52,5 +73,7 @@ time timeout -k 30 30m docker run \
     pip install tblib==3.1.0
     cd /workspace/vllm-omni
     pytest -v -s -m "core_model and xpu and B60"
+    pytest -v -s tests/diffusion/quantization/test_mxfp8_config.py
     pytest -v -s -m "advanced_model and xpu and B60"
+    pytest -v -s -m "omni and xpu and B60"
 '

.buildkite/nightly-release-pipeline.yaml

@@ -1,20 +0,0 @@
-steps:
-  - label: "Build and upload wheel"
-    key: "build-wheel"
-    agents:
-      queue: cpu_queue_release
-    commands:
-      - "curl -LsSf https://astral.sh/uv/install.sh | sh"
-      - 'export PATH="$HOME/.local/bin:$PATH"'
-      - "uv venv --python=3.12 && source .venv/bin/activate"
-      - "uv pip install --upgrade build"
-      - "python3 -m build"
-      - "bash .buildkite/scripts/upload-nightly-wheels.sh"
-
-  - label: "Generate and upload wheel indices"
-    depends_on: "build-wheel"
-    allow_dependency_failure: true
-    agents:
-      queue: small_cpu_queue_release
-    commands:
-      - "bash .buildkite/scripts/generate-and-upload-nightly-index.sh"

.buildkite/npu/bootstrap-upload-steps.yml

@@ -0,0 +1,181 @@
+# Uploaded by upload_pipeline.py --upload (not the Buildkite hook).
+# ``if`` is injected by step ``key`` from skip-ci at upload time.
+steps:
+  - label: ":buildkit: Build and Push NPU Test Image - A2"
+    key: image-build-a2
+    agents:
+      queue: "ascend-a2b3"
+      resource_class: "npu-1"
+    plugins:
+      - kubernetes:
+          metadata:
+            annotations:
+              vault.hashicorp.com/agent-init-first: "true"
+              vault.hashicorp.com/agent-inject: "true"
+              vault.hashicorp.com/agent-inject-perms-ca.pem: "0400"
+              vault.hashicorp.com/agent-inject-perms-cert.pem: "0400"
+              vault.hashicorp.com/agent-inject-perms-config.json: "0400"
+              vault.hashicorp.com/agent-inject-perms-key.pem: "0400"
+              vault.hashicorp.com/agent-inject-secret-ca.pem: internal/data/ascend/buildkitd
+              vault.hashicorp.com/agent-inject-secret-cert.pem: internal/data/ascend/buildkitd
+              vault.hashicorp.com/agent-inject-secret-config.json: internal/data/ascend/buildkitd
+              vault.hashicorp.com/agent-inject-secret-key.pem: internal/data/ascend/buildkitd
+              vault.hashicorp.com/agent-inject-template-ca.pem: "{{- with secret \"internal/data/ascend/buildkitd\" -}}\n{{ .Data.data.RootCA }}\n{{- end }}"
+              vault.hashicorp.com/agent-inject-template-cert.pem: "{{- with secret \"internal/data/ascend/buildkitd\" -}}\n{{ .Data.data.ClientCaCert }}\n{{- end }}"
+              vault.hashicorp.com/agent-inject-template-config.json: "{{- with secret \"internal/data/ascend/buildkitd\" -}}\n{{ .Data.data.dockerConfig }}\n{{- end }}"
+              vault.hashicorp.com/agent-inject-template-key.pem: "{{- with secret \"internal/data/ascend/buildkitd\" -}}\n{{ .Data.data.ClientCaKey }}\n{{- end }}"
+              vault.hashicorp.com/agent-pre-populate-only: "true"
+              vault.hashicorp.com/agent-run-as-group: "1000"
+              vault.hashicorp.com/agent-run-as-user: "1000"
+              vault.hashicorp.com/agent-service-account-token-volume-name: token-vol
+              vault.hashicorp.com/role: ascend-gha-runners
+              vault.hashicorp.com/secret-volume-path: /home/user/.docker/
+              vault.hashicorp.com/tls-skip-verify: "true"
+          podSpecPatch:
+            volumes:
+            - name: token-vol
+              projected:
+                defaultMode: 420
+                sources:
+                - serviceAccountToken:
+                    audience: api
+                    expirationSeconds: 600
+                    path: token
+    env:
+      VLLM_IMAGE_TAG: "${BUILDKITE_COMMIT}"
+      IMAGE_NAME: "vllm-omni-ci-npu"
+      IMAGE_REGISTRY: "swr.cn-southwest-2.myhuaweicloud.com/modelfoundry"
+      BUILDKITD_ADDR: "tcp://buildkitd-service.buildkitd:1234"
+    command: |
+      set -ex
+
+      echo "--- Building and pushing NPU Test Image"
+      echo "Image: $${IMAGE_REGISTRY}/$${IMAGE_NAME}:$${VLLM_IMAGE_TAG}"
+      echo "buildkitd address: $${BUILDKITD_ADDR}"
+
+      if ! command -v buildctl &> /dev/null; then
+        echo "Installing buildctl..."
+        mkdir -p /tmp/buildkit
+        BUILDKIT_VERSION="v0.29.0"
+        wget -q "https://github.com/moby/buildkit/releases/download/$${BUILDKIT_VERSION}/buildkit-$${BUILDKIT_VERSION}.linux-arm64.tar.gz" -O /tmp/buildkit.tar.gz
+        tar -xzf /tmp/buildkit.tar.gz -C /tmp/buildkit
+        cp /tmp/buildkit/bin/buildctl /usr/local/bin/
+      fi
+
+      export DOCKER_CONFIG=/home/user/.docker
+      buildctl \
+        --addr="$${BUILDKITD_ADDR}" \
+        --tlscacert=/home/user/.docker/ca.pem \
+        --tlscert=/home/user/.docker/cert.pem \
+        --tlskey=/home/user/.docker/key.pem \
+        build \
+        --frontend dockerfile.v0 \
+        --local context=. \
+        --local dockerfile=./docker \
+        --opt filename=Dockerfile.npu.ci \
+        --secret id=dockerconfig,src=/home/user/.docker/config.json \
+        --output type=image,name=$${IMAGE_REGISTRY}/$${IMAGE_NAME}:$${VLLM_IMAGE_TAG},push=true \
+        --progress=plain
+
+      echo "--- Image pushed successfully"
+      echo "$${IMAGE_REGISTRY}/$${IMAGE_NAME}:$${VLLM_IMAGE_TAG}"
+
+  - label: ":buildkit: Build and Push NPU Test Image - A3"
+    key: image-build-a3
+    agents:
+      queue: "ascend-a2b3"
+      resource_class: "npu-1"
+    plugins:
+      - kubernetes:
+          metadata:
+            annotations:
+              vault.hashicorp.com/agent-init-first: "true"
+              vault.hashicorp.com/agent-inject: "true"
+              vault.hashicorp.com/agent-inject-perms-ca.pem: "0400"
+              vault.hashicorp.com/agent-inject-perms-cert.pem: "0400"
+              vault.hashicorp.com/agent-inject-perms-config.json: "0400"
+              vault.hashicorp.com/agent-inject-perms-key.pem: "0400"
+              vault.hashicorp.com/agent-inject-secret-ca.pem: internal/data/ascend/buildkitd
+              vault.hashicorp.com/agent-inject-secret-cert.pem: internal/data/ascend/buildkitd
+              vault.hashicorp.com/agent-inject-secret-config.json: internal/data/ascend/buildkitd
+              vault.hashicorp.com/agent-inject-secret-key.pem: internal/data/ascend/buildkitd
+              vault.hashicorp.com/agent-inject-template-ca.pem: "{{- with secret \"internal/data/ascend/buildkitd\" -}}\n{{ .Data.data.RootCA }}\n{{- end }}"
+              vault.hashicorp.com/agent-inject-template-cert.pem: "{{- with secret \"internal/data/ascend/buildkitd\" -}}\n{{ .Data.data.ClientCaCert }}\n{{- end }}"
+              vault.hashicorp.com/agent-inject-template-config.json: "{{- with secret \"internal/data/ascend/buildkitd\" -}}\n{{ .Data.data.dockerConfig }}\n{{- end }}"
+              vault.hashicorp.com/agent-inject-template-key.pem: "{{- with secret \"internal/data/ascend/buildkitd\" -}}\n{{ .Data.data.ClientCaKey }}\n{{- end }}"
+              vault.hashicorp.com/agent-pre-populate-only: "true"
+              vault.hashicorp.com/agent-run-as-group: "1000"
+              vault.hashicorp.com/agent-run-as-user: "1000"
+              vault.hashicorp.com/agent-service-account-token-volume-name: token-vol
+              vault.hashicorp.com/role: ascend-gha-runners
+              vault.hashicorp.com/secret-volume-path: /home/user/.docker/
+              vault.hashicorp.com/tls-skip-verify: "true"
+          podSpecPatch:
+            volumes:
+            - name: token-vol
+              projected:
+                defaultMode: 420
+                sources:
+                - serviceAccountToken:
+                    audience: api
+                    expirationSeconds: 600
+                    path: token
+    env:
+      VLLM_IMAGE_TAG: "${BUILDKITE_COMMIT}-a3"
+      IMAGE_NAME: "vllm-omni-ci-npu"
+      IMAGE_REGISTRY: "swr.cn-southwest-2.myhuaweicloud.com/modelfoundry"
+      BUILDKITD_ADDR: "tcp://buildkitd-service.buildkitd:1234"
+    command: |
+      set -ex
+
+      echo "--- Building and pushing NPU Test Image"
+      echo "Image: $${IMAGE_REGISTRY}/$${IMAGE_NAME}:$${VLLM_IMAGE_TAG}"
+      echo "buildkitd address: $${BUILDKITD_ADDR}"
+
+      if ! command -v buildctl &> /dev/null; then
+        echo "Installing buildctl..."
+        mkdir -p /tmp/buildkit
+        BUILDKIT_VERSION="v0.29.0"
+        wget -q "https://github.com/moby/buildkit/releases/download/$${BUILDKIT_VERSION}/buildkit-$${BUILDKIT_VERSION}.linux-arm64.tar.gz" -O /tmp/buildkit.tar.gz
+        tar -xzf /tmp/buildkit.tar.gz -C /tmp/buildkit
+        cp /tmp/buildkit/bin/buildctl /usr/local/bin/
+      fi
+
+      export DOCKER_CONFIG=/home/user/.docker
+      buildctl \
+        --addr="$${BUILDKITD_ADDR}" \
+        --tlscacert=/home/user/.docker/ca.pem \
+        --tlscert=/home/user/.docker/cert.pem \
+        --tlskey=/home/user/.docker/key.pem \
+        build \
+        --frontend dockerfile.v0 \
+        --local context=. \
+        --local dockerfile=./docker \
+        --opt filename=Dockerfile.npu.ci.a3 \
+        --secret id=dockerconfig,src=/home/user/.docker/config.json \
+        --output type=image,name=$${IMAGE_REGISTRY}/$${IMAGE_NAME}:$${VLLM_IMAGE_TAG},push=true \
+        --progress=plain
+
+      echo "--- Image pushed successfully"
+      echo "$${IMAGE_REGISTRY}/$${IMAGE_NAME}:$${VLLM_IMAGE_TAG}"
+
+  - label: "Upload Ready Pipeline"
+    depends_on:
+      - image-build-a2
+      - image-build-a3
+    key: upload-ready-pipeline
+    agents:
+      queue: "cpu_queue_premerge"
+    command: |
+      python3 .buildkite/common/scripts/upload_pipeline.py --upload .buildkite/npu/test-npu-ready.yml
+
+  # L4 Test — main+NIGHTLY=1 (scheduled), or PR with specific label (e.g. add label then Rebuild)
+  - label: "Upload Nightly Pipeline"
+    depends_on:
+      - image-build-a2
+      - image-build-a3
+    key: upload-nightly-pipeline
+    commands:
+      - python3 .buildkite/common/scripts/upload_pipeline.py --upload .buildkite/npu/test-npu-nightly.yml
+    agents:
+      queue: "cpu_queue_premerge"

.buildkite/npu/pipeline-npu-a3.yml

@@ -0,0 +1,110 @@
+# steps:
+#   - label: ":buildkit: Build and Push NPU Test Image - A3"
+#     key: image-build-a3
+#     agents:
+#       queue: "ascend-a3"
+#       resource_class: "npu-2"
+#     plugins:
+#       - kubernetes:
+#           metadata:
+#             annotations:
+#               vault.hashicorp.com/agent-init-first: "true"
+#               vault.hashicorp.com/agent-inject: "true"
+#               vault.hashicorp.com/agent-inject-perms-ca.pem: "0400"
+#               vault.hashicorp.com/agent-inject-perms-cert.pem: "0400"
+#               vault.hashicorp.com/agent-inject-perms-config.json: "0400"
+#               vault.hashicorp.com/agent-inject-perms-key.pem: "0400"
+#               vault.hashicorp.com/agent-inject-secret-ca.pem: internal/data/ascend/buildkitd
+#               vault.hashicorp.com/agent-inject-secret-cert.pem: internal/data/ascend/buildkitd
+#               vault.hashicorp.com/agent-inject-secret-config.json: internal/data/ascend/buildkitd
+#               vault.hashicorp.com/agent-inject-secret-key.pem: internal/data/ascend/buildkitd
+#               vault.hashicorp.com/agent-inject-template-ca.pem: "{{- with secret \"internal/data/ascend/buildkitd\" -}}\n{{ .Data.data.RootCA }}\n{{- end }}"
+#               vault.hashicorp.com/agent-inject-template-cert.pem: "{{- with secret \"internal/data/ascend/buildkitd\" -}}\n{{ .Data.data.ClientCaCert }}\n{{- end }}"
+#               vault.hashicorp.com/agent-inject-template-config.json: "{{- with secret \"internal/data/ascend/buildkitd\" -}}\n{{ .Data.data.dockerConfig }}\n{{- end }}"
+#               vault.hashicorp.com/agent-inject-template-key.pem: "{{- with secret \"internal/data/ascend/buildkitd\" -}}\n{{ .Data.data.ClientCaKey }}\n{{- end }}"
+#               vault.hashicorp.com/agent-pre-populate-only: "true"
+#               vault.hashicorp.com/agent-run-as-group: "1000"
+#               vault.hashicorp.com/agent-run-as-user: "1000"
+#               vault.hashicorp.com/agent-service-account-token-volume-name: token-vol
+#               vault.hashicorp.com/role: ascend-gha-runners
+#               vault.hashicorp.com/secret-volume-path: /home/user/.docker/
+#               vault.hashicorp.com/tls-skip-verify: "true"
+#           podSpecPatch:
+#             volumes:
+#             - name: token-vol
+#               projected:
+#                 defaultMode: 420
+#                 sources:
+#                 - serviceAccountToken:
+#                     audience: api
+#                     expirationSeconds: 600
+#                     path: token
+#     env:
+#       VLLM_IMAGE_TAG: "${BUILDKITE_COMMIT}"
+#       IMAGE_NAME: "vllm-omni-ci-npu"
+#       IMAGE_REGISTRY: "swr.cn-southwest-2.myhuaweicloud.com/modelfoundry"
+#       BUILDKITD_ADDR: "tcp://buildkitd-service.buildkitd:1234"
+#     command: |
+#       set -ex
+
+#       echo "--- Building and pushing NPU Test Image"
+#       echo "Image: $${IMAGE_REGISTRY}/$${IMAGE_NAME}:$${VLLM_IMAGE_TAG}"
+#       echo "buildkitd address: $${BUILDKITD_ADDR}"
+
+#       git config --global url."https://gh-proxy.test.osinfra.cn/https://github.com/".insteadOf "https://github.com/"
+
+#       if ! command -v buildctl &> /dev/null; then
+#         echo "Installing buildctl..."
+#         mkdir -p /tmp/buildkit
+#         BUILDKIT_VERSION="v0.29.0"
+#         wget -q "https://gh-proxy.test.osinfra.cn/https://github.com/moby/buildkit/releases/download/$${BUILDKIT_VERSION}/buildkit-$${BUILDKIT_VERSION}.linux-arm64.tar.gz" -O /tmp/buildkit.tar.gz
+#         tar -xzf /tmp/buildkit.tar.gz -C /tmp/buildkit
+#         cp /tmp/buildkit/bin/buildctl /usr/local/bin/
+#       fi
+
+#       export DOCKER_CONFIG=/home/user/.docker
+#       buildctl \
+#         --addr="$${BUILDKITD_ADDR}" \
+#         --tlscacert=/home/user/.docker/ca.pem \
+#         --tlscert=/home/user/.docker/cert.pem \
+#         --tlskey=/home/user/.docker/key.pem \
+#         build \
+#         --frontend dockerfile.v0 \
+#         --local context=. \
+#         --local dockerfile=./docker \
+#         --opt filename=Dockerfile.npu.ci.a3 \
+#         --secret id=dockerconfig,src=/home/user/.docker/config.json \
+#         --output type=image,name=$${IMAGE_REGISTRY}/$${IMAGE_NAME}:$${VLLM_IMAGE_TAG},push=true \
+#         --progress=plain
+
+#       echo "--- Image pushed successfully"
+#       echo "$${IMAGE_REGISTRY}/$${IMAGE_NAME}:$${VLLM_IMAGE_TAG}"
+
+#   # - label: "Upload Ready Pipeline - A3"
+#   #   depends_on: image-build-a3
+#   #   key: upload-ready-pipeline-a3
+#   #   if: build.branch != "main" && build.pull_request.labels includes "npu-test"
+#   #   agents:
+#   #     queue: "ascend-a3"
+#   #     resource_class: "npu-2"
+#   #   command: |
+#   #     buildkite-agent pipeline upload .buildkite/npu/test-npu-ready.yml
+
+  # # L4 Test — main+NIGHTLY=1 (scheduled), or PR with specific label (e.g. add label then Rebuild)
+  # - label: "Upload Nightly Pipeline"
+  #   depends_on: image-build
+  #   key: upload-nightly-pipeline
+  #   if: >-
+  #     (build.branch == "main" && build.env("NIGHTLY") == "1") ||
+  #     (build.branch != "main" && (
+  #       build.pull_request.labels includes "nightly-test" ||
+  #       build.pull_request.labels includes "omni-test" ||
+  #       build.pull_request.labels includes "tts-test" ||
+  #       build.pull_request.labels includes "diffusion-x2iat-test" ||
+  #       build.pull_request.labels includes "diffusion-x2v-test"
+  #     ))
+  #   commands:
+  #     - buildkite-agent pipeline upload .buildkite/npu/test-npu-nightly.yml
+  #   agents:
+  #     queue: "ascend-a3"
+  #     resource_class: "npu-2"

.buildkite/npu/pipeline-npu.yml

@@ -0,0 +1,9 @@
+# Hook uploads this entry YAML; upload_pipeline.py --upload reads bootstrap-upload-steps.yml
+# and injects ``if`` by step ``key`` from skip-ci.
+steps:
+  - label: ":github: Resolve skip-ci & upload pipeline"
+    key: upload-ci-pipeline
+    commands:
+      - "python3 .buildkite/common/scripts/upload_pipeline.py --upload .buildkite/npu/bootstrap-upload-steps.yml"
+    agents:
+      queue: "cpu_queue_premerge"

.buildkite/npu/test-npu-nightly.yml

@@ -0,0 +1,58 @@
+steps:
+  - group: ":card_index_dividers: Diffusion X2V Model Test"
+    key: nightly-diffusion-x2v-group
+    depends_on: upload-nightly-pipeline
+    if: >-
+      build.env("NIGHTLY") == "1" ||
+      build.pull_request.labels includes "nightly-test" ||
+      build.pull_request.labels includes "diffusion-x2v-test"
+    steps:
+      - label: ":full_moon: Diffusion X2V · Function Test"
+        mirror_hardwares: a2b3_npu_4
+        env:
+          VLLM_WORKER_MULTIPROC_METHOD: spawn
+          HF_TOKEN: "${HF_TOKEN}"
+        commands:
+          - pytest -s -v tests/e2e/online_serving/test_wan22_expansion.py -k npu_i2v --run-level "full_model"
+  - group: ":card_index_dividers: TTS Model Test"
+    key: nightly-tts-test-group
+    depends_on: upload-nightly-pipeline
+    if: build.env("NIGHTLY") == "1" || build.pull_request.labels includes "nightly-test" || build.pull_request.labels includes "tts-test"
+    steps:
+      - label: ":full_moon: TTS · Perf Test"
+        mirror_hardwares: a3_npu_2
+        env:
+          VLLM_WORKER_MULTIPROC_METHOD: spawn
+          VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
+          HF_TOKEN: "${HF_TOKEN}"
+        commands:
+          - export BENCHMARK_DIR=tests/dfx/perf/results
+          - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
+          - |
+            set +e
+            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py -m npu --test-config-file tests/dfx/perf/tests/test_tts.json
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
+            exit $$EXIT
+
+  - group: ":card_index_dividers: Omni Model Test"
+    key: nightly-omni-group
+    depends_on: upload-nightly-pipeline
+    if: build.env("NIGHTLY") == "1" || build.pull_request.labels includes "nightly-test" || build.pull_request.labels includes "omni-test"
+    steps:
+      - label: ":robot: Omni · MiniCPM-o 4.5 Test"
+        agents:
+          queue: "ascend-a3"
+          resource_class: "npu-2"
+        image: "swr.cn-southwest-2.myhuaweicloud.com/modelfoundry/vllm-omni-ci-npu:${BUILDKITE_COMMIT}-a3"
+        plugins:
+          - kubernetes:
+              podSpecPatch:
+                imagePullSecrets:
+                  - name: swr-secret
+        env:
+          VLLM_WORKER_MULTIPROC_METHOD: spawn
+          VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
+          HF_TOKEN: "${HF_TOKEN}"
+        commands:
+          - pytest -s -v tests/e2e/offline_inference/test_minicpmo_4_5.py tests/e2e/online_serving/test_minicpmo_4_5.py tests/e2e/online_serving/test_minicpmo_4_5_expansion.py -m 'full_model' --run-level 'full_model'

.buildkite/npu/test-npu-ready.yml

@@ -0,0 +1,34 @@
+steps:
+  - group: ":card_index_dividers: Omni Model Test"
+    key: ready-omni-group
+    depends_on: upload-ready-pipeline
+    steps:
+      - label: ":robot: Omni · MiniCPM-o 4.5 Test"
+        agents:
+          queue: "ascend-a3"
+          resource_class: "npu-2"
+        image: "swr.cn-southwest-2.myhuaweicloud.com/modelfoundry/vllm-omni-ci-npu:${BUILDKITE_COMMIT}-a3"
+        plugins:
+          - kubernetes:
+              podSpecPatch:
+                imagePullSecrets:
+                  - name: swr-secret
+        env:
+          VLLM_WORKER_MULTIPROC_METHOD: spawn
+          VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
+          HF_TOKEN: "${HF_TOKEN}"
+        commands:
+          - pytest -s -v tests/e2e/offline_inference/test_minicpmo_4_5.py tests/e2e/online_serving/test_minicpmo_4_5.py -m 'core_model' --run-level 'core_model'
+
+  - group: ":card_index_dividers: TTS Model Test"
+    key: ready-tts-group
+    depends_on: upload-ready-pipeline
+    steps:
+      - label: ":sound: TTS · Qwen3-TTS CustomVoice Test"
+        mirror_hardwares: a2b3_npu_1
+        env:
+          VLLM_WORKER_MULTIPROC_METHOD: spawn
+          VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
+          HF_TOKEN: "${HF_TOKEN}"
+        commands:
+          - pytest -s -v tests/e2e/online_serving/test_qwen3_tts_customvoice.py -m 'core_model' --run-level 'core_model'

.buildkite/pipeline.yml

@@ -1,63 +0,0 @@
-# Document 1: Buildkite loads only this block on first parse. The next step resolves docs-only skip-ci
-# from git diff, then uploads document 2. When docs-only skip applies, image-build still runs if nightly-test
-# / main NIGHTLY so upload-nightly is not skipped together with test-ready/test-merge.
-#
-# Document 2: appended after `---`; same file, read by upload_pipeline_with_skip_ci.sh (not evaluated as a second pipeline by Buildkite).
-steps:
-  - label: ":github: Resolve skip-ci & upload pipeline"
-    key: upload-ci-pipeline
-    commands:
-      - "bash .buildkite/scripts/upload_pipeline_with_skip_ci.sh"
-    agents:
-      queue: "cpu_queue_premerge"
-
----
-steps:
-  - label: ":docker: Build image"
-    key: image-build
-    if: __IMAGE_BUILD_IF__
-    commands:
-      - "aws ecr-public get-login-password --region us-east-1 | docker login --username AWS --password-stdin public.ecr.aws/q9t5s3a7"
-      - "docker build --progress=plain --file docker/Dockerfile.ci -t vllm-omni-ci ."
-      - "docker tag vllm-omni-ci public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT"
-      - "docker push public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT"
-    agents:
-      queue: "cpu_queue_premerge"
-
-  # L2 Test
-  - label: "Upload Ready Pipeline"
-    depends_on: image-build
-    key: upload-ready-pipeline
-    if: __UPLOAD_READY_IF__
-    commands:
-      - buildkite-agent pipeline upload .buildkite/test-ready.yml
-    agents:
-      queue: "cpu_queue_premerge"
-
-  # L3 Test
-  - label: "Upload Merge Pipeline"
-    depends_on: image-build
-    key: upload-merge-pipeline
-    if: __UPLOAD_MERGE_IF__
-    commands:
-      - buildkite-agent pipeline upload .buildkite/test-merge.yml
-    agents:
-      queue: "cpu_queue_premerge"
-
-  # L4 Test — main+NIGHTLY=1 (scheduled), or PR with specific label (e.g. add label then Rebuild)
-  - label: "Upload Nightly Pipeline"
-    depends_on: image-build
-    key: upload-nightly-pipeline
-    if: >-
-      (build.branch == "main" && build.env("NIGHTLY") == "1") ||
-      (build.branch != "main" && (
-        build.pull_request.labels includes "nightly-test" ||
-        build.pull_request.labels includes "omni-test" ||
-        build.pull_request.labels includes "tts-test" ||
-        build.pull_request.labels includes "diffusion-x2iat-test" ||
-        build.pull_request.labels includes "diffusion-x2v-test"
-      ))
-    commands:
-      - buildkite-agent pipeline upload .buildkite/test-nightly.yml
-    agents:
-      queue: "cpu_queue_premerge"

.buildkite/release/release-pipeline.yml

@@ -0,0 +1,110 @@
+steps:
+  - input: "Provide Release version here"
+    id: input-release-version
+    fields:
+      - text: "What is the release version?"
+        key: release-version
+
+  - label: "Build and upload wheel"
+    depends_on: ~
+    key: "build-wheel"
+    agents:
+      queue: cpu_queue_release
+    commands:
+      - "curl -LsSf https://astral.sh/uv/install.sh | sh"
+      - 'export PATH="$HOME/.local/bin:$PATH"'
+      - "uv venv --python=3.12 && source .venv/bin/activate"
+      - "uv pip install --upgrade build"
+      - "python3 -m build"
+      - "bash .buildkite/release/scripts/upload-nightly-wheels.sh"
+
+  - label: "Generate and upload wheel indices"
+    depends_on: "build-wheel"
+    allow_dependency_failure: true
+    agents:
+      queue: small_cpu_queue_release
+    commands:
+      - "bash .buildkite/release/scripts/generate-and-upload-nightly-index.sh"
+
+  - block: "Upload release wheels to PyPI"
+    depends_on: "build-wheel"
+    key: block-upload-release-wheels
+
+  - label: "Upload release wheels to PyPI"
+    depends_on:
+      - block-upload-release-wheels
+      - input-release-version
+    key: upload-release-wheels
+    agents:
+      queue: small_cpu_queue_release
+    commands:
+      - "bash .buildkite/release/scripts/publish-release-wheels-pypi.sh"
+
+  - block: "Unblock to build release Docker images"
+    depends_on: ~
+    key: block-build-release-images
+
+  - group: "Build release Docker images"
+    key: "build-release-images"
+    depends_on: block-build-release-images
+    allow_dependency_failure: true
+    steps:
+      - label: "Build release image - x86_64"
+        key: build-release-image-x86
+        depends_on: ~
+        agents:
+          queue: cpu_queue_release
+        commands:
+          - "aws ecr-public get-login-password --region us-east-1 | docker login --username AWS --password-stdin public.ecr.aws/q9t5s3a7"
+          - |
+            DOCKER_BUILDKIT=1 docker build \
+              --build-arg max_jobs=16 \
+              --tag public.ecr.aws/q9t5s3a7/vllm-omni-release-repo:$BUILDKITE_COMMIT-$(uname -m) \
+              --progress plain \
+              -f docker/Dockerfile.cuda .
+          - "docker push public.ecr.aws/q9t5s3a7/vllm-omni-release-repo:$BUILDKITE_COMMIT-$(uname -m)"
+          # re-tag to default image tag and push, just in case arm64 build fails
+          - "docker tag public.ecr.aws/q9t5s3a7/vllm-omni-release-repo:$BUILDKITE_COMMIT-$(uname -m) public.ecr.aws/q9t5s3a7/vllm-omni-release-repo:$BUILDKITE_COMMIT"
+          - "docker push public.ecr.aws/q9t5s3a7/vllm-omni-release-repo:$BUILDKITE_COMMIT"
+        env:
+          DOCKER_BUILDKIT: "1"
+
+      - label: "Build release image - aarch64"
+        key: build-release-image-arm64
+        depends_on: ~
+        agents:
+          queue: arm64_cpu_queue_release
+        commands:
+          - "aws ecr-public get-login-password --region us-east-1 | docker login --username AWS --password-stdin public.ecr.aws/q9t5s3a7"
+          - |
+            DOCKER_BUILDKIT=1 docker build \
+              --build-arg max_jobs=16 \
+              --tag public.ecr.aws/q9t5s3a7/vllm-omni-release-repo:$BUILDKITE_COMMIT-$(uname -m) \
+              --progress plain \
+              -f docker/Dockerfile.cuda .
+          - "docker push public.ecr.aws/q9t5s3a7/vllm-omni-release-repo:$BUILDKITE_COMMIT-$(uname -m)"
+        env:
+          DOCKER_BUILDKIT: "1"
+
+  - block: "Publish release images to DockerHub"
+    key: block-publish-release-images
+    depends_on:
+      - build-release-image-x86
+      - build-release-image-arm64
+      - input-release-version
+
+  - label: "Publish release images to DockerHub"
+    depends_on:
+      - block-publish-release-images
+    key: publish-release-images-dockerhub
+    agents:
+      queue: small_cpu_queue_release
+    commands:
+      - "bash .buildkite/release/scripts/publish-release-images.sh"
+    plugins:
+      - docker-login#v3.0.0:
+          username: vllmbot
+          password-env: DOCKERHUB_TOKEN
+    env:
+      DOCKER_BUILDKIT: "1"
+      DOCKERHUB_USERNAME: "vllmbot"

.buildkite/release/scripts/generate-and-upload-nightly-index.sh

@@ -36,12 +36,12 @@ mkdir -p "$INDICES_OUTPUT_DIR"
 
 # HACK: we do not need regex module here, but it is required by pre-commit hook
 # To avoid any external dependency, we simply replace it back to the stdlib re module
-sed -i.bak 's/import regex as re/import re/g' .buildkite/scripts/generate-nightly-index.py && rm -f .buildkite/scripts/generate-nightly-index.py.bak
+sed -i.bak 's/import regex as re/import re/g' .buildkite/release/scripts/generate-nightly-index.py && rm -f .buildkite/release/scripts/generate-nightly-index.py.bak
 
 # Generate indices -- the version is just the commit hash (not omni/{commit})
 # because relative paths are computed between the index and wheel directories,
 # both of which live under the omni/ prefix in S3.
-$PYTHON .buildkite/scripts/generate-nightly-index.py \
+$PYTHON .buildkite/release/scripts/generate-nightly-index.py \
     --version "$BUILDKITE_COMMIT" \
     --current-objects "$obj_json" \
     --output-dir "$INDICES_OUTPUT_DIR" \
@@ -78,7 +78,7 @@ if [[ "$version" != *"dev"* ]]; then
     rm -rf "${INDICES_OUTPUT_DIR:?}"
     mkdir -p "$INDICES_OUTPUT_DIR"
     # wheel-dir is overridden to be the commit directory, so that the indices point to the correct wheel path
-    $PYTHON .buildkite/scripts/generate-nightly-index.py \
+    $PYTHON .buildkite/release/scripts/generate-nightly-index.py \
         --version "$s3_version" \
         --wheel-dir "$BUILDKITE_COMMIT" \
         --current-objects "$obj_json" \

.buildkite/release/scripts/publish-release-images.sh

@@ -0,0 +1,50 @@
+#!/bin/bash
+# SPDX-License-Identifier: Apache-2.0
+#
+# Publish release Docker images from ECR to DockerHub.
+# Pulls per-arch images, tags with `latest` and versioned tags, pushes them,
+# then creates and pushes a multi-arch manifest.
+
+set -euo pipefail
+
+RELEASE_VERSION=$(buildkite-agent meta-data get release-version --default "" | sed 's/^v//')
+if [ -z "${RELEASE_VERSION}" ]; then
+  echo "ERROR: release-version metadata not set"
+  exit 1
+fi
+
+COMMIT="$BUILDKITE_COMMIT"
+DOCKERHUB_REPO="vllm/vllm-omni"
+ECR_REPO="public.ecr.aws/q9t5s3a7/vllm-omni-release-repo"
+
+echo "========================================"
+echo "Publishing release images v${RELEASE_VERSION}"
+echo "  Commit: ${COMMIT}"
+echo "========================================"
+
+# Login to ECR to pull staging images
+aws ecr-public get-login-password --region us-east-1 | \
+  docker login --username AWS --password-stdin public.ecr.aws/q9t5s3a7
+
+docker pull "${ECR_REPO}:${COMMIT}-x86_64"
+docker pull "${ECR_REPO}:${COMMIT}-aarch64"
+
+docker tag "${ECR_REPO}:${COMMIT}-x86_64" "${DOCKERHUB_REPO}:latest-x86_64"
+docker tag "${ECR_REPO}:${COMMIT}-x86_64" "${DOCKERHUB_REPO}:v${RELEASE_VERSION}-x86_64"
+docker push "${DOCKERHUB_REPO}:latest-x86_64"
+docker push "${DOCKERHUB_REPO}:v${RELEASE_VERSION}-x86_64"
+
+docker tag "${ECR_REPO}:${COMMIT}-aarch64" "${DOCKERHUB_REPO}:latest-aarch64"
+docker tag "${ECR_REPO}:${COMMIT}-aarch64" "${DOCKERHUB_REPO}:v${RELEASE_VERSION}-aarch64"
+docker push "${DOCKERHUB_REPO}:latest-aarch64"
+docker push "${DOCKERHUB_REPO}:v${RELEASE_VERSION}-aarch64"
+
+docker manifest rm "${DOCKERHUB_REPO}:latest" || true
+docker manifest rm "${DOCKERHUB_REPO}:v${RELEASE_VERSION}" || true
+docker manifest create "${DOCKERHUB_REPO}:latest" "${DOCKERHUB_REPO}:latest-x86_64" "${DOCKERHUB_REPO}:latest-aarch64"
+docker manifest create "${DOCKERHUB_REPO}:v${RELEASE_VERSION}" "${DOCKERHUB_REPO}:v${RELEASE_VERSION}-x86_64" "${DOCKERHUB_REPO}:v${RELEASE_VERSION}-aarch64"
+docker manifest push "${DOCKERHUB_REPO}:latest"
+docker manifest push "${DOCKERHUB_REPO}:v${RELEASE_VERSION}"
+
+echo ""
+echo "Successfully published release images for v${RELEASE_VERSION}"

.buildkite/release/scripts/publish-release-wheels-pypi.sh

@@ -0,0 +1,52 @@
+#!/usr/bin/env bash
+# SPDX-License-Identifier: Apache-2.0
+
+set -e
+
+BUCKET="vllm-wheels"
+SUBPATH="omni/$BUILDKITE_COMMIT"
+S3_COMMIT_PREFIX="s3://$BUCKET/$SUBPATH/"
+
+RELEASE_VERSION=$(buildkite-agent meta-data get release-version | sed 's/^v//')
+if [[ -z "$RELEASE_VERSION" ]]; then
+  echo "[FATAL] release-version metadata not set."
+  exit 1
+fi
+echo "Release version: $RELEASE_VERSION"
+
+if [[ -z "$PYPI_TOKEN" ]]; then
+  echo "[FATAL] PYPI_TOKEN is not set."
+  exit 1
+else
+  export TWINE_USERNAME="__token__"
+  export TWINE_PASSWORD="$PYPI_TOKEN"
+fi
+
+set -x
+
+if ! command -v uv &> /dev/null; then
+  curl -LsSf https://astral.sh/uv/install.sh | sh
+  export PATH="$HOME/.local/bin:$PATH"
+fi
+
+uv venv --python 3.12 /tmp/vllm-omni-release-env
+source /tmp/vllm-omni-release-env/bin/activate
+uv pip install twine
+
+DIST_DIR=/tmp/vllm-omni-release-dist
+mkdir -p "$DIST_DIR"
+
+echo "Downloading wheels from S3:"
+aws s3 ls "$S3_COMMIT_PREFIX"
+aws s3 cp --recursive --exclude "*" --include "vllm_omni-${RELEASE_VERSION}*.whl" --exclude "*dev*" "$S3_COMMIT_PREFIX" "$DIST_DIR"
+ls -la "$DIST_DIR"
+
+PYPI_WHEEL_FILES=$(find "$DIST_DIR" -name "vllm_omni-${RELEASE_VERSION}*.whl" -not -name "*+*")
+if [[ -z "$PYPI_WHEEL_FILES" ]]; then
+  echo "[FATAL] No wheels found for version ${RELEASE_VERSION}"
+  exit 1
+fi
+
+python3 -m twine check $PYPI_WHEEL_FILES
+python3 -m twine upload --non-interactive --verbose $PYPI_WHEEL_FILES
+echo "Wheels uploaded to PyPI"

.buildkite/scripts/docker_login_ecr_public.sh

@@ -1,73 +0,0 @@
-#!/bin/bash
-# Helper function to safely login to ECR Public with per-job config isolation
-# Uses DOCKER_CONFIG environment variable to prevent race conditions
-#
-# This script prevents the "device or resource busy" error by giving each
-# Buildkite job its own isolated Docker config directory.
-#
-# Usage:
-#   source docker_login_ecr_public.sh && safe_docker_login_ecr_public
-
-set -euo pipefail
-
-# Configuration
-ECR_REGISTRY="public.ecr.aws"
-
-setup_isolated_docker_config() {
-    # Use BUILDKITE_JOB_ID for job-specific isolation
-    # Fallback to PID if running outside Buildkite
-    local job_id="${BUILDKITE_JOB_ID:-$$}"
-
-    # Set Docker config to job-specific directory
-    export DOCKER_CONFIG="/tmp/docker-config-${job_id}"
-
-    # Create directory if it doesn't exist
-    mkdir -p "$DOCKER_CONFIG"
-
-    echo "[docker-config] Using isolated Docker config: $DOCKER_CONFIG"
-}
-
-check_docker_auth() {
-    # Check if already authenticated to the given registry
-    # Returns 0 if authenticated, 1 if not
-    local registry="$1"
-
-    # Check if credentials exist in the isolated config
-    if [[ -f "$DOCKER_CONFIG/config.json" ]]; then
-        # Check if registry is present in config
-        if grep -q "$registry" "$DOCKER_CONFIG/config.json" 2>/dev/null; then
-            return 0
-        fi
-    fi
-
-    return 1
-}
-
-safe_docker_login_ecr_public() {
-    # Setup isolated config first
-    setup_isolated_docker_config
-
-    local registry="$ECR_REGISTRY"
-
-    # Check if already authenticated (within this job)
-    if check_docker_auth "$registry"; then
-        echo "[docker-login] Already authenticated to $registry in this job"
-        return 0
-    fi
-
-    # Perform login to isolated config directory
-    echo "[docker-login] Logging in to $ECR_REGISTRY (isolated config)..."
-    if aws ecr-public get-login-password --region us-east-1 | docker login --username AWS --password-stdin "$ECR_REGISTRY"; then
-        echo "[docker-login] Login successful (config: $DOCKER_CONFIG)"
-        return 0
-    else
-        local exit_code=$?
-        echo "[docker-login] ERROR: Login failed with exit code $exit_code" >&2
-        return $exit_code
-    fi
-}
-
-# Execute if run as script (not sourced)
-if [[ "${BASH_SOURCE[0]}" == "${0}" ]]; then
-    safe_docker_login_ecr_public
-fi

.buildkite/scripts/hardware_ci/run_npu_test.sh

@@ -1,145 +0,0 @@
-#!/bin/bash
-
-# This script build the Ascend NPU docker image and run the offline inference inside the container.
-# It serves a sanity check for compilation and basic model usage.
-set -ex
-
-# Base ubuntu image with basic ascend development libraries and python installed
-VLLM_OMNI_REPO="https://github.com/vllm-project/vllm-omni.git"
-BASE_IMAGE_NAME="quay.nju.edu.cn/ascend/vllm-ascend:v0.11.0rc2"
-image_name="npu/vllm-omni-ci:${BUILDKITE_COMMIT}_${EPOCHSECONDS}"
-# image_name="npu/vllm-ci:${BUILDKITE_COMMIT}_${EPOCHSECONDS}"
-container_name="npu_${BUILDKITE_COMMIT}_$(tr -dc A-Za-z0-9 < /dev/urandom | head -c 10; echo)"
-
-# BUILDKITE_AGENT_NAME format is {hostname}-{agent_idx}-{npu_card_num}cards
-agent_idx=$(echo "${BUILDKITE_AGENT_NAME}" | awk -F'-' '{print $(NF-1)}')
-echo "agent_idx: ${agent_idx}"
-builder_name="cachebuilder${agent_idx}"
-builder_cache_dir="/mnt/docker-cache${agent_idx}"
-mkdir -p ${builder_cache_dir}
-
-# Try building the docker image
-cat <<EOF | DOCKER_BUILDKIT=1 docker build \
-    --add-host pypi-cache:${PYPI_CACHE_HOST} \
-    --builder ${builder_name} --cache-from type=local,src=${builder_cache_dir} \
-                           --cache-to type=local,dest=${builder_cache_dir},mode=max \
-    --build-arg BUILDKITE_PULL_REQUEST="${BUILDKITE_PULL_REQUEST}" \
-    --build-arg BUILDKITE_PULL_REQUEST_REPO="${BUILDKITE_PULL_REQUEST_REPO}" \
-    --progress=plain --load -t ${image_name} -f - .
-FROM ${BASE_IMAGE_NAME}
-
-# Define environments
-ENV DEBIAN_FRONTEND=noninteractive
-
-RUN pip config set global.index-url http://pypi-cache:${PYPI_CACHE_PORT}/pypi/simple && \
-    pip config set global.trusted-host pypi-cache && \
-    apt-get update -y && \
-    apt-get install -y python3-pip git vim wget net-tools gcc g++ cmake libnuma-dev && \
-    rm -rf /var/cache/apt/* && \
-    rm -rf /var/lib/apt/lists/*
-
-# Install for pytest to make the docker build cache layer always valid
-RUN --mount=type=cache,target=/root/.cache/pip \
-    pip install pytest>=6.0  pytest-cov modelscope
-
-COPY . .
-
-# Install vllm-omni
-WORKDIR /workspace
-ARG VLLM_OMNI_REPO=https://github.com/vllm-project/vllm-omni.git
-ARG VLLM_OMNI_TAG=main
-ARG BUILDKITE_PULL_REQUEST
-ARG BUILDKITE_PULL_REQUEST_REPO
-RUN git config --global url."https://gh-proxy.test.osinfra.cn/https://github.com/".insteadOf "https://github.com/" && \
-    if [ "\$BUILDKITE_PULL_REQUEST" != "false" ] && [ -n "\$BUILDKITE_PULL_REQUEST" ]; then \
-        echo "Cloning and checking out PR #\$BUILDKITE_PULL_REQUEST..." && \
-        git clone \$VLLM_OMNI_REPO /workspace/vllm-omni && \
-        cd /workspace/vllm-omni && \
-        git fetch origin pull/\$BUILDKITE_PULL_REQUEST/head:pr-\$BUILDKITE_PULL_REQUEST && \
-        git checkout pr-\$BUILDKITE_PULL_REQUEST; \
-    else \
-        echo "Not a PR build, using main branch" && \
-        git clone --depth 1 \$VLLM_OMNI_REPO /workspace/vllm-omni; \
-    fi
-
-RUN --mount=type=cache,target=/root/.cache/pip \
-    export PIP_EXTRA_INDEX_URL=https://mirrors.huaweicloud.com/ascend/repos/pypi && \
-    source /usr/local/Ascend/ascend-toolkit/set_env.sh && \
-    source /usr/local/Ascend/nnal/atb/set_env.sh && \
-    export LD_LIBRARY_PATH=\$LD_LIBRARY_PATH:/usr/local/Ascend/ascend-toolkit/latest/`uname -i`-linux/devlib && \
-    python3 -m pip install -v -e /workspace/vllm-omni/
-
-ENV VLLM_WORKER_MULTIPROC_METHOD=spawn
-
-WORKDIR /workspace/vllm-omni
-CMD ["/bin/bash"]
-
-EOF
-
-# Setup cleanup
-remove_docker_container() {
-  docker rm -f "${container_name}" || true;
-  docker image rm -f "${image_name}" || true;
-  docker system prune -f || true;
-}
-trap remove_docker_container EXIT
-
-# Generate corresponding --device args based on BUILDKITE_AGENT_NAME
-# Ascend NPU BUILDKITE_AGENT_NAME format is {hostname}-{agent_idx}-{npu_card_num}cards, and agent_idx starts from 1.
-#   e.g. atlas-a2-001-1-2cards means this is the 1-th agent on atlas-a2-001 host, and it has 2 NPU cards.
-#   returns --device /dev/davinci0 --device /dev/davinci1
-parse_and_gen_devices() {
-    local input="$1"
-    local index cards_num
-    if [[ "$input" =~ ([0-9]+)-([0-9]+)cards$ ]]; then
-        index="${BASH_REMATCH[1]}"
-        cards_num="${BASH_REMATCH[2]}"
-    else
-        echo "parse error" >&2
-        return 1
-    fi
-
-    local devices=""
-    local i=0
-    while (( i < cards_num )); do
-        local dev_idx=$(((index - 1)*cards_num + i ))
-        devices="$devices --device /dev/davinci${dev_idx}"
-        ((i++))
-    done
-
-    # trim leading space
-    devices="${devices#"${devices%%[![:space:]]*}"}"
-    # Output devices: assigned to the caller variable
-    printf '%s' "$devices"
-}
-
-devices=$(parse_and_gen_devices "${BUILDKITE_AGENT_NAME}") || exit 1
-
-# Run the image and execute the Out-Of-Tree (OOT) platform interface test case on Ascend NPU hardware.
-# This test checks whether the OOT platform interface is functioning properly in conjunction with
-# the hardware plugin vllm-ascend.
-hf_model_cache_dir=/mnt/hf_cache${agent_idx}
-ms_model_cache_dir=/mnt/modelscope${agent_idx}
-mkdir -p ${hf_model_cache_dir}
-mkdir -p ${ms_model_cache_dir}
-docker run \
-    --init \
-    ${devices} \
-    --device /dev/davinci_manager \
-    --device /dev/devmm_svm \
-    --device /dev/hisi_hdc \
-    -v /usr/local/dcmi:/usr/local/dcmi \
-    -v /usr/local/bin/npu-smi:/usr/local/bin/npu-smi \
-    -v /usr/local/Ascend/driver/lib64/:/usr/local/Ascend/driver/lib64/ \
-    -v /usr/local/Ascend/driver/version.info:/usr/local/Ascend/driver/version.info \
-    -v /etc/ascend_install.info:/etc/ascend_install.info \
-    -v ${hf_model_cache_dir}:/root/.cache/huggingface \
-    -v ${ms_model_cache_dir}:/root/.cache/modelscope \
-    --network host \
-    --entrypoint="" \
-    --name "${container_name}" \
-    "${image_name}" \
-    bash -c '
-    set -e
-    VLLM_USE_MODELSCOPE=True pytest -s -v tests/e2e/offline_inference/test_qwen2_5_omni.py
-'

.buildkite/scripts/upload_pipeline_with_skip_ci.sh

@@ -1,137 +0,0 @@
-#!/usr/bin/env bash
-# Evaluate docs-only skip-ci and upload continuation steps from the same `.buildkite/pipeline.yml`
-# (YAML document after the first `---`). Buildkite `if` is evaluated at upload time.
-set -euo pipefail
-
-ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/../.." && pwd)"
-PIPELINE_YML="${ROOT}/.buildkite/pipeline.yml"
-
-# Prints a single digit to stdout: 1 = skip image CI, 0 = run. Logs go to stderr.
-is_docs_only_change() {
-  local file_path
-  local has_any=0
-
-  while IFS= read -r file_path; do
-    [[ -z "${file_path}" ]] && continue
-    has_any=1
-
-    if [[ "${file_path}" == docs/* ]]; then
-      continue
-    fi
-    if [[ "${file_path}" == *.md ]]; then
-      continue
-    fi
-    if [[ "${file_path}" == "mkdocs.yaml" ]]; then
-      continue
-    fi
-    return 1
-  done
-
-  [[ "${has_any}" -eq 1 ]]
-}
-
-resolve_skip_ci() {
-  local is_pr_build=0
-  local files
-  local base_branch base_ref
-
-  if [[ "${BUILDKITE_PULL_REQUEST:-false}" != "false" && -n "${BUILDKITE_PULL_REQUEST:-}" ]]; then
-    is_pr_build=1
-  fi
-
-  if [[ "${is_pr_build}" -eq 1 ]]; then
-    base_branch="${BUILDKITE_PULL_REQUEST_BASE_BRANCH:-main}"
-    if ! git rev-parse --verify "origin/${base_branch}" >/dev/null 2>&1; then
-      echo "resolve_skip_ci: origin/${base_branch} not found locally; trying fetch" >&2
-      git fetch --depth=200 origin "${base_branch}" >/dev/null 2>&1 || true
-    fi
-
-    base_ref=""
-    if git rev-parse --verify "origin/${base_branch}" >/dev/null 2>&1; then
-      base_ref="origin/${base_branch}"
-    elif git rev-parse --verify "${base_branch}" >/dev/null 2>&1; then
-      base_ref="${base_branch}"
-    else
-      echo "resolve_skip_ci: cannot resolve PR base ${base_branch}; skip-ci=0" >&2
-      echo -n 0
-      return 0
-    fi
-
-    if ! files="$(git diff --name-only "${base_ref}...${BUILDKITE_COMMIT}" 2>/dev/null)"; then
-      echo "resolve_skip_ci: failed to compute PR changed files; skip-ci=0" >&2
-      echo -n 0
-      return 0
-    fi
-  elif [[ "${BUILDKITE_BRANCH:-}" == "main" ]]; then
-    if ! git rev-parse --verify "${BUILDKITE_COMMIT}^" >/dev/null 2>&1; then
-      echo "resolve_skip_ci: commit has no parent on main; skip-ci=0" >&2
-      echo -n 0
-      return 0
-    fi
-    if ! files="$(git diff --name-only "${BUILDKITE_COMMIT}^..${BUILDKITE_COMMIT}" 2>/dev/null)"; then
-      echo "resolve_skip_ci: failed to compute main changed files; skip-ci=0" >&2
-      echo -n 0
-      return 0
-    fi
-  else
-    echo "resolve_skip_ci: not PR/main build; skip-ci=0" >&2
-    echo -n 0
-    return 0
-  fi
-
-  if is_docs_only_change <<< "${files}"; then
-    echo "resolve_skip_ci: docs-only change detected; skip-ci=1" >&2
-    echo -n 1
-    return 0
-  fi
-
-  echo "resolve_skip_ci: non-doc changes detected; skip-ci=0" >&2
-  echo -n 0
-}
-
-SKIP_CI="$(resolve_skip_ci)"
-
-if [[ ! -f "${PIPELINE_YML}" ]]; then
-  echo "upload_pipeline_with_skip_ci: missing ${PIPELINE_YML}" >&2
-  exit 1
-fi
-
-export ROOT SKIP_CI PIPELINE_YML
-python3 <<'PY' | buildkite-agent pipeline upload
-import os
-import pathlib
-
-path = pathlib.Path(os.environ["PIPELINE_YML"])
-text = path.read_text(encoding="utf-8")
-sep = "\n---\n"
-if sep not in text:
-    raise SystemExit(
-        "upload_pipeline_with_skip_ci: .buildkite/pipeline.yml must contain a '\\n---\\n' separator "
-        "(document 1 = bootstrap, document 2 = uploaded steps)"
-    )
-_, continuation = text.split(sep, 1)
-
-skip = os.environ.get("SKIP_CI") == "1"
-# When docs-only skip-ci: skip default CI image, but still build for L4 nightly (PR label nightly-test or
-# main NIGHTLY=1), otherwise upload-nightly (depends_on image-build) would be skipped too.
-nightly_only = (
-    '(build.pull_request.labels includes "nightly-test") '
-    '|| (build.branch == "main" && build.env("NIGHTLY") == "1")'
-)
-# Placeholder in pipeline.yml is `if: __IMAGE_BUILD_IF__` (valid YAML); replace value only.
-if skip:
-    rep = f"'{nightly_only}'"
-    ready_rep = "'false'"
-    merge_rep = "'false'"
-else:
-    rep = "'true'"
-    ready_rep = "'build.branch != \"main\" && build.pull_request.labels includes \"ready\"'"
-    merge_rep = "'(build.branch == \"main\" && build.env(\"NIGHTLY\") != \"1\") || (build.branch != \"main\" && build.pull_request.labels includes \"merge-test\")'"
-rendered = (
-    continuation
-    .replace("__IMAGE_BUILD_IF__", rep)
-    .replace("__UPLOAD_READY_IF__", ready_rep)
-    .replace("__UPLOAD_MERGE_IF__", merge_rep)
-)
-print(rendered, end="")
-PY

.buildkite/test-amd-merge.yml

@@ -1,221 +0,0 @@
-steps:
-
-- label: "Simple Unit Test"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export VLLM_ROCM_USE_AITER=0
-    - "timeout 20m pytest -v -s -m 'core_model and cpu' --cov=vllm_omni --cov-branch --cov-report=term-missing --cov-report=html --cov-report=xml"
-
-- label: "Diffusion Model Test"
-  agent_pool: mi325_2
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export GPU_ARCHS=gfx942
-    - timeout 20m pytest -s -v tests/e2e/offline_inference/test_t2i_model.py -m "advanced_model and diffusion" --run-level "advanced_model"
-
-- label: "Diffusion Images API LoRA E2E"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - timeout 20m pytest -s -v tests/e2e/online_serving/test_images_generations_lora.py
-
-- label: "Diffusion Model CPU offloading Test"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export VLLM_LOGGING_LEVEL=DEBUG
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - |
-      timeout 20m bash -c '
-        set +e
-        pytest -s -v tests/e2e/offline_inference/test_diffusion_cpu_offload.py
-        EXIT1=\$?
-        pytest -s -v tests/e2e/offline_inference/test_diffusion_layerwise_offload.py
-        EXIT2=\$?
-        exit \$((EXIT1 | EXIT2))
-      '
-
-## ISSUE depends on `diffusers` package: https://github.com/huggingface/diffusers/issues/13274
-# - label: "Audio Generation Model Test"
-#   agent_pool: mi325_1
-#   depends_on: amd-build
-#   mirror_hardwares: [amdproduction]
-#   grade: Blocking
-#   commands:
-#     - export GPU_ARCHS=gfx942
-#     - export VLLM_LOGGING_LEVEL=DEBUG
-#     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-#     - timeout 20m pytest -s -v tests/e2e/offline_inference/test_stable_audio_expansion.py -m "advanced_model and diffusion and L4" --run-level advanced_model
-
-- label: "Diffusion Cache Backend Test"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export VLLM_LOGGING_LEVEL=DEBUG
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - timeout 15m pytest -s -v -m "core_model and cache and diffusion and not distributed_cuda and L4"
-
-- label: "Diffusion Sequence Parallelism Test (Need 4 GPUs)"
-  agent_pool: mi325_4
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export VLLM_LOGGING_LEVEL=DEBUG
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - timeout 20m pytest -s -v tests/e2e/offline_inference/test_sequence_parallel.py
-    - timeout 20m pytest -s -v tests/diffusion/distributed/test_ulysses_uaa_perf.py
-
-# merge-only tests
-- label: "Diffusion Tensor Parallelism Test"
-  agent_pool: mi325_2
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - pytest -s -v tests/e2e/offline_inference/test_zimage_parallelism.py
-
-- label: "Diffusion GPU Worker Test"
-  agent_pool: mi325_2
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - timeout 20m pytest -s -v tests/diffusion/test_diffusion_worker.py
-
-- label: "Engine Test"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - timeout 20m pytest -s -v tests/engine/test_async_omni_engine_abort.py
-
-- label: "Omni Model Test Qwen2-5-Omni"
-  agent_pool: mi325_2
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export VLLM_LOGGING_LEVEL=DEBUG
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - timeout 20m pytest -s -v tests/e2e/offline_inference/test_qwen2_5_omni.py
-    - timeout 20m pytest -s -v tests/e2e/online_serving/test_qwen2_5_omni.py -m "advanced_model" --run-level "advanced_model"
-
-- label: "Omni Model Test Qwen3-Omni"
-  agent_pool: mi325_2
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export VLLM_LOGGING_LEVEL=DEBUG
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - export VLLM_TEST_CLEAN_GPU_MEMORY=1
-    - timeout 30m pytest -s -v tests/e2e/offline_inference/test_qwen3_omni.py tests/e2e/online_serving/test_qwen3_omni.py tests/e2e/online_serving/test_mimo_audio.py -m "advanced_model" --run-level "advanced_model"
-
-- label: "Qwen3-TTS CustomVoice E2E Test"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - |
-      timeout 20m bash -c '
-        export VLLM_LOGGING_LEVEL=DEBUG
-        export VLLM_WORKER_MULTIPROC_METHOD=spawn
-        export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
-        pytest -s -v tests/e2e/online_serving/test_qwen3_tts_customvoice.py tests/e2e/offline_inference/test_qwen3_tts_customvoice.py -m "advanced_model" --run-level "advanced_model"
-      '
-
-- label: "Qwen3-TTS Base E2E Test"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - |
-      timeout 30m bash -c '
-        export VLLM_LOGGING_LEVEL=DEBUG
-        export VLLM_WORKER_MULTIPROC_METHOD=spawn
-        export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
-        pytest -s -v tests/e2e/online_serving/test_qwen3_tts_base.py tests/e2e/offline_inference/test_qwen3_tts_base.py -m "advanced_model" --run-level "advanced_model"
-      '
-
-- label: "Diffusion Image Edit Test"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export GPU_ARCHS=gfx942
-    - export VLLM_LOGGING_LEVEL=DEBUG
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - timeout 20m pytest -s -v tests/e2e/online_serving/test_image_gen_edit.py
-
-# TODO: Bagel test on ROCm is very unstable. @tjtanaa
-# Need to debug before reneable numerical changes across large PRs
-# # split Bagel Model Test with H100 (Real Weights) into three tests
-# - label: "Bagel Text2Img Model Test (1/3)"
-#   agent_pool: mi325_1
-#   depends_on: amd-build
-#   mirror_hardwares: [amdproduction]
-#   grade: Blocking
-#   commands:
-#     - export GPU_ARCHS=gfx942
-#     - export VLLM_TEST_CLEAN_GPU_MEMORY=1
-#     - export VLLM_LOGGING_LEVEL=DEBUG
-#     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-#     - export VLLM_ROCM_USE_AITER_RMSNORM=0
-#     - timeout 30m pytest -s -v tests/e2e/offline_inference/test_bagel_text2img.py -m "advanced_model" --run-level "advanced_model" -k "shared_memory" -k "rocm"
-
-# - label: "Bagel Img2Img Model Test (2/3)"
-#   agent_pool: mi325_1
-#   depends_on: amd-build
-#   mirror_hardwares: [amdproduction]
-#   grade: Blocking
-#   commands:
-#     - export GPU_ARCHS=gfx942
-#     - export VLLM_TEST_CLEAN_GPU_MEMORY=1
-#     - export VLLM_LOGGING_LEVEL=DEBUG
-#     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-#     - export VLLM_ROCM_USE_AITER_RMSNORM=0
-#     - timeout 30m pytest -s -v tests/e2e/offline_inference/test_bagel_img2img.py -m "advanced_model" --run-level "advanced_model" -k "rocm"
-
-# - label: "Bagel Online Serving Test (3/3)"
-#   agent_pool: mi325_1
-#   depends_on: amd-build
-#   mirror_hardwares: [amdproduction]
-#   grade: Blocking
-#   commands:
-#     - export GPU_ARCHS=gfx942
-#     - export VLLM_TEST_CLEAN_GPU_MEMORY=1
-#     - export VLLM_IMAGE_FETCH_TIMEOUT=60
-#     - export VLLM_LOGGING_LEVEL=DEBUG
-#     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-#     - export VLLM_ROCM_USE_AITER_RMSNORM=0
-#     - timeout 40m pytest -s -v tests/e2e/online_serving/test_bagel_online.py -m "advanced_model" --run-level "advanced_model" -k "rocm"
-
-- label: "Voxtral-TTS E2E Test"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - |
-      timeout 20m bash -c '
-        export VLLM_LOGGING_LEVEL=DEBUG
-        export VLLM_WORKER_MULTIPROC_METHOD=spawn
-        pytest -s -v tests/e2e/online_serving/test_voxtral_tts.py tests/e2e/offline_inference/test_voxtral_tts.py -m "advanced_model" --run-level "advanced_model"
-      '

.buildkite/test-amd-ready.yaml

@@ -1,240 +0,0 @@
-steps:
-
-- label: "Simple Unit Test"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export VLLM_ROCM_USE_AITER=0
-    - "timeout 20m pytest -v -s -m 'core_model and cpu' --cov=vllm_omni --cov-branch --cov-report=term-missing --cov-report=html --cov-report=xml"
-
-- label: "Voxtral TTS CUDA Unit Test"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - timeout 10m pytest -s -v tests/model_executor/models/voxtral_tts/test_cuda_graph_acoustic_transformer.py
-
-- label: "Diffusion Model Test"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - timeout 30m pytest -s -v tests/e2e/offline_inference/test_t2i_model.py -m "core_model and diffusion" --run-level "core_model"
-
-- label: "Diffusion Batching Test"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - timeout 20m pytest -s -v tests/e2e/offline_inference/test_qwen_image_diffusion_batching.py -m "core_model and diffusion" --run-level "core_model"
-
-- label: "Custom Pipeline Test"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - timeout 20m pytest -s -v tests/e2e/offline_inference/custom_pipeline/ -m "core_model"
-
-- label: "Diffusion Model CPU offloading Test"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export VLLM_LOGGING_LEVEL=DEBUG
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - |
-      timeout 20m bash -c '
-        set +e
-        pytest -s -v tests/e2e/offline_inference/test_diffusion_cpu_offload.py
-        EXIT1=\$?
-        pytest -s -v tests/e2e/offline_inference/test_diffusion_layerwise_offload.py
-        EXIT2=\$?
-        exit \$((EXIT1 | EXIT2))
-      '
-
-## ISSUE depends on `diffusers` package: https://github.com/huggingface/diffusers/issues/13274
-# - label: "Audio Generation Model Test"
-#   agent_pool: mi325_1
-#   depends_on: amd-build
-#   mirror_hardwares: [amdproduction]
-#   grade: Blocking
-#   commands:
-#     - export GPU_ARCHS=gfx942
-#     - export VLLM_LOGGING_LEVEL=DEBUG
-#     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-#     - timeout 20m pytest -s -v tests/e2e/offline_inference/test_stable_audio_expansion.py -m "advanced_model and diffusion and L4" --run-level advanced_model
-
-- label: "Diffusion Cache Backend Test"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export GPU_ARCHS=gfx942
-    - export VLLM_LOGGING_LEVEL=DEBUG
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - timeout 15m pytest -s -v -m "core_model and cache and diffusion and not distributed_cuda and L4"
-
-- label: "Diffusion Sequence Parallelism Test"
-  agent_pool: mi325_2
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export VLLM_LOGGING_LEVEL=DEBUG
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - timeout 20m pytest -s -v tests/e2e/offline_inference/test_sequence_parallel.py -m core_model
-
-- label: "Diffusion GPU Worker Test"
-  agent_pool: mi325_2
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - timeout 20m pytest -s -v tests/diffusion/test_diffusion_worker.py
-
-- label: "Engine Test"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - |
-      timeout 15m bash -c '
-              pytest -s -v tests/engine/test_async_omni_engine_abort.py
-      '
-
-
-# NOTE: This test is not running any thing. It is skipped and deselected.
-# Currently it is = 1 skipped, 1 deselected, 17 warnings in 0.03s ======
-# - label: "Omni Model Test Qwen2-5-Omni"
-#   agent_pool: mi325_2
-#   depends_on: amd-build
-#   mirror_hardwares: [amdproduction]
-#   grade: Blocking
-#   commands:
-#     - export VLLM_LOGGING_LEVEL=DEBUG
-#     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-#     - timeout 20m pytest -s -v tests/e2e/offline_inference/test_qwen2_5_omni.py -m "core_model" --run-level "core_model"
-
-# - label: "Omni Model Test Qwen3-Omni"
-#   agent_pool: mi325_2
-#   depends_on: amd-build
-#   mirror_hardwares: [amdproduction]
-#   grade: Blocking
-#   commands:
-#     - export VLLM_LOGGING_LEVEL=DEBUG
-#     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-#     - export VLLM_TEST_CLEAN_GPU_MEMORY=1
-#     - timeout 10m pytest -s -v tests/e2e/offline_inference/test_qwen3_omni.py
-#     - timeout 20m pytest -s -v tests/e2e/online_serving/test_qwen3_omni.py -m "core_model" --run-level "core_model"
-
-- label: "MiMo-Audio E2E Test with H100"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - |
-      timeout 30m bash -c '
-        export VLLM_LOGGING_LEVEL=DEBUG
-        export VLLM_WORKER_MULTIPROC_METHOD=spawn
-        pytest -s -v tests/e2e/online_serving/test_mimo_audio.py -m "core_model" --run-level "core_model"
-      '
-
-- label: "Qwen3-TTS E2E Test"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export VLLM_LOGGING_LEVEL=DEBUG
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
-    - timeout 30m pytest -s -v tests/e2e/online_serving/test_qwen3_tts_customvoice.py -m "core_model" --run-level "core_model"
-
-- label: "Voxtral-TTS E2E Test"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - |
-      timeout 20m bash -c '
-        export VLLM_LOGGING_LEVEL=DEBUG
-        export VLLM_WORKER_MULTIPROC_METHOD=spawn
-        pytest -s -v tests/e2e/online_serving/test_voxtral_tts.py -m "advanced_model" --run-level "advanced_model"
-        pytest -s -v tests/e2e/offline_inference/test_voxtral_tts.py -m "advanced_model" --run-level "advanced_model"
-      '
-
-- label: "Diffusion Image Edit Test"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export GPU_ARCHS=gfx942
-    - export VLLM_LOGGING_LEVEL=DEBUG
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - timeout 20m pytest -s -v tests/e2e/online_serving/test_image_gen_edit.py
-
-# TODO: Bagel test on ROCm is very unstable. @tjtanaa
-# Need to debug before reneable numerical changes across large PRs
-# - label: "Bagel Text2Img Model Test"
-#   agent_pool: mi325_1
-#   depends_on: amd-build
-#   mirror_hardwares: [amdproduction]
-#   grade: Blocking
-#   commands:
-#     - export GPU_ARCHS=gfx942
-#     - export VLLM_TEST_CLEAN_GPU_MEMORY=1
-#     - export VLLM_LOGGING_LEVEL=DEBUG
-#     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-#     - export VLLM_ROCM_USE_AITER_RMSNORM=0
-#     - timeout 30m pytest -s -v tests/e2e/offline_inference/test_bagel_text2img.py -m "core_model" --run-level "core_model" -k "rocm"
-
-# - label: "Bagel Img2Img Model Test"
-#   agent_pool: mi325_1
-#   depends_on: amd-build
-#   mirror_hardwares: [amdproduction]
-#   grade: Blocking
-#   commands:
-#     - export GPU_ARCHS=gfx942
-#     - export VLLM_TEST_CLEAN_GPU_MEMORY=1
-#     - export VLLM_LOGGING_LEVEL=DEBUG
-#     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-#     - export VLLM_ROCM_USE_AITER_RMSNORM=0
-#     - timeout 30m pytest -s -v tests/e2e/offline_inference/test_bagel_img2img.py -m "core_model" --run-level "core_model" -k "rocm"
-
-# - label: "Bagel Online Serving Test"
-#   agent_pool: mi325_1
-#   depends_on: amd-build
-#   mirror_hardwares: [amdproduction]
-#   grade: Blocking
-#   commands:
-#     - export GPU_ARCHS=gfx942
-#     - export VLLM_TEST_CLEAN_GPU_MEMORY=1
-#     - export VLLM_IMAGE_FETCH_TIMEOUT=60
-#     - export VLLM_LOGGING_LEVEL=DEBUG
-#     - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-#     - export VLLM_ROCM_USE_AITER_RMSNORM=0
-#     - timeout 40m pytest -s -v tests/e2e/online_serving/test_bagel_online.py -m "core_model" --run-level "core_model" -k "rocm"
-
-- label: "CosyVoice3-TTS E2E Test"
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - |
-      timeout 20m bash -c '
-        pytest -s -v tests/e2e/online_serving/test_cosyvoice3_tts.py -m "core_model" --run-level "core_model"
-      '

.buildkite/test-amd.yaml

@@ -1,133 +0,0 @@
-steps:
-
-- label: "Diffusion Model Test"
-  timeout_in_minutes: 30
-  agent_pool: mi325_2
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export GPU_ARCHS=gfx942
-    - pytest -s -v tests/e2e/offline_inference/test_t2i_model.py
-
-- label: "Diffusion Images API LoRA E2E"
-  timeout_in_minutes: 30
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export GPU_ARCHS=gfx942
-    - export VLLM_LOGGING_LEVEL=DEBUG
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - pytest -s -v tests/e2e/online_serving/test_images_generations_lora.py
-
-- label: "Diffusion Model CPU offloading Test"
-  timeout_in_minutes: 20
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export GPU_ARCHS=gfx942
-    - export VLLM_LOGGING_LEVEL=DEBUG
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - pytest -s -v tests/e2e/offline_inference/test_diffusion_cpu_offload.py
-    - pytest -s -v tests/e2e/offline_inference/test_diffusion_layerwise_offload.py
-
-- label: "Diffusion Cache Backend Test"
-  timeout_in_minutes: 15
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export GPU_ARCHS=gfx942
-    - export VLLM_LOGGING_LEVEL=DEBUG
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - pytest -s -v -m 'core_model and cache and diffusion and not distributed_rocm and MI325'
-
-- label: "Diffusion Sequence Parallelism Test"
-  timeout_in_minutes: 20
-  agent_pool: mi325_2
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export GPU_ARCHS=gfx942
-    - export VLLM_LOGGING_LEVEL=DEBUG
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - pytest -s -v tests/e2e/offline_inference/test_sequence_parallel.py -m core_model
-
-- label: "Diffusion Tensor Parallelism Test"
-  timeout_in_minutes: 20
-  agent_pool: mi325_2
-  depends_on: amd-build
-  commands:
-    - export GPU_ARCHS=gfx942
-    - export VLLM_LOGGING_LEVEL=DEBUG
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - pytest -s -v tests/e2e/offline_inference/test_zimage_parallelism.py
-
-- label: "Diffusion GPU Worker Test"
-  timeout_in_minutes: 20
-  agent_pool: mi325_2
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - pytest -s -v tests/diffusion/test_diffusion_worker.py
-
-- label: "Omni Model Test Qwen2-5-Omni"
-  timeout_in_minutes: 15
-  agent_pool: mi325_2
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export GPU_ARCHS=gfx942
-    - export VLLM_LOGGING_LEVEL=DEBUG
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - pytest -s -v tests/e2e/offline_inference/test_qwen2_5_omni.py
-    - pytest -s -v tests/engine/test_async_omni_engine_abort.py
-
-- label: "Omni Model Test Qwen3-Omni"
-  timeout_in_minutes: 15
-  agent_pool: mi325_2
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export VLLM_LOGGING_LEVEL=DEBUG
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - export VLLM_TEST_CLEAN_GPU_MEMORY="1"
-    - export VLLM_ROCM_USE_AITER=0
-    - pytest -s -v tests/e2e/offline_inference/test_qwen3_omni.py
-    - pytest -s -v tests/e2e/online_serving/test_qwen3_omni.py
-
-
-- label: "Diffusion Image Edit Test"
-  timeout_in_minutes: 15
-  agent_pool: mi325_1
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export GPU_ARCHS=gfx942
-    - export VLLM_LOGGING_LEVEL=DEBUG
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - pytest -s -v tests/e2e/online_serving/test_image_gen_edit.py
-
-
-- label: "Omni Sleep Mode Test"
-  timeout_in_minutes: 40
-  agent_pool: mi325_2
-  depends_on: amd-build
-  mirror_hardwares: [amdproduction]
-  grade: Blocking
-  commands:
-    - export GPU_ARCHS=gfx942
-    - export VLLM_LOGGING_LEVEL=DEBUG
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - export VLLM_TEST_CLEAN_GPU_MEMORY="1"
-    - pytest -s -v tests/e2e/offline_inference/test_omni_sleep_mode.py -m "advanced_model and omni and MI325" --run-level "advanced_model"

.buildkite/test-merge.yml

@@ -1,444 +0,0 @@
-env:
-  VLLM_WORKER_MULTIPROC_METHOD: spawn
-  HF_HUB_DOWNLOAD_TIMEOUT: 300
-  HF_HUB_ETAG_TIMEOUT: 60
-
-steps:
-  - label: "Simple Unit Test"
-    depends_on: upload-merge-pipeline
-    commands:
-      - "pytest -v -s -m 'core_model and cpu' --cov=vllm_omni --cov-branch --cov-report=term-missing --cov-report=html --cov-report=xml"
-    agents:
-      queue: "gpu_1_queue"
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Diffusion Model Test"
-    timeout_in_minutes: 30
-    depends_on: upload-merge-pipeline
-    commands:
-      - pytest -s -v tests/e2e/offline_inference/test_t2i_model.py -m "advanced_model and diffusion" --run-level "advanced_model"
-    agents:
-      queue: "gpu_1_queue" # g6.4xlarge instance on AWS, has 1 L4 GPU
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Diffusion Images API LoRA E2E"
-    timeout_in_minutes: 30
-    depends_on: upload-merge-pipeline
-    commands:
-      - pytest -s -v tests/e2e/online_serving/test_images_generations_lora.py
-    agents:
-      queue: "gpu_1_queue" # g6.4xlarge instance on AWS, has 1 L4 GPU
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Diffusion Model CPU offloading Test"
-    timeout_in_minutes: 20
-    depends_on: upload-merge-pipeline
-    commands:
-      # Single pytest session for one combined summary at end of log.
-      - pytest -s -v tests/e2e/offline_inference/test_diffusion_cpu_offload.py tests/e2e/offline_inference/test_diffusion_layerwise_offload.py
-    agents:
-      queue: "gpu_1_queue" # g6.4xlarge instance on AWS, has 1 L4 GPU
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Diffusion Cache Backend Test"
-    timeout_in_minutes: 15
-    depends_on: upload-merge-pipeline
-    commands:
-      - pytest -s -v -m 'core_model and cache and diffusion and not distributed_cuda and L4'
-    agents:
-      queue: "gpu_1_queue" # g6.4xlarge instance on AWS, has 1 L4 GPU
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Diffusion Sequence Parallelism Test"
-    timeout_in_minutes: 25
-    depends_on: upload-merge-pipeline
-    commands:
-      - pytest -s -v tests/e2e/offline_inference/test_sequence_parallel.py tests/diffusion/distributed/test_ulysses_uaa_perf.py
-    agents:
-      queue: "gpu_4_queue" # g6.12xlarge instance on AWS, has 4 L4 GPU
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          shm-size: "8gb"
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Diffusion Tensor Parallelism Test"
-    timeout_in_minutes: 20
-    depends_on: upload-merge-pipeline
-    commands:
-      - pytest -s -v tests/e2e/offline_inference/test_zimage_parallelism.py
-    agents:
-      queue: "gpu_4_queue" # g6.12xlarge instance on AWS, has 4 L4 GPU
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          shm-size: "8gb"
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Diffusion GPU Worker Test"
-    timeout_in_minutes: 20
-    depends_on: upload-merge-pipeline
-    commands:
-      - pytest -s -v tests/diffusion/test_diffusion_worker.py
-    agents:
-      queue: "gpu_4_queue" # g6.12xlarge instance on AWS, has 4 L4 GPU
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          shm-size: "8gb"
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Engine Test"
-    depends_on: upload-merge-pipeline
-    commands:
-      - |
-        timeout 15m bash -c '
-                pytest -s -v tests/engine/test_async_omni_engine_abort.py
-        '
-    agents:
-      queue: "gpu_1_queue" # g6.12xlarge instance on AWS, has 4 L4 GPU
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          shm-size: "8gb"
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Omni Model Test"
-    timeout_in_minutes: 30
-    depends_on: upload-merge-pipeline
-    commands:
-      - export VLLM_LOGGING_LEVEL=DEBUG
-      - pytest -s -v tests/e2e/offline_inference/test_qwen2_5_omni.py tests/e2e/online_serving/test_qwen2_5_omni.py -m "advanced_model" --run-level "advanced_model"
-    agents:
-      queue: "gpu_4_queue" # g6.12xlarge instance on AWS, has 4 L4 GPU
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Qwen3-TTS CustomVoice E2E Test"
-    depends_on: upload-merge-pipeline
-    commands:
-      - |
-        timeout 20m bash -c '
-          export VLLM_LOGGING_LEVEL=DEBUG
-          export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
-          pytest -s -v tests/e2e/online_serving/test_qwen3_tts_customvoice.py tests/e2e/offline_inference/test_qwen3_tts_customvoice.py -m "advanced_model" --run-level "advanced_model"
-        '
-    agents:
-      queue: "gpu_1_queue"
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          shm-size: "8gb"
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Qwen3-TTS Base E2E Test"
-    depends_on: upload-merge-pipeline
-    commands:
-      - |
-        timeout 20m bash -c '
-          export VLLM_LOGGING_LEVEL=DEBUG
-          export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
-          pytest -s -v tests/e2e/online_serving/test_qwen3_tts_base.py tests/e2e/offline_inference/test_qwen3_tts_base.py -m "advanced_model" --run-level "advanced_model"
-        '
-    agents:
-      queue: "gpu_1_queue"
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          shm-size: "8gb"
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Omni Model Test with H100"
-    timeout_in_minutes: 30
-    depends_on: upload-merge-pipeline
-    commands:
-      - export VLLM_TEST_CLEAN_GPU_MEMORY="1"
-      - pytest -s -v tests/e2e/offline_inference/test_qwen3_omni.py tests/e2e/online_serving/test_qwen3_omni.py tests/e2e/online_serving/test_mimo_audio.py -m "advanced_model" --run-level "advanced_model"
-    agents:
-      queue: "mithril-h100-pool"
-    plugins:
-      - kubernetes:
-          podSpec:
-            containers:
-              - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                resources:
-                  limits:
-                    nvidia.com/gpu: 2
-                volumeMounts:
-                  - name: devshm
-                    mountPath: /dev/shm
-                  - name: hf-cache
-                    mountPath: /root/.cache/huggingface
-                env:
-                  - name: HF_HOME
-                    value: /root/.cache/huggingface
-                  - name: HF_TOKEN
-                    valueFrom:
-                      secretKeyRef:
-                        name: hf-token-secret
-                        key: token
-            nodeSelector:
-              node.kubernetes.io/instance-type: gpu-h100-sxm
-            volumes:
-              - name: devshm
-                emptyDir:
-                  medium: Memory
-              - name: hf-cache
-                hostPath:
-                  path: /mnt/hf-cache
-                  type: DirectoryOrCreate
-
-  - label: "Diffusion Image Edit Test with H100 (1 GPU)"
-    timeout_in_minutes: 20
-    depends_on: upload-merge-pipeline
-    commands:
-      - pytest -s -v tests/e2e/online_serving/test_image_gen_edit.py
-    agents:
-      queue: "mithril-h100-pool"
-    plugins:
-      - kubernetes:
-          podSpec:
-            containers:
-              - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                resources:
-                  limits:
-                    nvidia.com/gpu: 1
-                volumeMounts:
-                  - name: devshm
-                    mountPath: /dev/shm
-                  - name: hf-cache
-                    mountPath: /root/.cache/huggingface
-                env:
-                  - name: HF_HOME
-                    value: /root/.cache/huggingface
-                  - name: HF_TOKEN
-                    valueFrom:
-                      secretKeyRef:
-                        name: hf-token-secret
-                        key: token
-            nodeSelector:
-              node.kubernetes.io/instance-type: gpu-h100-sxm
-            volumes:
-              - name: devshm
-                emptyDir:
-                  medium: Memory
-              - name: hf-cache
-                hostPath:
-                  path: /mnt/hf-cache
-                  type: DirectoryOrCreate
-
-  - label: "Bagel Model Test with H100 (Real Weights)"
-    timeout_in_minutes: 60
-    depends_on: upload-merge-pipeline
-    commands:
-      - |
-        timeout 55m bash -c '
-          set -e
-          export VLLM_TEST_CLEAN_GPU_MEMORY=1
-          export VLLM_IMAGE_FETCH_TIMEOUT=60
-          pytest -s -v tests/e2e/offline_inference/test_bagel_text2img.py -m "advanced_model" --run-level "advanced_model" -k "shared_memory"
-          pytest -s -v tests/e2e/offline_inference/test_bagel_img2img.py tests/e2e/online_serving/test_bagel_online.py -m "advanced_model" --run-level "advanced_model"
-        '
-    agents:
-      queue: "mithril-h100-pool"
-    plugins:
-      - kubernetes:
-          podSpec:
-            containers:
-              - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                resources:
-                  limits:
-                    nvidia.com/gpu: 1
-                volumeMounts:
-                  - name: devshm
-                    mountPath: /dev/shm
-                  - name: hf-cache
-                    mountPath: /root/.cache/huggingface
-                env:
-                  - name: HF_HOME
-                    value: /root/.cache/huggingface
-                  - name: HF_TOKEN
-                    valueFrom:
-                      secretKeyRef:
-                        name: hf-token-secret
-                        key: token
-            nodeSelector:
-              node.kubernetes.io/instance-type: gpu-h100-sxm
-            volumes:
-              - name: devshm
-                emptyDir:
-                  medium: Memory
-              - name: hf-cache
-                hostPath:
-                  path: /mnt/hf-cache
-                  type: DirectoryOrCreate
-
-  - label: "Omni Sleep Mode Test with H100"
-    timeout_in_minutes: 30
-    depends_on: upload-merge-pipeline
-    commands:
-      - export VLLM_TEST_CLEAN_GPU_MEMORY="1"
-      - pytest -s -v tests/e2e/offline_inference/test_omni_sleep_mode.py -m "advanced_model and H100 and omni" --run-level "advanced_model"
-    agents:
-      queue: "mithril-h100-pool"
-    plugins:
-      - kubernetes:
-          podSpec:
-            containers:
-              - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                resources:
-                  limits:
-                    nvidia.com/gpu: 2
-                volumeMounts:
-                  - name: devshm
-                    mountPath: /dev/shm
-                  - name: hf-cache
-                    mountPath: /root/.cache/huggingface
-                env:
-                  - name: HF_HOME
-                    value: /root/.cache/huggingface
-                  - name: HF_TOKEN
-                    valueFrom:
-                      secretKeyRef:
-                        name: hf-token-secret
-                        key: token
-            nodeSelector:
-              node.kubernetes.io/instance-type: gpu-h100-sxm
-            volumes:
-              - name: devshm
-                emptyDir:
-                  medium: Memory
-              - name: hf-cache
-                hostPath:
-                  path: /mnt/hf-cache
-                  type: DirectoryOrCreate
-
-  - label: "Voxtral-TTS E2E Test"
-    timeout_in_minutes: 20
-    depends_on: upload-merge-pipeline
-    commands:
-      - |
-        timeout 20m bash -c '
-          export VLLM_LOGGING_LEVEL=DEBUG
-          pytest -s -v tests/e2e/online_serving/test_voxtral_tts.py tests/e2e/offline_inference/test_voxtral_tts.py -m "advanced_model" --run-level "advanced_model"
-        '
-    agents:
-      queue: "mithril-h100-pool"
-    plugins:
-      - kubernetes:
-          podSpec:
-            containers:
-              - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                resources:
-                  limits:
-                    nvidia.com/gpu: 1
-                volumeMounts:
-                  - name: devshm
-                    mountPath: /dev/shm
-                  - name: hf-cache
-                    mountPath: /root/.cache/huggingface
-                env:
-                  - name: HF_HOME
-                    value: /root/.cache/huggingface
-                  - name: HF_TOKEN
-                    valueFrom:
-                      secretKeyRef:
-                        name: hf-token-secret
-                        key: token
-            nodeSelector:
-              node.kubernetes.io/instance-type: gpu-h100-sxm
-            volumes:
-              - name: devshm
-                emptyDir:
-                  medium: Memory
-              - name: hf-cache
-                hostPath:
-                  path: /mnt/hf-cache
-                  type: DirectoryOrCreate

.buildkite/test-nightly.yml

@@ -1,655 +0,0 @@
-env:
-  VLLM_WORKER_MULTIPROC_METHOD: spawn
-  HF_HUB_DOWNLOAD_TIMEOUT: 300
-  HF_HUB_ETAG_TIMEOUT: 60
-
-steps:
-  # Group: collapses under one heading in the Buildkite UI; child steps still run in parallel.
-  - group: ":card_index_dividers: Omni Model Test"
-    key: nightly-omni-test-group
-    depends_on: upload-nightly-pipeline
-    if: build.env("NIGHTLY") == "1" || build.pull_request.labels includes "nightly-test" || build.pull_request.labels includes "omni-test"
-    steps:
-      - label: ":full_moon: Omni · Function Test"
-        timeout_in_minutes: 90
-        commands:
-          - pytest -s -v tests/e2e/online_serving/test_*_expansion.py -m "advanced_model and H100 and omni" --run-level "advanced_model"
-        agents:
-          queue: "mithril-h100-pool"
-        plugins:
-          - kubernetes:
-              podSpec:
-                containers:
-                  - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                    resources:
-                      limits:
-                        nvidia.com/gpu: 2
-                    volumeMounts:
-                      - name: devshm
-                        mountPath: /dev/shm
-                      - name: hf-cache
-                        mountPath: /root/.cache/huggingface
-                    env:
-                      - name: HF_HOME
-                        value: /root/.cache/huggingface
-                      - name: HF_TOKEN
-                        valueFrom:
-                          secretKeyRef:
-                            name: hf-token-secret
-                            key: token
-                nodeSelector:
-                  node.kubernetes.io/instance-type: gpu-h100-sxm
-                volumes:
-                  - name: devshm
-                    emptyDir:
-                      medium: Memory
-                  - name: hf-cache
-                    hostPath:
-                      path: /mnt/hf-cache
-                      type: DirectoryOrCreate
-
-      - label: ":full_moon: Omni · Doc Test with L4"
-        timeout_in_minutes: 90
-        commands:
-          - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
-          - pytest -s -v tests/examples/ -m "advanced_model and omni and L4" --run-level "advanced_model"
-        agents:
-          queue: "gpu_4_queue" # g6.12xlarge instance on AWS, has 4 L4 GPU
-        plugins:
-          - docker#v5.2.0:
-              image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-              always-pull: true
-              propagate-environment: true
-              shm-size: "8gb"
-              environment:
-                - "HF_HOME=/fsx/hf_cache"
-                - "HF_TOKEN"
-              volumes:
-                - "/fsx/hf_cache:/fsx/hf_cache"
-
-      - label: ":full_moon: Omni · Doc Test with H100"
-        timeout_in_minutes: 90
-        commands:
-          - pytest -s -v tests/examples/ -m "advanced_model and omni and H100" --run-level "advanced_model"
-        agents:
-          queue: "mithril-h100-pool"
-        plugins:
-          - kubernetes:
-              podSpec:
-                containers:
-                  - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                    resources:
-                      limits:
-                        nvidia.com/gpu: 2
-                    volumeMounts:
-                      - name: devshm
-                        mountPath: /dev/shm
-                      - name: hf-cache
-                        mountPath: /root/.cache/huggingface
-                    env:
-                      - name: HF_HOME
-                        value: /root/.cache/huggingface
-                      - name: HF_TOKEN
-                        valueFrom:
-                          secretKeyRef:
-                            name: hf-token-secret
-                            key: token
-                nodeSelector:
-                  node.kubernetes.io/instance-type: gpu-h100-sxm
-                volumes:
-                  - name: devshm
-                    emptyDir:
-                      medium: Memory
-                  - name: hf-cache
-                    hostPath:
-                      path: /mnt/hf-cache
-                      type: DirectoryOrCreate
-
-      - label: ":full_moon: Omni · Perf Test"
-        key: nightly-omni-performance
-        timeout_in_minutes: 180
-        commands:
-          - export BENCHMARK_DIR=tests/dfx/perf/results
-          - |
-            set +e
-            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen_omni.json
-            EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
-            exit $$EXIT
-        agents:
-          queue: "mithril-h100-pool"
-        plugins:
-          - kubernetes:
-              podSpec:
-                containers:
-                  - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                    resources:
-                      limits:
-                        nvidia.com/gpu: 2
-                    volumeMounts:
-                      - name: devshm
-                        mountPath: /dev/shm
-                      - name: hf-cache
-                        mountPath: /root/.cache/huggingface
-                    env:
-                      - name: HF_HOME
-                        value: /root/.cache/huggingface
-                      - name: HF_TOKEN
-                        valueFrom:
-                          secretKeyRef:
-                            name: hf-token-secret
-                            key: token
-                nodeSelector:
-                  node.kubernetes.io/instance-type: gpu-h100-sxm
-                volumes:
-                  - name: devshm
-                    emptyDir:
-                      medium: Memory
-                  - name: hf-cache
-                    hostPath:
-                      path: /mnt/hf-cache
-                      type: DirectoryOrCreate
-
-
-  - group: ":card_index_dividers: TTS Model Test"
-    key: nightly-tts-test-group
-    depends_on: upload-nightly-pipeline
-    if: build.env("NIGHTLY") == "1" || build.pull_request.labels includes "nightly-test" || build.pull_request.labels includes "tts-test"
-    steps:
-      - label: ":full_moon: TTS · Function Test"
-        timeout_in_minutes: 90
-        commands:
-          - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
-          - pytest -s -v tests/e2e/online_serving/test_*_expansion.py -m "advanced_model and L4 and omni" --run-level "advanced_model"
-        agents:
-          queue: "gpu_1_queue" # g6.12xlarge instance on AWS, has 4 L4 GPU
-        plugins:
-          - docker#v5.2.0:
-              image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-              always-pull: true
-              propagate-environment: true
-              shm-size: "8gb"
-              environment:
-                - "HF_HOME=/fsx/hf_cache"
-                - "HF_TOKEN"
-              volumes:
-                - "/fsx/hf_cache:/fsx/hf_cache"
-
-      - label: ":full_moon: TTS · Perf Test"
-        key: nightly-tts-performance
-        timeout_in_minutes: 180
-        commands:
-          - export BENCHMARK_DIR=tests/dfx/perf/results
-          - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
-          - |
-            set +e
-            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_tts.json
-            EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
-            exit $$EXIT
-        agents:
-          queue: "mithril-h100-pool"
-        plugins:
-          - kubernetes:
-              podSpec:
-                containers:
-                  - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                    resources:
-                      limits:
-                        nvidia.com/gpu: 1
-                    volumeMounts:
-                      - name: devshm
-                        mountPath: /dev/shm
-                      - name: hf-cache
-                        mountPath: /root/.cache/huggingface
-                    env:
-                      - name: HF_HOME
-                        value: /root/.cache/huggingface
-                      - name: HF_TOKEN
-                        valueFrom:
-                          secretKeyRef:
-                            name: hf-token-secret
-                            key: token
-                nodeSelector:
-                  node.kubernetes.io/instance-type: gpu-h100-sxm
-                volumes:
-                  - name: devshm
-                    emptyDir:
-                      medium: Memory
-                  - name: hf-cache
-                    hostPath:
-                      path: /mnt/hf-cache
-                      type: DirectoryOrCreate
-
-  # Diffusion X2I suite: x2i / x2a / x2t and related non-video paths; x2v is only in "Diffusion X2V Model Test" below.
-  - group: ":card_index_dividers: Diffusion X2I(&A&T) Model Test"
-    key: nightly-diffusion-x2iat-group
-    depends_on: upload-nightly-pipeline
-    if: >-
-      build.env("NIGHTLY") == "1" ||
-      build.pull_request.labels includes "nightly-test" ||
-      build.pull_request.labels includes "diffusion-x2iat-test"
-    steps:
-      - label: ":full_moon: Diffusion X2I(&A&T) · Function Test with H100"
-        timeout_in_minutes: 120
-        commands:
-          - pytest -s -v tests/e2e/online_serving/test_*_expansion.py -k "not test_wan22_expansion and not test_wan_2_1_vace_expansion and not hunyuan" -m "advanced_model and diffusion and H100" --run-level "advanced_model"
-        agents:
-          queue: "mithril-h100-pool"
-        plugins:
-          - kubernetes:
-              podSpec:
-                containers:
-                  - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                    resources:
-                      limits:
-                        nvidia.com/gpu: 2
-                    volumeMounts:
-                      - name: devshm
-                        mountPath: /dev/shm
-                      - name: hf-cache
-                        mountPath: /root/.cache/huggingface
-                    env:
-                      - name: HF_HOME
-                        value: /root/.cache/huggingface
-                      - name: HF_TOKEN
-                        valueFrom:
-                          secretKeyRef:
-                            name: hf-token-secret
-                            key: token
-                nodeSelector:
-                  node.kubernetes.io/instance-type: gpu-h100-sxm
-                volumes:
-                  - name: devshm
-                    emptyDir:
-                      medium: Memory
-                  - name: hf-cache
-                    hostPath:
-                      path: /mnt/hf-cache
-                      type: DirectoryOrCreate
-
-      - label: ":full_moon: Diffusion X2I(&A&T) · Function Test with L4"
-        timeout_in_minutes: 60
-        commands:
-          - pytest -s -v tests/e2e/online_serving/test_*_expansion.py -k "not test_wan22_expansion and not test_wan_2_1_vace_expansion and not hunyuan" -m "advanced_model and diffusion and L4" --run-level "advanced_model"
-        agents:
-          queue: "gpu_4_queue" # g6.12xlarge instance on AWS, has 4 L4 GPU
-        plugins:
-          - docker#v5.2.0:
-              image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-              always-pull: true
-              propagate-environment: true
-              shm-size: "8gb"
-              environment:
-                - "HF_HOME=/fsx/hf_cache"
-                - "HF_TOKEN"
-              volumes:
-                - "/fsx/hf_cache:/fsx/hf_cache"
-
-      - label: ":full_moon: Diffusion X2I(&A&T) · Doc Test"
-        timeout_in_minutes: 60
-        commands:
-          - export VLLM_TEST_CLEAN_GPU_MEMORY="1"
-          - pytest -s -v tests/examples/*/test_text_to_image.py -m "advanced_model and example and H100" --run-level "advanced_model"
-        agents:
-          queue: "mithril-h100-pool"
-        plugins:
-          - kubernetes:
-              podSpec:
-                containers:
-                  - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                    resources:
-                      limits:
-                        nvidia.com/gpu: 2
-                    volumeMounts:
-                      - name: devshm
-                        mountPath: /dev/shm
-                      - name: hf-cache
-                        mountPath: /root/.cache/huggingface
-                    env:
-                      - name: HF_HOME
-                        value: /root/.cache/huggingface
-                      - name: HF_TOKEN
-                        valueFrom:
-                          secretKeyRef:
-                            name: hf-token-secret
-                            key: token
-                nodeSelector:
-                  node.kubernetes.io/instance-type: gpu-h100-sxm
-                volumes:
-                  - name: devshm
-                    emptyDir:
-                      medium: Memory
-                  - name: hf-cache
-                    hostPath:
-                      path: /mnt/hf-cache
-                      type: DirectoryOrCreate
-
-      - label: ":full_moon: Diffusion X2I(&A&T) · GEBench Accuracy Test"
-        timeout_in_minutes: 60
-        commands:
-          - pytest -s -v tests/e2e/accuracy/test_gebench_h100_smoke.py --run-level advanced_model --gebench-model Qwen/Qwen-Image-2512 --accuracy-judge-model QuantTrio/Qwen3-VL-30B-A3B-Instruct-AWQ --accuracy-gpu 0 --gebench-port 8093 --accuracy-workers 1
-          - buildkite-agent artifact upload "tests/e2e/accuracy/artifacts/gebench_qwen-image-2512/summary*.json"
-        agents:
-          queue: "mithril-h100-pool"
-        plugins:
-          - kubernetes:
-              podSpec:
-                containers:
-                  - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                    resources:
-                      limits:
-                        nvidia.com/gpu: 1
-                    volumeMounts:
-                      - name: devshm
-                        mountPath: /dev/shm
-                      - name: hf-cache
-                        mountPath: /root/.cache/huggingface
-                    env:
-                      - name: HF_HOME
-                        value: /root/.cache/huggingface
-                      - name: HF_TOKEN
-                        valueFrom:
-                          secretKeyRef:
-                            name: hf-token-secret
-                            key: token
-                nodeSelector:
-                  node.kubernetes.io/instance-type: gpu-h100-sxm
-                volumes:
-                  - name: devshm
-                    emptyDir:
-                      medium: Memory
-                  - name: hf-cache
-                    hostPath:
-                      path: /mnt/hf-cache
-                      type: DirectoryOrCreate
-
-      - label: ":full_moon: Diffusion X2I(&A&T) · GEdit-Bench Accuracy Test"
-        timeout_in_minutes: 60
-        commands:
-          - pytest -s -v tests/e2e/accuracy/test_gedit_bench_h100_smoke.py --run-level advanced_model --gedit-model Qwen/Qwen-Image-Edit --accuracy-judge-model QuantTrio/Qwen3-VL-30B-A3B-Instruct-AWQ --accuracy-gpu 0 --gedit-port 8093 --gedit-samples-per-group 20 --accuracy-workers 1
-          - buildkite-agent artifact upload "tests/e2e/accuracy/artifacts/gedit_scores_qwen-image-edit/qwen-image-edit_all_all_vie_score_*.csv"
-          - buildkite-agent artifact upload "tests/e2e/accuracy/artifacts/gedit_scores_qwen-image-edit/qwen-image-edit_all_all_summary_*.json"
-        agents:
-          queue: "mithril-h100-pool"
-        plugins:
-          - kubernetes:
-              podSpec:
-                containers:
-                  - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                    resources:
-                      limits:
-                        nvidia.com/gpu: 1
-                    volumeMounts:
-                      - name: devshm
-                        mountPath: /dev/shm
-                      - name: hf-cache
-                        mountPath: /root/.cache/huggingface
-                    env:
-                      - name: HF_HOME
-                        value: /root/.cache/huggingface
-                      - name: VLLM_HTTP_TIMEOUT_KEEP_ALIVE
-                        value: "120"
-                      - name: HF_TOKEN
-                        valueFrom:
-                          secretKeyRef:
-                            name: hf-token-secret
-                            key: token
-                nodeSelector:
-                  node.kubernetes.io/instance-type: gpu-h100-sxm
-                volumes:
-                  - name: devshm
-                    emptyDir:
-                      medium: Memory
-                  - name: hf-cache
-                    hostPath:
-                      path: /mnt/hf-cache
-                      type: DirectoryOrCreate
-
-      - label: ":full_moon: Diffusion X2I(&A&T) · Accuracy Test"
-        timeout_in_minutes: 180
-        commands:
-          - pytest -s -v tests/e2e/accuracy/test_qwen_image*.py --run-level advanced_model
-        agents:
-          queue: "mithril-h100-pool"
-        plugins:
-          - kubernetes:
-              podSpec:
-                containers:
-                  - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                    resources:
-                      limits:
-                        nvidia.com/gpu: 1
-                    volumeMounts:
-                      - name: devshm
-                        mountPath: /dev/shm
-                      - name: hf-cache
-                        mountPath: /root/.cache/huggingface
-                    env:
-                      - name: HF_HOME
-                        value: /root/.cache/huggingface
-                      - name: VLLM_HTTP_TIMEOUT_KEEP_ALIVE
-                        value: "120"
-                      - name: HF_TOKEN
-                        valueFrom:
-                          secretKeyRef:
-                            name: hf-token-secret
-                            key: token
-                nodeSelector:
-                  node.kubernetes.io/instance-type: gpu-h100-sxm
-                volumes:
-                  - name: devshm
-                    emptyDir:
-                      medium: Memory
-                  - name: hf-cache
-                    hostPath:
-                      path: /mnt/hf-cache
-                      type: DirectoryOrCreate
-
-      - label: ":full_moon: Diffusion X2I(&A&T) · Perf Test"
-        key: nightly-diffusion-x2iat-performance
-        timeout_in_minutes: 180
-        commands:
-          - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
-          - export DIFFUSION_ATTENTION_BACKEND=FLASH_ATTN
-          - export CACHE_DIT_VERSION=1.3.0
-          # [HACK]: run upload in the same command block as pytest.
-          # Because `exit` aborts the entire commands list.
-          - |
-            set +e
-            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen_image_vllm_omni.json
-            EXIT1=$$?
-            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen_image_edit_vllm_omni.json
-            EXIT2=$$?
-            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen_image_edit_2509_vllm_omni.json
-            EXIT3=$$?
-            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen_image_layered_vllm_omni.json
-            EXIT4=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/diffusion_result_*.json"
-            buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
-            exit $$((EXIT1 | EXIT2 | EXIT3 | EXIT4))
-        agents:
-          queue: "mithril-h100-pool"
-        plugins:
-          - kubernetes:
-              podSpec:
-                containers:
-                  - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                    resources:
-                      limits:
-                        nvidia.com/gpu: 4
-                    volumeMounts:
-                      - name: devshm
-                        mountPath: /dev/shm
-                      - name: hf-cache
-                        mountPath: /root/.cache/huggingface
-                    env:
-                      - name: HF_HOME
-                        value: /root/.cache/huggingface
-                      - name: HF_TOKEN
-                        valueFrom:
-                          secretKeyRef:
-                            name: hf-token-secret
-                            key: token
-                nodeSelector:
-                  node.kubernetes.io/instance-type: gpu-h100-sxm
-                volumes:
-                  - name: devshm
-                    emptyDir:
-                      medium: Memory
-                  - name: hf-cache
-                    hostPath:
-                      path: /mnt/hf-cache
-                      type: DirectoryOrCreate
-
-  # Diffusion x2v only (Wan, HunyuanVideo, …). x2i/x2a/x2t live in the X2I group above, not here.
-  - group: ":card_index_dividers: Diffusion X2V Model Test"
-    key: nightly-diffusion-x2v-group
-    depends_on: upload-nightly-pipeline
-    if: >-
-      build.env("NIGHTLY") == "1" ||
-      build.pull_request.labels includes "nightly-test" ||
-      build.pull_request.labels includes "diffusion-x2v-test"
-    steps:
-      - label: ":full_moon: Diffusion X2V · Function Test"
-        timeout_in_minutes: 90
-        commands:
-          - pytest -s -v tests/e2e/online_serving/test_wan22_expansion.py tests/e2e/online_serving/test_wan_2_1_vace_expansion.py tests/e2e/online_serving/test_hunyuan_video_15_expansion.py -m "advanced_model" --run-level "advanced_model"
-        agents:
-          queue: "mithril-h100-pool"
-        plugins:
-          - kubernetes:
-              podSpec:
-                containers:
-                  - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                    resources:
-                      limits:
-                        nvidia.com/gpu: 2
-                    volumeMounts:
-                      - name: devshm
-                        mountPath: /dev/shm
-                      - name: hf-cache
-                        mountPath: /root/.cache/huggingface
-                    env:
-                      - name: HF_HOME
-                        value: /root/.cache/huggingface
-                      - name: HF_TOKEN
-                        valueFrom:
-                          secretKeyRef:
-                            name: hf-token-secret
-                            key: token
-                nodeSelector:
-                  node.kubernetes.io/instance-type: gpu-h100-sxm
-                volumes:
-                  - name: devshm
-                    emptyDir:
-                      medium: Memory
-                  - name: hf-cache
-                    hostPath:
-                      path: /mnt/hf-cache
-                      type: DirectoryOrCreate
-
-      - label: ":full_moon: Diffusion X2V · Accuracy Test"
-        timeout_in_minutes: 180
-        commands:
-          - pytest -s -v tests/e2e/accuracy/wan22_i2v/test_wan22_i2v_video_similarity.py -m advanced_model --run-level advanced_model
-        agents:
-          queue: "mithril-h100-pool"
-        plugins:
-          - kubernetes:
-              podSpec:
-                containers:
-                  - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                    resources:
-                      limits:
-                        nvidia.com/gpu: 2
-                    volumeMounts:
-                      - name: devshm
-                        mountPath: /dev/shm
-                      - name: hf-cache
-                        mountPath: /root/.cache/huggingface
-                    env:
-                      - name: HF_HOME
-                        value: /root/.cache/huggingface
-                      - name: HF_TOKEN
-                        valueFrom:
-                          secretKeyRef:
-                            name: hf-token-secret
-                            key: token
-                nodeSelector:
-                  node.kubernetes.io/instance-type: gpu-h100-sxm
-                volumes:
-                  - name: devshm
-                    emptyDir:
-                      medium: Memory
-                  - name: hf-cache
-                    hostPath:
-                      path: /mnt/hf-cache
-                      type: DirectoryOrCreate
-
-  - label: ":bar_chart: Testcase Statistics"
-    key: nightly-testcase-statistics
-    timeout_in_minutes: 120
-    depends_on: upload-nightly-pipeline
-    if: build.env("NIGHTLY") == "1" || build.pull_request.labels includes "nightly-test"
-    commands:
-      - python tools/nightly/buildkite_testcase_statistics.py -o tests/dfx/perf/results/buildkite_testcase_statistics.html
-      - buildkite-agent artifact upload "tests/dfx/perf/results/*.html"
-    agents:
-      queue: "mithril-h100-pool"
-    plugins:
-      - kubernetes:
-          podSpec:
-            containers:
-              - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                resources:
-                  limits:
-                    nvidia.com/gpu: 2
-                volumeMounts:
-                  - name: devshm
-                    mountPath: /dev/shm
-                  - name: hf-cache
-                    mountPath: /root/.cache/huggingface
-                env:
-                  - name: HF_HOME
-                    value: /root/.cache/huggingface
-                  - name: HF_TOKEN
-                    valueFrom:
-                      secretKeyRef:
-                        name: hf-token-secret
-                        key: token
-            nodeSelector:
-              node.kubernetes.io/instance-type: gpu-h100-sxm
-            volumes:
-              - name: devshm
-                emptyDir:
-                  medium: Memory
-              - name: hf-cache
-                hostPath:
-                  path: /mnt/hf-cache
-                  type: DirectoryOrCreate
-
-  # No need to run this step for PRs with label nightly-test
-  - label: ":email: Nightly Collection & Email"
-    key: nightly-perf-distribution
-    depends_on:
-      - nightly-omni-performance
-      - nightly-tts-performance
-      - nightly-diffusion-x2iat-performance
-      - nightly-testcase-statistics
-    if: build.env("NIGHTLY") == "1"
-    commands:
-      - pip install openpyxl
-      - export DEFAULT_INPUT_DIR=tests/dfx/perf/results
-      - export DEFAULT_OUTPUT_DIR=tests/dfx/perf/results
-      - buildkite-agent artifact download "tests/dfx/perf/results/*.json" . --step nightly-tts-performance
-      - buildkite-agent artifact download "tests/dfx/perf/results/*.json" . --step nightly-omni-performance
-      - buildkite-agent artifact download "tests/dfx/perf/results/*.json" . --step nightly-diffusion-x2iat-performance
-      - buildkite-agent artifact download "tests/dfx/perf/results/*.html" . --step nightly-testcase-statistics
-      - python tools/nightly/generate_nightly_perf_excel.py
-      - python tools/nightly/generate_nightly_perf_html.py
-      - python tools/nightly/send_nightly_email.py --report-file "tests/dfx/perf/results/*.xlsx, tests/dfx/perf/results/*.html"
-      - buildkite-agent artifact upload "tests/dfx/perf/results/*.xlsx"
-      - buildkite-agent artifact upload "tests/dfx/perf/results/*.html"
-    agents:
-      queue: "cpu_queue_premerge"

.buildkite/test-ready.yml

@@ -1,638 +0,0 @@
-env:
-  VLLM_WORKER_MULTIPROC_METHOD: spawn
-  HF_HUB_DOWNLOAD_TIMEOUT: 300
-  HF_HUB_ETAG_TIMEOUT: 60
-
-steps:
-  - label: "Simple Unit Test"
-    depends_on: upload-ready-pipeline
-    commands:
-      - "timeout 20m pytest -v -s -m 'core_model and cpu' --cov=vllm_omni --cov-branch --cov-report=term-missing --cov-report=html --cov-report=xml"
-    agents:
-      queue: "gpu_1_queue"
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "CUDA Unit Test with single card"
-    depends_on: upload-ready-pipeline
-    commands:
-      - timeout 10m pytest -v -s -m 'core_model and cuda and L4 and not distributed_cuda' --ignore=tests/e2e --ignore=tests/engine/test_async_omni_engine_abort.py --cov=vllm_omni --cov-branch --cov-report=term-missing --cov-report=html --cov-report=xml
-    agents:
-      queue: "gpu_1_queue"
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "CUDA Unit Test with multi cards"
-    depends_on: upload-ready-pipeline
-    commands:
-      - timeout 10m pytest -v -s -m 'core_model and cuda and L4 and distributed_cuda' --ignore=tests/e2e --cov=vllm_omni --cov-branch --cov-report=term-missing --cov-report=html --cov-report=xml
-    agents:
-      queue: "gpu_4_queue"
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Diffusion Model Test"
-    depends_on: upload-ready-pipeline
-    commands:
-      - timeout 30m pytest -s -v tests/e2e/offline_inference/test_t2i_model.py -m "core_model and diffusion" --run-level "core_model"
-    agents:
-      queue: "gpu_1_queue"
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Diffusion Batching Test"
-    depends_on: upload-ready-pipeline
-    commands:
-      - timeout 20m pytest -s -v tests/e2e/offline_inference/test_qwen_image_diffusion_batching.py -m "core_model and diffusion" --run-level "core_model"
-    agents:
-      queue: "gpu_1_queue" # g6.4xlarge instance on AWS, has 1 L4 GPU
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Custom Pipeline Test"
-    depends_on: upload-ready-pipeline
-    commands:
-      - timeout 20m pytest -s -v tests/e2e/offline_inference/custom_pipeline/ -m "core_model"
-    agents:
-      queue: "gpu_1_queue"
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Diffusion Model CPU offloading Test"
-    depends_on: upload-ready-pipeline
-    commands:
-      - timeout 10m pytest -s -v tests/e2e/offline_inference/test_diffusion_cpu_offload.py
-    agents:
-      queue: "gpu_1_queue" # g6.4xlarge instance on AWS, has 1 L4 GPU
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Audio Generation Model Test"
-    depends_on: upload-ready-pipeline
-    commands:
-      - timeout 20m pytest -s -v tests/e2e/offline_inference/test_stable_audio_expansion.py -m "advanced_model and diffusion and L4" --run-level advanced_model
-    agents:
-      queue: "gpu_1_queue" # g6.4xlarge instance on AWS, has 1 L4 GPU
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Diffusion Cache Backend Test"
-    depends_on: upload-ready-pipeline
-    commands:
-      - timeout 15m pytest -s -v -m 'core_model and cache and diffusion and not distributed_cuda and L4'
-    agents:
-      queue: "gpu_1_queue" # g6.4xlarge instance on AWS, has 1 L4 GPU
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Diffusion Sequence Parallelism Test"
-    depends_on: upload-ready-pipeline
-    commands:
-      - timeout 20m pytest -s -v tests/e2e/offline_inference/test_sequence_parallel.py -m core_model
-    agents:
-      queue: "gpu_4_queue" # g6.12xlarge instance on AWS, has 4 L4 GPU
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          shm-size: "8gb"
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-
-  - label: "Engine Test"
-    depends_on: upload-ready-pipeline
-    commands:
-      - |
-        timeout 15m bash -c '
-                pytest -s -v tests/engine/test_async_omni_engine_abort.py
-        '
-    agents:
-      queue: "gpu_1_queue" # g6.12xlarge instance on AWS, has 4 L4 GPU
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          shm-size: "8gb"
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Omni Model Test with H100"
-    depends_on: upload-ready-pipeline
-    commands:
-      - |
-        timeout 20m bash -c '
-          pytest -s -v tests/e2e/online_serving/test_qwen3_omni.py -m "core_model" --run-level "core_model"
-        '
-    agents:
-      queue: "mithril-h100-pool"
-    plugins:
-      - kubernetes:
-          podSpec:
-            containers:
-              - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                resources:
-                  limits:
-                    nvidia.com/gpu: 2
-                volumeMounts:
-                  - name: devshm
-                    mountPath: /dev/shm
-                  - name: hf-cache
-                    mountPath: /root/.cache/huggingface
-                env:
-                  - name: HF_HOME
-                    value: /root/.cache/huggingface
-            nodeSelector:
-              node.kubernetes.io/instance-type: gpu-h100-sxm
-            volumes:
-              - name: devshm
-                emptyDir:
-                  medium: Memory
-              - name: hf-cache
-                hostPath:
-                  path: /mnt/hf-cache
-                  type: DirectoryOrCreate
-
-  - label: "MiMo-Audio E2E Test with H100"
-    depends_on: upload-ready-pipeline
-    commands:
-      - |
-        timeout 30m bash -c '
-          export VLLM_LOGGING_LEVEL=DEBUG
-          pytest -s -v tests/e2e/online_serving/test_mimo_audio.py -m "core_model" --run-level "core_model"
-        '
-    agents:
-      queue: "mithril-h100-pool"
-    plugins:
-      - kubernetes:
-          podSpec:
-            containers:
-              - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                resources:
-                  limits:
-                    nvidia.com/gpu: 1
-                volumeMounts:
-                  - name: devshm
-                    mountPath: /dev/shm
-                  - name: hf-cache
-                    mountPath: /root/.cache/huggingface
-                env:
-                  - name: HF_HOME
-                    value: /root/.cache/huggingface
-                  - name: HF_TOKEN
-                    valueFrom:
-                      secretKeyRef:
-                        name: hf-token-secret
-                        key: token
-            nodeSelector:
-              node.kubernetes.io/instance-type: gpu-h100-sxm
-            volumes:
-              - name: devshm
-                emptyDir:
-                  medium: Memory
-              - name: hf-cache
-                hostPath:
-                  path: /mnt/hf-cache
-                  type: DirectoryOrCreate
-
-  - label: "Qwen3-TTS E2E Test"
-    depends_on: upload-ready-pipeline
-    commands:
-      - |
-        timeout 20m bash -c '
-          export VLLM_LOGGING_LEVEL=DEBUG
-          export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
-          pytest -s -v tests/e2e/online_serving/test_qwen3_tts_customvoice.py -m "core_model" --run-level "core_model"
-        '
-    agents:
-      queue: "gpu_1_queue"
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          shm-size: "8gb"
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "VoxCPM E2E Test"
-    timeout_in_minutes: 20
-    depends_on: upload-ready-pipeline
-    commands:
-      - |
-        timeout 20m bash -c '
-          pip install voxcpm
-          export VLLM_LOGGING_LEVEL=DEBUG
-          export VLLM_WORKER_MULTIPROC_METHOD=spawn
-          pytest -s -v tests/e2e/offline_inference/test_voxcpm.py -m "core_model" --run-level "core_model"
-        '
-    agents:
-      queue: "gpu_1_queue"
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          shm-size: "8gb"
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "VoxCPM2 Native AR E2E Test"
-    timeout_in_minutes: 20
-    depends_on: upload-ready-pipeline
-    commands:
-      - |
-        timeout 20m bash -c '
-          pip install voxcpm
-          export VLLM_LOGGING_LEVEL=DEBUG
-          export VLLM_WORKER_MULTIPROC_METHOD=spawn
-          pytest -s -v tests/e2e/offline_inference/test_voxcpm2.py -m "core_model" --run-level "core_model"
-        '
-    agents:
-      queue: "gpu_1_queue"
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          shm-size: "8gb"
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "OmniVoice E2E Test"
-    timeout_in_minutes: 20
-    depends_on: upload-ready-pipeline
-    commands:
-      - |
-        timeout 20m bash -c '
-          export VLLM_LOGGING_LEVEL=DEBUG
-          pytest -s -v tests/e2e/online_serving/test_omnivoice.py -m "core_model" --run-level "core_model"
-        '
-    agents:
-      queue: "gpu_1_queue"
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          shm-size: "8gb"
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Qwen3-TTS Base E2E Test (ModelRunner V2)"
-    depends_on: upload-ready-pipeline
-    soft_fail:
-      - exit_status: 1
-    commands:
-      - |
-        timeout 20m bash -c '
-          export VLLM_LOGGING_LEVEL=DEBUG
-          export VLLM_WORKER_MULTIPROC_METHOD=spawn
-          export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
-          export VLLM_OMNI_USE_V2_RUNNER="1"
-          pytest -s -v tests/e2e/online_serving/test_qwen3_tts_base.py -m "core_model" --run-level "core_model"
-        '
-    agents:
-      queue: "gpu_1_queue"
-    plugins:
-      - docker#v5.2.0:
-          image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-          always-pull: true
-          propagate-environment: true
-          shm-size: "8gb"
-          environment:
-            - "HF_HOME=/fsx/hf_cache"
-            - "HF_TOKEN"
-          volumes:
-            - "/fsx/hf_cache:/fsx/hf_cache"
-
-  - label: "Voxtral-TTS E2E Test"
-    timeout_in_minutes: 20
-    depends_on: upload-ready-pipeline
-    commands:
-      - |
-        timeout 20m bash -c '
-          export VLLM_LOGGING_LEVEL=DEBUG
-          pytest -s -v tests/e2e/online_serving/test_voxtral_tts.py -m "core_model" --run-level "core_model"
-        '
-    agents:
-      queue: "mithril-h100-pool"
-    plugins:
-      - kubernetes:
-          podSpec:
-            containers:
-              - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                resources:
-                  limits:
-                    nvidia.com/gpu: 1
-                volumeMounts:
-                  - name: devshm
-                    mountPath: /dev/shm
-                  - name: hf-cache
-                    mountPath: /root/.cache/huggingface
-                env:
-                  - name: HF_HOME
-                    value: /root/.cache/huggingface
-            nodeSelector:
-              node.kubernetes.io/instance-type: gpu-h100-sxm
-            volumes:
-              - name: devshm
-                emptyDir:
-                  medium: Memory
-              - name: hf-cache
-                hostPath:
-                  path: /mnt/hf-cache
-                  type: DirectoryOrCreate
-
-  # - label: "Diffusion Image Edit Test with H100 (1 GPU)"
-  #   depends_on: upload-ready-pipeline
-  #   commands:
-  #     - |
-  #       timeout 20m bash -c '
-  #         pytest -s -v tests/e2e/online_serving/test_image_gen_edit.py
-  #       '
-  #   agents:
-  #     queue: "mithril-h100-pool"
-  #   plugins:
-  #     - kubernetes:
-  #         podSpec:
-  #           containers:
-  #             - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-  #               resources:
-  #                 limits:
-  #                   nvidia.com/gpu: 1
-  #               volumeMounts:
-  #                 - name: devshm
-  #                   mountPath: /dev/shm
-  #                 - name: hf-cache
-  #                   mountPath: /root/.cache/huggingface
-  #               env:
-  #                 - name: HF_HOME
-  #                   value: /root/.cache/huggingface
-  #           nodeSelector:
-  #             node.kubernetes.io/instance-type: gpu-h100-sxm
-  #           volumes:
-  #             - name: devshm
-  #               emptyDir:
-  #                 medium: Memory
-  #             - name: hf-cache
-  #               hostPath:
-  #                 path: /mnt/hf-cache
-  #                 type: DirectoryOrCreate
-
-  - label: "Bagel Text2Img Model Test with H100"
-    depends_on: upload-ready-pipeline
-    commands:
-      - |
-        timeout 30m bash -c '
-          export VLLM_TEST_CLEAN_GPU_MEMORY=1
-          pytest -s -v tests/e2e/offline_inference/test_bagel_text2img.py -m "core_model" --run-level "core_model"
-        '
-    agents:
-      queue: "mithril-h100-pool"
-    plugins:
-      - kubernetes:
-          podSpec:
-            containers:
-              - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                resources:
-                  limits:
-                    nvidia.com/gpu: 1
-                volumeMounts:
-                  - name: devshm
-                    mountPath: /dev/shm
-                  - name: hf-cache
-                    mountPath: /root/.cache/huggingface
-                env:
-                  - name: HF_HOME
-                    value: /root/.cache/huggingface
-                  - name: HF_TOKEN
-                    valueFrom:
-                      secretKeyRef:
-                        name: hf-token-secret
-                        key: token
-            nodeSelector:
-              node.kubernetes.io/instance-type: gpu-h100-sxm
-            volumes:
-              - name: devshm
-                emptyDir:
-                  medium: Memory
-              - name: hf-cache
-                hostPath:
-                  path: /mnt/hf-cache
-                  type: DirectoryOrCreate
-
-  - label: "Bagel Img2Img Model Test with H100"
-    depends_on: upload-ready-pipeline
-    commands:
-      - |
-        timeout 30m bash -c '
-          export VLLM_TEST_CLEAN_GPU_MEMORY=1
-          pytest -s -v tests/e2e/offline_inference/test_bagel_img2img.py -m "core_model" --run-level "core_model"
-        '
-    agents:
-      queue: "mithril-h100-pool"
-    plugins:
-      - kubernetes:
-          podSpec:
-            containers:
-              - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                resources:
-                  limits:
-                    nvidia.com/gpu: 1
-                volumeMounts:
-                  - name: devshm
-                    mountPath: /dev/shm
-                  - name: hf-cache
-                    mountPath: /root/.cache/huggingface
-                env:
-                  - name: HF_HOME
-                    value: /root/.cache/huggingface
-                  - name: HF_TOKEN
-                    valueFrom:
-                      secretKeyRef:
-                        name: hf-token-secret
-                        key: token
-            nodeSelector:
-              node.kubernetes.io/instance-type: gpu-h100-sxm
-            volumes:
-              - name: devshm
-                emptyDir:
-                  medium: Memory
-              - name: hf-cache
-                hostPath:
-                  path: /mnt/hf-cache
-                  type: DirectoryOrCreate
-
-  - label: "Bagel Online Serving Test with H100"
-    depends_on: upload-ready-pipeline
-    commands:
-      - |
-        timeout 40m bash -c '
-          export VLLM_TEST_CLEAN_GPU_MEMORY=1
-          export VLLM_IMAGE_FETCH_TIMEOUT=60
-          pytest -s -v tests/e2e/online_serving/test_bagel_online.py -m "core_model" --run-level "core_model"
-        '
-    agents:
-      queue: "mithril-h100-pool"
-    plugins:
-      - kubernetes:
-          podSpec:
-            containers:
-              - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                resources:
-                  limits:
-                    nvidia.com/gpu: 1
-                volumeMounts:
-                  - name: devshm
-                    mountPath: /dev/shm
-                  - name: hf-cache
-                    mountPath: /root/.cache/huggingface
-                env:
-                  - name: HF_HOME
-                    value: /root/.cache/huggingface
-                  - name: HF_TOKEN
-                    valueFrom:
-                      secretKeyRef:
-                        name: hf-token-secret
-                        key: token
-            nodeSelector:
-              node.kubernetes.io/instance-type: gpu-h100-sxm
-            volumes:
-              - name: devshm
-                emptyDir:
-                  medium: Memory
-              - name: hf-cache
-                hostPath:
-                  path: /mnt/hf-cache
-                  type: DirectoryOrCreate
-
-  - label: "CosyVoice3-TTS E2E Test"
-    timeout_in_minutes: 20
-    depends_on: upload-ready-pipeline
-    commands:
-      - |
-        timeout 20m bash -c '
-          pytest -s -v tests/e2e/online_serving/test_cosyvoice3_tts.py -m "core_model" --run-level "core_model"
-        '
-    agents:
-      queue: "mithril-h100-pool"
-    plugins:
-      - kubernetes:
-          podSpec:
-            containers:
-              - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                resources:
-                  limits:
-                    nvidia.com/gpu: 1
-                volumeMounts:
-                  - name: devshm
-                    mountPath: /dev/shm
-                  - name: hf-cache
-                    mountPath: /root/.cache/huggingface
-                env:
-                  - name: HF_HOME
-                    value: /root/.cache/huggingface
-                  - name: HF_TOKEN
-                    valueFrom:
-                      secretKeyRef:
-                        name: hf-token-secret
-                        key: token
-            nodeSelector:
-              node.kubernetes.io/instance-type: gpu-h100-sxm
-            volumes:
-              - name: devshm
-                emptyDir:
-                  medium: Memory
-              - name: hf-cache
-                hostPath:
-                  path: /mnt/hf-cache
-                  type: DirectoryOrCreate

.buildkite/test-template-amd-omni.j2

@@ -1,62 +0,0 @@
-{# vllm-omni customized version
-   Based on: https://github.com/vllm-project/ci-infra/blob/main/buildkite/test-template-amd.j2
-   Last synced: 2025-12-15
-   Modifications: Removed unused CUDA/NVIDIA logic, keeping only AMD tests
-#}
-{% set docker_image_amd = "rocm/vllm-omni:$BUILDKITE_COMMIT" %}
-{% set default_working_dir = "/app/vllm-omni" %}
-
-  - group: "AMD Tests"
-    depends_on: ~
-    steps:
-      - label: "AMD: :docker: build image"
-        depends_on: ~
-        soft_fail: false
-        commands:
-          - "docker build -f docker/Dockerfile.rocm -t {{ docker_image_amd }} --target test --build-arg ARG_PYTORCH_ROCM_ARCH=gfx942 --progress plain ."
-          - "docker push {{ docker_image_amd }}"
-        key: "amd-build"
-        env:
-          DOCKER_BUILDKIT: "1"
-        retry:
-          automatic:
-            - exit_status: -1  # Agent was lost
-              limit: 2
-            - exit_status: -10  # Agent was lost
-              limit: 2
-            - exit_status: 128  # Git connectivity issues
-              limit: 2
-            - exit_status: 1  # Machine occasionally fail
-              limit: 1
-        agents:
-          queue: amd-cpu
-
-    {% for step in steps %}
-    {% if step.mirror_hardwares and mirror_hw in step.mirror_hardwares %}
-      - label: "{{ step.agent_pool }}: {{ step.label }}"
-        depends_on: amd-build
-        agents:
-          {% if step.agent_pool %}
-          queue: amd_{{ step.agent_pool }}
-          {% else %}
-          queue: amd_mi325_1
-          {% endif %}
-{% set cmd_body = (step.command or (step.commands | join("\n"))) | trim %}
-{% set indented_cmd = cmd_body | replace("\n", "\n            ") %}
-        command: bash .buildkite/scripts/hardware_ci/run-amd-test.sh
-        env:
-          DOCKER_BUILDKIT: "1"
-          TEST_COMMAND: |-
-            (command rocm-smi || true) && cd {{ (step.working_dir or default_working_dir) | safe }}
-{% if "mi250" in step.agent_pool %}
-            python3 -m pip uninstall -y amd-aiter
-{% endif %}
-            {{ indented_cmd | safe }}
-        priority: 100
-        {% if step.grade and step.grade == "Blocking" %}
-        soft_fail: false
-        {% else %}
-        soft_fail: true
-        {% endif%}
-    {% endif %}
-    {% endfor %}

.claude/skills/add-diffusion-model/SKILL.md

@@ -24,7 +24,7 @@ Check the model's HF repo for `model_index.json`. This determines your path:
 
 | Scenario | How to identify | Migration path |
 |----------|----------------|----------------|
-| **Already supported** | `_class_name` in `model_index.json` matches a key in `_DIFFUSION_MODELS` in `registry.py` | Skip to Step 5 (test) and Step 7 (docs) |
+| **Already supported** | `_class_name` in `model_index.json` matches a key in `_DIFFUSION_MODELS` in `registry.py` | Skip implementation, then validate model-specific examples, tests, and docs as needed |
 | **Diffusers-based** | Has standard `model_index.json` with `_diffusers_version`, subfolders for `transformer/`, `vae/`, etc. | Follow **Path A** below |
 | **Custom/private repo** | No diffusers `model_index.json`, weights in non-standard format, custom model code in a separate git repo | Follow **Path B** below |
 | **Hybrid** | Has some diffusers components (VAE) but custom transformer/fusion | Mix of Path A and Path B |
@@ -52,7 +52,7 @@ vllm_omni/diffusion/models/your_model_name/
 2. Replace attention with `vllm_omni.diffusion.attention.layer.Attention` (QKV shape: `[B, seq, heads, head_dim]`).
 3. Add `od_config: OmniDiffusionConfig | None = None` to `__init__`.
 4. Add `load_weights()` method mapping diffusers weight names to vllm-omni names.
-5. Add class attributes: `_repeated_blocks`, `_layerwise_offload_blocks_attr`.
+5. Add class attributes for acceleration features such as `_repeated_blocks` and `_layerwise_offload_blocks_attrs` (see [references/transformer-adaptation.md](references/transformer-adaptation.md) for examples).
 
 ### A4. Adapt pipeline
 
@@ -78,13 +78,34 @@ class YourPipeline(nn.Module):
 
 Add post/pre-process functions in the same pipeline file. Register them in `registry.py`.
 
+### A4.1 Add progress bar support (recommended)
+
+For pipelines with a standard denoising loop, prefer the existing progress bar pattern instead of hand-rolled logging.
+
+```python
+from vllm_omni.diffusion.models.progress_bar import ProgressBarMixin
+
+class YourPipeline(nn.Module, ProgressBarMixin):
+    def forward(self, req: OmniDiffusionRequest) -> DiffusionOutput:
+        # ... prepare timesteps / latents ...
+        with self.progress_bar(total=len(timesteps)) as progress_bar:
+            for i, t in enumerate(timesteps):
+                # predict noise / scheduler step
+                latents = ...
+                progress_bar.update()
+
+        return DiffusionOutput(output=output)
+```
+
+For custom loop structures, follow `vllm_omni/diffusion/models/progress_bar.py` and existing pipelines using `ProgressBarMixin`.
+
 ### A5. Register, test, docs → continue at Step 4 below.
 
 ---
 
 ## Path B: Custom/Private Repo Model
 
-For models without a diffusers pipeline — weights in custom format, model code in a private repo. Real examples: DreamID-Omni, BAGEL, HunyuanImage3.
+For models without a diffusers pipeline — weights in custom format, model code in a private repo. Real examples: DreamID-Omni, BAGEL, HunyuanImage3. See [references/custom-model-patterns.md](references/custom-model-patterns.md) for concrete custom-model integration patterns.
 
 ### B1. Understand the reference repo
 
@@ -134,7 +155,7 @@ Copy the essential model files into `vllm_omni/diffusion/models/<name>/` and ada
 
 ### B4. Handle custom weight loading
 
-Custom models have two patterns for weight loading:
+Custom models have two common patterns for weight loading:
 
 **Pattern 1: Bypass standard loader** (DreamID-Omni style)
 
@@ -149,8 +170,10 @@ class CustomPipeline(nn.Module):
         self.vae = custom_init_vae(model, device=self.device)
         self.text_encoder = custom_init_text_encoder(model, device=self.device)
         self.transformer = CustomFusionModel(CONFIG)
-        load_custom_checkpoint(self.transformer,
-            checkpoint_path=os.path.join(model, "model.safetensors"))
+        load_custom_checkpoint(
+            self.transformer,
+            checkpoint_path=os.path.join(model, "model.safetensors"),
+        )
         # NO weights_sources defined — bypasses standard loader
 
     def load_weights(self, weights):
@@ -162,6 +185,8 @@ class CustomPipeline(nn.Module):
 When weights are in safetensors format but need name remapping:
 
 ```python
+from vllm.model_executor.model_loader.weight_utils import default_weight_loader
+
 class CustomPipeline(nn.Module):
     def __init__(self, *, od_config, prefix=""):
         super().__init__()
@@ -187,7 +212,9 @@ class CustomPipeline(nn.Module):
             # Remap original weight names to vllm-omni module names
             name = self._remap_weight_name(name)
             if name in params:
-                default_weight_loader(params[name], tensor)
+                param = params[name]
+                weight_loader = getattr(param, "weight_loader", default_weight_loader)
+                weight_loader(param, tensor)
                 loaded.add(name)
         return loaded
 ```
@@ -281,15 +308,43 @@ For Omni or custom models, create:
 ### Step 7: Update Documentation
 
 Required updates:
-1. `docs/user_guide/diffusion/parallelism_acceleration.md` — parallelism support table
-2. `docs/user_guide/diffusion/teacache.md` — if TeaCache supported
-3. `docs/user_guide/diffusion/cache_dit_acceleration.md` — if Cache-DiT supported
-4. `examples/offline_inference/xxx/README.md` — offline example docs
-5. `examples/online_serve/xxx/README.md` — online serve docs
+1. `docs/user_guide/diffusion/parallelism/overview.md` — parallelism support overview/table
+2. `docs/user_guide/diffusion/cpu_offload_diffusion.md` — if CPU offload supported (add to supported models table)
+3. `docs/user_guide/diffusion/cache_acceleration/teacache.md` — if TeaCache supported
+4. `docs/user_guide/diffusion/cache_acceleration/cache_dit.md` — if Cache-DiT supported
+5. Offline example docs under `examples/offline_inference/<name>/` (`README.md` or category-specific `.md`)
+6. `examples/online_serving/<name>/README.md` — online serving docs
 
-### Step 8: Add E2E Tests (Recommended)
+### Step 8: Add E2E Tests
 
-Create `tests/e2e/online_serving/test_your_model_expansion.py`.
+**Follow the [vllm-omni-test skill](../vllm-omni-test/SKILL.md)** for markers, file naming, Buildkite wiring, and run commands. Also read [l4_functionality_tests.inc.md](https://github.com/vllm-project/vllm-omni/blob/main/docs/contributing/ci/test_examples/l4_functionality_tests.inc.md), [test_system_overview.md](https://github.com/vllm-project/vllm-omni/blob/main/docs/contributing/ci/test_system_overview.md), and [test_writing_guide.md](https://github.com/vllm-project/vllm-omni/blob/main/docs/contributing/ci/test_writing_guide.md).
+
+Classify the model's **CI priority** first:
+
+| Priority | Required test levels | Files & markers |
+|----------|---------------------|-----------------|
+| **High** (listed in [#1832](https://github.com/vllm-project/vllm-omni/issues/1832) or on the diffusion hot path) | **L1** · **L2** online · **L3** online + offline · **L4** feature + performance | See table below |
+| **Medium** (*normal priority* in L4 docs) | **L3** online + offline · **L4** feature only | Fewer L4 parametrized rows |
+| **Low** | **L4** feature only | One or two `*_expansion.py` cases |
+
+**Per-level deliverables (diffusion / `pytest.mark.diffusion`):**
+
+| Level | Location | Marker | CI pipeline | Notes |
+|-------|----------|--------|-------------|-------|
+| **L1** | `tests/diffusion/models/{slug}/`, `tests/diffusion/cache/`, transformer unit tests | `core_model` + `cpu` | `test-ready.yml` | Weight remap, `_sp_plan`, cache enabler registration, shape contracts |
+| **L2** | `tests/e2e/online_serving/test_{slug}.py` (and offline if the category is offline-first) | **`core_model` + `advanced_model`** (both on baseline smoke) + `diffusion` + `@hardware_test` / `hardware_marks` | `test-ready.yml` | Default deploy smoke — minimal `num_inference_steps`, single prompt |
+| **L3** | `tests/e2e/online_serving/test_{slug}.py` **and** `tests/e2e/offline_inference/test_{slug}.py` when offline matters | Baseline smoke: **`core_model` + `advanced_model`**; heavier cases: `advanced_model` only (+ `diffusion`) | `test-merge.yml` **or** merged into nightly diffusion function job | Real weights, streaming/API paths, LoRA/offload smoke |
+| **L4** | `tests/e2e/online_serving/test_{slug}_expansion.py` (+ offline expansion if needed) | `full_model` + `diffusion` | `test-nightly.yml` (X2I / X2V / X2A function groups) | Feature combos per [#1832](https://github.com/vllm-project/vllm-omni/issues/1832); perf → `tests/dfx/perf/tests/test_{model}_vllm_omni.json` with per-case **`mark`** (`hardware_marks` + `full_model` + `diffusion`) |
+
+**L2 & L3 online — same file, dual marks on the baseline smoke:** The **first / simplest** case in `test_{slug}.py` (default deploy, minimal steps, single prompt) should carry **both** `@pytest.mark.core_model` and `@pytest.mark.advanced_model` on the **same** function so L2 (`test-ready.yml`) and L3 (`test-merge.yml`) share one smoke test. Heavier deploy variants or API paths in the same file use **`advanced_model` only**. When L3 moves to nightly, migrate those heavier cases into `test_{slug}_expansion.py` with `full_model` and remove the dedicated `test-merge.yml` job (see `test_longcat_image_expansion.py`, `test_qwen_image_expansion.py`).
+
+**L4 design (high priority):** Combine multiple supported features (Cache-DiT, TP, USP, CFG, HSDP, CPU offload, quantization) into **few parametrized** `OmniServerParams` rows so each feature appears in at least one case without exploding GPU jobs. Shard single-GPU vs multi-GPU cases across the nightly X2I/X2V function steps (1-GPU vs `distributed_cuda`).
+
+**L4 design (medium / low):** One or two parametrized rows covering the best quality/perf trade-off; skip perf JSON unless the model is high priority.
+
+**Reference implementations:** `tests/e2e/online_serving/test_qwen_image_edit_expansion.py`, `tests/e2e/online_serving/test_longcat_image_expansion.py`, `tests/e2e/online_serving/test_hunyuan_video_15_expansion.py`.
+
+**Keep model-specific code inside test modules — not `tests/helpers/{slug}.py`:** deploy constants, prompts, sampling dicts, and inline `request_config` / `form_data` belong in each `test_{slug}.py` and `test_{slug}_expansion.py`. Do not add per-model files under `tests/helpers/`; reuse only repo-wide harness (`mark`, `media`, `runtime`, `stage_config`, `assertions`). **L2+ online/offline e2e:** reuse or add `send_*_request` in `tests/helpers/runtime.py` — tests call the handler, not raw `omni.generate` / HTTP. See [vllm-omni-test skill](../vllm-omni-test/SKILL.md) § **Runtime send helpers**.
 
 ### Step 9: Add Cache-DiT Acceleration
 
@@ -363,7 +418,8 @@ def enable_cache_for_your_model(pipeline, cache_config):
 
 #### 9d. Register the custom enabler
 
-Add your enabler to `CUSTOM_DIT_ENABLERS` in `vllm_omni/diffusion/cache/cache_dit_backend.py`:
+Add your enabler to `CUSTOM_DIT_ENABLERS` in
+`vllm_omni/diffusion/cache/cachedit/model_specific.py`:
 
 ```python
 CUSTOM_DIT_ENABLERS = {
@@ -509,9 +565,49 @@ class YourTransformer(nn.Module):
 #### 10e. Update parallelism documentation
 
 After adding parallelism support, update:
-1. `docs/user_guide/diffusion/parallelism_acceleration.md` — add your model to the support table
+1. `docs/user_guide/diffusion/parallelism/overview.md` — add your model to the support overview/table
 2. Record which parallelism methods are supported (USP, Ring, CFG, TP, HSDP, VAE-Patch)
 
+### Step 11: Add CPU Offload Support
+
+Implement `SupportsComponentDiscovery` on your pipeline class to enable
+`--enable-cpu-offload` and `--enable-layerwise-offload`. The protocol
+declares which submodules the offloader should manage:
+
+```python
+from typing import ClassVar
+from vllm_omni.diffusion.models.interface import SupportsComponentDiscovery
+
+class YourPipeline(nn.Module, SupportsComponentDiscovery):
+    _dit_modules: ClassVar[list[str]] = ["transformer"]
+    _encoder_modules: ClassVar[list[str]] = ["text_encoder"]
+    _vae_modules: ClassVar[list[str]] = ["vae"]
+    _resident_modules: ClassVar[list[str]] = []  # optional
+```
+
+- `_dit_modules`: denoising submodules (kept on GPU during diffusion loop)
+- `_encoder_modules`: encoder/vision submodules (offloaded to CPU during diffusion loop)
+- `_vae_modules`: VAE(s) (handled by both sequential and layerwise backends)
+- `_resident_modules`: additional modules to pin on GPU during layerwise
+  offloading (e.g. embedders, connectors). Only used by the layerwise
+  backend. Optional — defaults to `[]`.
+
+All attribute names support dotted paths for nested submodules
+(e.g. `"pipe.transformer"`, `"bagel.time_embedder"`).
+
+Pipelines without `SupportsComponentDiscovery` fall back to scanning
+well-known attribute names (`transformer`, `text_encoder`, `vae`,
+etc.), which fails for non-standard names.
+
+### Step 12: Performance Profiling
+
+After verifying correctness and implementing parallelism/caching, profile the model's performance to identify bottlenecks and ensure optimal execution.
+
+See the [Profiling Single-Stage Diffusion](../../../docs/contributing/profiling.md#3-profiling-single-stage-diffusion) guide for detailed instructions on:
+1. Using the PyTorch profiler (`profiler: "torch"`) to capture detailed CPU/CUDA traces.
+2. Using Nsight Systems (`nsys`) with `profiler: "cuda"` for low-overhead CUDA traces.
+3. Controlling profiling via `omni.start_profile()` and `omni.stop_profile()`.
+
 ---
 
 ## Iterative Development Tips
@@ -527,6 +623,7 @@ After adding parallelism support, update:
 
 ## Reference Files
 
+- [vllm-omni-test skill](../vllm-omni-test/SKILL.md) — L1–L4 markers, naming, Buildkite wiring, run commands
 - [Transformer Adaptation](references/transformer-adaptation.md) — porting transformers from diffusers
 - [Custom Model Patterns](references/custom-model-patterns.md) — patterns for non-diffusers models
 - [Parallelism Patterns](references/parallelism-patterns.md) — TP, SP/USP, CFG parallel, HSDP implementation details

.claude/skills/add-diffusion-model/references/cache-dit-patterns.md

@@ -29,8 +29,13 @@ vLLM-Omni integrates cache-dit through `CacheDiTBackend`:
 | `cache_dit.refresh_context()` | Updates cache context when `num_inference_steps` changes |
 
 **Source files**:
-- `vllm_omni/diffusion/cache/cache_dit_backend.py` — `CacheDiTBackend`, enablers, `CUSTOM_DIT_ENABLERS`
-- `vllm_omni/diffusion/cache/` — cache backend implementations
+- `vllm_omni/diffusion/cache/cachedit/__init__.py` — public package API
+- `vllm_omni/diffusion/cache/cachedit/backend.py` — generic backend lifecycle and integration
+- `vllm_omni/diffusion/cache/cachedit/config.py` — Cache-DiT configuration types
+- `vllm_omni/diffusion/cache/cachedit/model_specific.py` — model-specific adapters, enablers, and registry
+
+Outside `vllm_omni/diffusion/cache`, import Cache-DiT symbols only from
+`vllm_omni.diffusion.cache.cachedit`.
 
 ## Standard Models: Automatic Support
 
@@ -175,7 +180,8 @@ Inspect your block's `forward()` return type and residual connection pattern to
 
 ## Registering Custom Enablers
 
-Add your enabler to `CUSTOM_DIT_ENABLERS` in `vllm_omni/diffusion/cache/cache_dit_backend.py`:
+Add your enabler to `CUSTOM_DIT_ENABLERS` in
+`vllm_omni/diffusion/cache/cachedit/model_specific.py`:
 
 ```python
 CUSTOM_DIT_ENABLERS = {
@@ -248,7 +254,7 @@ Models listed in `_NO_CACHE_ACCELERATION` in `vllm_omni/diffusion/registry.py` d
 
 | Model | Path | Notes |
 |-------|------|-------|
-| Standard DiT | `cache_dit_backend.py::enable_cache_for_dit` | Default enabler, automatic |
-| Wan2.2 | `cache_dit_backend.py::enable_cache_for_wan22` | Dual-transformer, auto-detects mode |
-| LongCat | `cache_dit_backend.py::enable_cache_for_longcat_image` | Multi-block-list |
-| BAGEL | `cache_dit_backend.py::enable_cache_for_bagel` | Complex omni model |
+| Standard DiT | `cachedit.backend::enable_cache_for_dit` | Default enabler, automatic |
+| Wan2.2 | `cachedit.model_specific::enable_cache_for_wan22` | Dual-transformer, auto-detects mode |
+| LongCat | `cachedit.config::CacheDiTAdapterConfig` | Declarative multi-block-list adapter |
+| BAGEL | `cachedit.model_specific::BagelCachedAdapter` | Custom cached adapter |

.claude/skills/add-diffusion-model/references/parallelism-patterns.md

@@ -92,6 +92,7 @@ class TPSelfAttention(nn.Module):
             softmax_scale=1.0 / (self.head_dim ** 0.5),
             causal=False,
             num_kv_heads=self.to_qkv.num_kv_heads,  # Local KV heads per GPU
+            role="self",
         )
 
     def forward(self, x):
@@ -568,4 +569,4 @@ Check which parallelism methods a model supports:
 | **TP** | Uses `ColumnParallelLinear` / `QKVParallelLinear`. Search: `grep -r 'ParallelLinear\|QKVParallel' vllm_omni/diffusion/models/<model>/` |
 | **HSDP** | Transformer defines `_hsdp_shard_conditions`. Search: `grep -r '_hsdp_shard_conditions' vllm_omni/diffusion/models/` |
 
-The canonical per-model support table is in `docs/user_guide/diffusion/parallelism_acceleration.md`.
+The canonical per-model support overview/table is in `docs/user_guide/diffusion/parallelism/overview.md`.

.claude/skills/add-diffusion-model/references/transformer-adaptation.md

@@ -61,6 +61,7 @@ class SelfAttentionBlock(nn.Module):
             softmax_scale=1.0 / (self.head_dim ** 0.5),
             causal=False,
             num_kv_heads=num_heads,
+            role="self",
         )
 
     def forward(self, x, attn_mask=None):
@@ -75,6 +76,24 @@ class SelfAttentionBlock(nn.Module):
         return self.to_out(out)
 ```
 
+**Declaring the attention role.** Always pass a `role` string to `Attention(...)`. It is the lookup key users target with `--diffusion-attention-config.per_role.<role>.backend` to swap kernels per site without touching model code.
+
+| Convention | When to use |
+|---|---|
+| `role="self"` | Q/K/V come from the same hidden state |
+| `role="cross"` | K/V come from a separate `encoder_hidden_states`; pair with `skip_sequence_parallel=True` if K/V is replicated across SP ranks |
+
+For multi-modal sites that don't fit `self` / `cross`, use a dot-namespaced role and provide `role_category` so user config can fall back to the generic category:
+
+```python
+self.audio_to_video_attn = Attention(
+    num_heads=num_heads, head_size=self.head_dim,
+    softmax_scale=1.0 / (self.head_dim ** 0.5), causal=False,
+    role="mymodel.audio_to_video",
+    role_category="cross",
+)
+```
+
 #### Fused QKV with TP (Advanced)
 
 For tensor parallelism, use vLLM's parallel linear layers:
@@ -104,6 +123,7 @@ class TPSelfAttention(nn.Module):
             softmax_scale=1.0 / (self.head_dim ** 0.5),
             causal=False,
             num_kv_heads=num_heads,
+            role="self",
         )
 ```
 
@@ -199,13 +219,14 @@ def load_weights(self, weights):
 
 ### Class-Level Attributes for Features
 
+
 ```python
 class MyTransformer(nn.Module):
     # torch.compile: list block class names that repeat and can be compiled
     _repeated_blocks = ["MyTransformerBlock"]
 
     # CPU offload: attribute name of the nn.ModuleList containing blocks
-    _layerwise_offload_blocks_attr = "blocks"
+    _layerwise_offload_blocks_attrs = ["blocks"]
 
     # LoRA: mapping of fused param names to original param names
     packed_modules_mapping = {"to_qkv": ["to_q", "to_k", "to_v"]}

.claude/skills/add-diffusion-model/references/troubleshooting.md

@@ -68,7 +68,7 @@ out = out.reshape(batch, seq_len, -1)
 ### Different Output vs Diffusers Reference
 
 **Common causes**:
-1. **Attention backend difference**: FlashAttention vs SDPA may produce slightly different results. Set `DIFFUSION_ATTENTION_BACKEND=TORCH_SDPA` to match diffusers
+1. **Attention backend difference**: FlashAttention vs SDPA may produce slightly different results. Force SDPA via `--diffusion-attention-backend TORCH_SDPA` (or `DIFFUSION_ATTENTION_BACKEND=TORCH_SDPA`) to match diffusers; for finer-grained per-role overrides see `--diffusion-attention-config.per_role.<role>.backend`
 2. **Float precision**: vLLM-Omni may use bfloat16 where diffusers uses float32 for some operations
 3. **Missing normalization**: Check all LayerNorm/RMSNorm are preserved
 4. **Scheduler rounding**: Some schedulers have numerical sensitivity

.claude/skills/add-tts-model/SKILL.md

@@ -8,10 +8,90 @@ description: "Integrate a new text-to-speech model into vLLM-Omni from HuggingFa
 ## Overview
 
 ```
-HF Reference -> Stage Separation -> Online Serving -> Async Chunk -> CUDA Graph
-   (Phase 1)      (Phase 2)          (Phase 3)        (Phase 4)     (Phase 5)
+HF Reference -> Stage Separation -> Online Serving -> Async Chunk -> CUDA Graph -> Pre-commit/DCO
+   (Phase 1)      (Phase 2)          (Phase 3)        (Phase 4)     (Phase 5)      (Phase 6)
 ```
 
+Three architecture patterns are supported:
+
+- **Two-stage pipeline** (e.g. Qwen3-TTS, Fish Speech, CosyVoice3): AR
+  code-predictor → audio decoder, connected via async_chunk for low-latency
+  streaming. Use this for maximum performance.
+- **Single-stage AR via generator** (e.g. MOSS-TTS-Nano): entire model runs
+  inside one AR worker, streaming audio chunks through a per-request
+  `inference_stream()` generator. Use this when the upstream model bundles AR
+  + codec inseparably. See [references/single-stage-ar.md](references/single-stage-ar.md).
+- **Single-stage, vLLM-native base LM + side computation** (e.g. VoxCPM2):
+  the base language model runs under vLLM's PagedAttention as a normal AR
+  model; diffusion / VAE / side computations run outside vLLM and are
+  attached via the runner post-processing hook. This is a distinct pattern
+  from the generator approach above — do not confuse the two.
+
+The single-stage variants skip Phase 4 (async_chunk) but Phase 5 (CUDA graph)
+is still encouraged for the inner AR loop.
+
+## Cross-Cutting Invariants
+
+These rules apply to every TTS model regardless of architecture (AR vs AR+diffusion, single-stage vs two-stage, codec-based vs VAE-based). They surface repeatedly across PRs — check them at the end of every phase.
+
+### I1. Streaming output contract
+
+Pick exactly one per-step semantics for `forward()` and document it in the docstring:
+
+- **Delta**: yield only new audio samples produced this step. Preferred — linear cost, low memory.
+- **Cumulative**: re-decode from step 0 every call. O(N²); only acceptable if the codec has no streaming decode path.
+
+If you choose **delta**, verify the full emit→consolidate→consume chain:
+
+1. `forward()` returns `{"model_outputs": <new_chunk_only>, ...}`
+2. `_consolidate_multimodal_tensors()` in `vllm_omni/engine/output_processor.py` concatenates the audio key into one tensor at finish. If it skips the key (`continue`), offline consumers receive only the final chunk. See `output_processor.py` for the concrete list of handled modality keys.
+3. Streaming consumers (SSE, Gradio) receive per-step deltas; offline consumers (`engine.generate()`) receive a single concatenated tensor.
+
+Cumulative-vs-delta mismatch is the most common silent bug — offline RTF benchmarks pass, but users hear replays or truncation.
+
+### I2. Multimodal output consumer hygiene
+
+`outputs[0].outputs[0].multimodal_output[<key>]` can be any of `Tensor`, `list[Tensor]` (pre-consolidation snapshot), `np.ndarray`, or scalar. When writing tests, examples, and benchmarks:
+
+- **Never** use `dict.get("a") or dict.get("b")` on tensor values — Python evaluates the tensor's boolean, raising `RuntimeError: Boolean value of Tensor with more than one value is ambiguous`. Use explicit `if x is None` chains.
+- Always defensively handle the list form: `if isinstance(x, list): x = torch.cat([t.reshape(-1) for t in x], dim=0)`.
+- Assert `shape` / `dtype` / `duration` explicitly; do not rely on truthiness for presence checks.
+
+### I3. Hot-loop GPU discipline
+
+Inside any per-step model loop (AR decode, diffusion solver, CFM Euler, vocoder block loop):
+
+- No `tensor.item()`, `.cpu()`, or `.tolist()` — each triggers a GPU→CPU sync; at 10 steps × 60 frames × 4 ops that is 2400 syncs per request.
+- Prefer `dst.copy_(src)` over `dst.fill_(src.item())` when writing a scalar tensor into a buffer.
+- Prefer `torch.compile(Model.forward, fullgraph=False)` on the whole forward over per-submodule compile — fewer dispatch boundaries, larger fusion regions. Measure before choosing granularity.
+- No Python-side control flow that depends on tensor values; use `torch.where` / masking instead.
+
+Profile first, optimize second. See the profiling docs / project memory for the trace-analysis workflow.
+
+### I4. Validation pyramid
+
+Offline RTF alone is necessary but not sufficient. Every new TTS model must pass all three:
+
+| Layer | Catches | Tool |
+|-------|---------|------|
+| Offline RTF / duration check | Throughput regressions, missing audio, wrong sample rate | `end2end.py`, pytest e2e |
+| Browser streaming playback | Delta/cumulative bugs, chunk boundary glitches, TTFP regressions | Gradio demo over `/v1/audio/speech?stream=true` |
+| Concurrent requests | Per-request state leaks, codec window round-robin gaps | `max_num_seqs>1` smoke test with 4+ parallel prompts |
+
+Declaring a model "done" without all three has shipped regressions more than once.
+
+### I5. Per-request state is owned by the request, not the model
+
+If the model caches *anything* across `forward()` calls (streaming generators, codec buffers, sliding-window pads, CUDA graph state), key it by request ID:
+
+```python
+self._state: dict[str, YourState] = {}    # request_key → state
+# fetch: request_key = str(info.get("_omni_req_id", "0"))
+# free on finish: del self._state[request_key]
+```
+
+A shared buffer silently corrupts audio across concurrent requests — the symptom is crosstalk or truncation only under load.
+
 ## Phase 1: HuggingFace Reference
 
 **Goal**: Understand the reference implementation and verify it produces correct audio.
@@ -85,12 +165,54 @@ When audio output is wrong, check in this order:
 6. **Tensor layout**: Codebook-major vs frame-major ordering
 7. **dtype**: Float32 for codec decoders (autocast can corrupt audio)
 
+### Streaming Correctness Rules (single-stage and two-stage)
+
+These bugs appear in almost every new TTS PR. Check all before the first push. See also the cross-cutting invariants I1 (output contract) and I5 (per-request state) above — the rules below are the Phase 2-specific instances of those invariants:
+
+- **Accumulate codes across AR steps** — each `forward()` appends new codes; do not reset between steps or audio will be truncated (fish speech: `fix: accumulate audio_codes across steps`)
+- **Emit delta audio, not full waveform** — in streaming mode yield only the new chunk per step, not the re-decoded full waveform from step 0 (fish speech: `fix: emit delta audio not full waveform`)
+- **All return paths must emit `model_outputs`** — if any early-return branch skips setting `model_outputs`, the serving layer silently drops that step's audio (fish speech: `fix: ensure ALL return paths emit model_outputs`)
+- **Per-request state isolation** — for batched concurrent requests, key all state by request ID; a shared buffer corrupts audio across requests (fish speech: `fix: per-request vocode + delta emission`)
+- **Codec tensor device** — move codec codes to the codec decoder's device before calling decode; mismatches cause silent CPU fallback or crashes (fish speech: `fix: use model device for CUDA stream`)
+- **AR stage `max_num_seqs`** — set to at least 4 in production configs; for single-stage models this is the only stage. For two-stage models, Stage 0 (AR) needs `max_num_seqs ≥ 4` to pipeline concurrent requests; Stage 1 (codec decoder) typically uses `max_num_seqs: 1` intentionally. Default of 1 everywhere causes audio gaps under concurrency because the codec window round-robins across requests (RFC #2568)
+
+### Optional Dependency Handling
+
+Patch optional dependencies (`torchaudio` / `torchcodec` / `soundfile`) at
+the top of `load_weights()`, not at module import. Failures to do so cause
+cryptic errors only on environments missing the optional package — after
+the model is already deployed. See
+[references/optional-deps.md](references/optional-deps.md) for the full
+pattern, signature constraints, and MOSS-TTS-Nano reference.
+
+### Single-Stage AR Pattern (alternative to two-stage)
+
+When the upstream model cannot be cleanly split into an AR stage and a
+separate decoder, run the full pipeline inside a single AR worker and
+stream audio through a per-request `inference_stream()` generator keyed by
+`_omni_req_id`. Stage config must set `worker_type: ar`,
+`engine_output_type: audio`, `final_output: true`, `is_comprehension: true`,
+and `async_chunk: false` at the top level. Only extract params from
+`additional_information` that you actually forward, or pre-commit fails
+`ruff F841`.
+
+Full walkthrough with the complete `forward()` / `_create_stream_gen()`
+skeleton and stage-config fields:
+[references/single-stage-ar.md](references/single-stage-ar.md). For an
+in-tree reference, look for any single-stage AR model under
+`vllm_omni/model_executor/models/` — e.g. the MOSS-TTS-Nano integration when
+it lands.
+
+**VoxCPM2 is a different pattern** and should not reuse this skeleton — it
+runs the base LM under vLLM PagedAttention with external side-computation.
+See `plan/voxcpm2_native_ar_design.md`.
+
 ### Deliverables
 
 - Model files in `vllm_omni/model_executor/models/<model_name>/`
 - Stage config YAML
-- Working `end2end.py` with correct audio output
-- README.md in the example directory
+- Working `end2end.py` at `examples/offline_inference/text_to_speech/<model>/end2end.py`
+- New section in `examples/offline_inference/text_to_speech/README.md` (table row + per-model section). Do **not** create a top-level `examples/offline_inference/<model>/` dir or a per-model `README.md` inside `text_to_speech/<model>/` — the hub README is the documented surface and the mkdocs `generate_examples` hook only descends one level into `examples/<category>/`.
 
 ## Phase 3: Online Serving
 
@@ -98,10 +220,66 @@ When audio output is wrong, check in this order:
 
 ### Steps
 
-1. **Register in `serving_speech.py`**:
-   - Add model stage name to `_TTS_MODEL_STAGES` set
-   - Add model detection flag (e.g., `_is_fish_speech`)
-   - Implement prompt builder method (e.g., `_build_fish_speech_prompt()`)
+1. **Register in `serving_speech.py`** — add all 5 points in a **single commit**;
+   partial integration causes hard-to-debug failures. This file is modified by every
+   model PR and is the most common source of rebase conflicts — see conflict note below.
+
+   **Point 1** — stage constant (near the top, alongside the other `_*_TTS_MODEL_STAGES` sets):
+   ```python
+   _YOUR_MODEL_TTS_MODEL_STAGES = {"your_stage_key"}
+   ```
+
+   **Point 2** — union into `_TTS_MODEL_STAGES`:
+   ```python
+   _TTS_MODEL_STAGES: set[str] = (
+       ...
+       | _YOUR_MODEL_TTS_MODEL_STAGES
+   )
+   ```
+
+   **Point 3** — model type detection in `_detect_tts_model_type()`:
+   ```python
+   if model_stage in _YOUR_MODEL_TTS_MODEL_STAGES:
+       return "your_model"
+   ```
+
+   **Point 4** — validation dispatch in `_validate_tts_request()`:
+   ```python
+   if self._tts_model_type == "your_model":
+       return self._validate_your_model_request(request)
+   ```
+
+   **Point 5** — validation + parameter-builder methods:
+   ```python
+   def _validate_your_model_request(self, request) -> str | None:
+       if not request.input or not request.input.strip():
+           return "Input text cannot be empty"
+       return None
+
+   def _build_your_model_params(self, request) -> dict:
+       params = {"text": [request.input]}
+       if request.voice is not None:
+           params["voice"] = [request.voice]
+       return params
+   ```
+   Wire `_build_your_model_params` into `_create_tts_request()` alongside the other
+   model-specific param builders.
+
+   > **Two dispatch patterns coexist**: Fish Speech uses a `self._is_fish_speech` boolean
+   > instance attribute checked before `elif self._is_tts`, while all newer models
+   > (CosyVoice3, MOSS-TTS-Nano) use the `_tts_model_type` string returned by
+   > `_detect_tts_model_type()`. For new models, always use the `_tts_model_type` string
+   > pattern — do not add new `_is_*` flags.
+
+   > **Unused variable rule**: only extract fields in `_build_your_model_params` that
+   > are actually forwarded to the model. Unused extractions fail `ruff F841`.
+   > For voice-cloning fields (`ref_audio` → `prompt_audio_path`, `ref_text` →
+   > `prompt_text`), add them to the param builder and verify they reach the model call.
+
+   **Rebase conflict note**: when rebasing onto `main` after another model was merged,
+   `serving_speech.py` will conflict. Resolution: always keep *both* the upstream
+   model's additions and your own — never discard either side.
+
 2. **Handle model-specific parameters**:
    - Voice cloning: `ref_audio` encoding and prompt injection
    - `max_new_tokens` override in sampling params
@@ -127,12 +305,89 @@ def build_voice_clone_prompt(ref_audio_path: str, text: str, codec) -> list:
     ]
 ```
 
+### Test Case Writing (CI Levels)
+
+**Follow the [vllm-omni-test skill](../vllm-omni-test/SKILL.md)** for markers, file naming (`test_{slug}.py` / `test_{slug}_expansion.py`), Buildkite wiring, and copy-paste run commands. Also read [test_system_overview.md](https://github.com/vllm-project/vllm-omni/blob/main/docs/contributing/ci/test_system_overview.md) and [test_writing_guide.md](https://github.com/vllm-project/vllm-omni/blob/main/docs/contributing/ci/test_writing_guide.md).
+
+Classify the model's **CI priority** first (high / medium / low). High-priority TTS models are typically those on the integration hot path or listed in tracking issues such as [#1832](https://github.com/vllm-project/vllm-omni/issues/1832); medium and low tiers cover the long tail. When unsure, ask the reviewer which tier applies.
+
+| Priority | Required test levels | Files & markers |
+|----------|---------------------|-----------------|
+| **High** | **L1** unit/logic · **L2** online smoke · **L3** online + offline integration · **L4** feature + performance | See table below |
+| **Medium** | **L3** online + offline · **L4** feature only | Skip dedicated L1/L2 unless fixing a logic bug |
+| **Low** | **L4** feature only | One or two `*_expansion.py` parametrized cases |
+
+**Per-level deliverables (TTS / `pytest.mark.tts`):**
+
+| Level | Location | Marker | CI pipeline | Notes |
+|-------|----------|--------|-------------|-------|
+| **L1** | `tests/model_executor/…`, `tests/entrypoints/openai_api/…`, stage-processor tests | `core_model` + `cpu` | `test-ready.yml` | Prompt assembly, async_chunk helpers, `serving_speech` validation — no GPU |
+| **L2** | `tests/e2e/online_serving/test_{slug}.py` | **`core_model` + `advanced_model`** (both on baseline smoke) + `tts` + `@hardware_test(...)` | `test-ready.yml` (`ready` label) | Default deploy smoke: single `/v1/audio/speech` or offline `OmniRunner` path |
+| **L3** | `tests/e2e/online_serving/test_{slug}.py` **and** `tests/e2e/offline_inference/test_{slug}.py` | Baseline smoke: **`core_model` + `advanced_model`**; heavier cases: `advanced_model` only (+ `tts`) | `test-merge.yml` **or** merged into nightly TTS function job | Streaming, voice clone, batch/queue, async_chunk |
+| **L4** | `tests/e2e/online_serving/test_{slug}_expansion.py`, optional offline expansion | `full_model` + `tts` | `test-nightly.yml` (`:full_moon: TTS · Function Test with L4`) | Feature matrix; perf → `tests/dfx/perf/tests/test_tts.json` |
+
+**L2 & L3 online — same file, dual marks on the baseline smoke:** The **first / simplest** case in `test_{slug}.py` (default deploy, single non-streaming `/v1/audio/speech` or equivalent offline path) should carry **both** `@pytest.mark.core_model` **and** `@pytest.mark.advanced_model` on the **same** function so it runs in L2 (`test-ready.yml`, `--run-level core_model`, basic validation) and L3 (`test-merge.yml`, `--run-level advanced_model`, deeper validation) without duplicating the test. In-tree examples: `test_voxcpm2_tts.py::test_text_to_audio_001`, `test_qwen3_tts_customvoice.py::test_text_to_audio_001`.
+
+Heavier scenarios in the same file use **`advanced_model` only** (streaming, extra languages, concurrency, async_chunk, batch). Example: `test_voice_clone_en_streaming_001` → `advanced_model` only. When migrating L3 to nightly, move those heavier cases into `test_{slug}_expansion.py` with `full_model` and drop the dedicated merge job (see `test_ming_tts_expansion.py`, `test_glm_tts_expansion.py`).
+
+```python
+@pytest.mark.core_model
+@pytest.mark.advanced_model
+@pytest.mark.tts
+@hardware_test(res={"cuda": "L4"}, num_cards=1)
+@pytest.mark.parametrize("omni_server", tts_server_params, indirect=True)
+def test_voice_clone_en_non_streaming_001(omni_server, openai_client) -> None:
+    openai_client.send_audio_speech_request({...})
+```
+
+**L4 consolidation:** Prefer parametrized `OmniServerParams` rows (`default`, `async_chunk`, feature flags) in one expansion module rather than many merge-only files ([#1832](https://github.com/vllm-project/vllm-omni/issues/1832)).
+
+**L4 performance (high-priority models):** Add latency / throughput / stress rows in `tests/dfx/perf/tests/test_tts.json`, or a dedicated `tests/dfx/perf/tests/test_{slug}.json` when the model must not join the shared nightly server matrix before integration lands (see VoxCPM2 / Coqui XTTS pattern). Register the model in `benchmarks/tts/model_configs.yaml` for local `bench_tts.py`. Wire a **separate** `test-nightly.yml` Perf Test step when the JSON is not merged into `test_tts.json` yet.
+
+**Keep model-specific code inside test modules — not `tests/helpers/{slug}.py`:**
+
+- Put `MODEL`, deploy path, vendored `REF_AUDIO_URL`, `get_prompt()`, and inline `request_config` dicts **in each** `test_{slug}.py`, `test_{slug}_expansion.py`, offline `test_{slug}.py`, and L1 `test_{slug}_*.py` as needed.
+- **Do not** add `tests/helpers/{slug}.py` (or `tests/helpers/{model_name}.py`) to deduplicate constants or request builders across those files. A little duplication is intentional; follow in-tree references such as `tests/e2e/online_serving/test_glm_tts.py` and `tests/e2e/online_serving/test_cosyvoice3_tts_expansion.py`.
+- `tests/helpers/` is for **repo-wide** harness code only (`mark.py`, `media.py`, `runtime.py`, `stage_config.py`, `assertions.py`, `fixtures/`). Import those; do not extend the tree with per-model modules.
+
+**Runtime send helpers (`tests/helpers/runtime.py`) — online and offline e2e:**
+
+| Path | Fixture | Call |
+|------|---------|------|
+| Online `/v1/*` | `openai_client` | `openai_client.send_*_request(request_config)` |
+| Offline inference | `omni_runner_handler` | `omni_runner_handler.send_*_request(request_config)` |
+
+1. **Grep `runtime.py` first** — reuse `send_omni_request`, `send_diffusion_request`, `send_audio_speech_request` (online + offline Qwen-style TTS), `send_single_stage_tts_request` (Coqui XTTS / MOSS-TTS-Nano offline), etc.
+2. **No matching helper** → add `send_<feature>_request` (or `send_<route>_http_request` for negative/dfx) in **`runtime.py`** with general `assert_*` bundled inside, **then** call it from the test.
+3. **Test file** holds `request_config` dicts only — not `omni.generate`, not `_collect_audio()`, not raw HTTP/SDK.
+4. Wire `tests/helpers/runtime.py` into Buildkite `source_file_dependencies` when you add helpers.
+
+See [vllm-omni-test skill](../vllm-omni-test/SKILL.md) § **Runtime send helpers** for full tables and exceptions.
+
 ### Deliverables
 
-- Updated `serving_speech.py` with model-specific prompt builder
-- Client scripts and server launcher
-- Gradio demo with streaming and voice cloning UI
-- Documentation (offline + online serving docs)
+- Updated `serving_speech.py` with all 5 integration points (single commit)
+- Client scripts and server launcher under `examples/online_serving/text_to_speech/<model>/`
+- Gradio demo with streaming and voice cloning UI in the same dir
+- E2E tests per **Test Case Writing (CI Levels)** above (priority tier determines L1–L4 scope)
+- Buildkite wired per level: `test-ready.yml` (L1/L2), `test-merge.yml` or nightly function job (L3), `test-nightly.yml` (L4) — see [vllm-omni-test skill](../vllm-omni-test/SKILL.md)
+- New section in `examples/online_serving/text_to_speech/README.md` (table row + per-model section). Do **not** create a top-level `examples/online_serving/<model>/` dir or a per-model `README.md` inside `text_to_speech/<model>/`.
+
+### E2E test pitfalls to avoid
+
+- **One `OmniServerParams` set per file.** `omni_server` is module-scoped; a second
+  id in the same file forces mid-module teardown/restart and exposes startup
+  races (`APIConnectionError` on the first request post-restart). Split variants
+  into separate files instead.
+- **No external URL fetches from the server.** CI and some dev hosts can't
+  reach `raw.githubusercontent.com` over TLS. Inline ref audio as
+  `data:audio/wav;base64,...`; the serving layer accepts both URL and data URL.
+- **Use the harness readiness gate.** The fixture waits for HTTP 200 on
+  `/health`; don't add `time.sleep` in tests. If warmup is incomplete, make
+  `/health` return non-200 until you're actually ready.
+- **Mark tests per the CI Levels table** — baseline smoke: `core_model` + `advanced_model`; heavier cases: `advanced_model` only; L4 expansion: `full_model`
+- **No per-model helper modules** — do not create `tests/helpers/{slug}.py`; keep constants and `request_config` payloads in the test file
+- **Online and offline e2e go through `runtime.py`** — `openai_client.send_audio_speech_request` (online); `omni_runner_handler.send_audio_speech_request` (Qwen-style offline) or `send_single_stage_tts_request` (single-stage offline). Add a new `send_*_request` in `runtime.py` when none fits; do not embed `omni.generate` or HTTP in tests
 
 ## Phase 4: Async Chunk (Streaming)
 
@@ -199,35 +454,10 @@ Stage 0 (AR)                    Stage 1 (Decoder)
    - Fixed batch size (fall back to eager for other sizes)
    - No dynamic control flow inside the graph
 
-### Example: Code Predictor CUDA Graph (Qwen3-TTS)
-
-```python
-import torch
-
-class CodePredictorGraph:
-    """Captures the 16-step code predictor AR loop as a single CUDA graph."""
-
-    def setup_graph(self, device: torch.device, kv_heads: int = 4, head_dim: int = 64):
-        self.num_steps = 16
-        self.kv_cache = torch.zeros(1, kv_heads, self.num_steps, head_dim, device=device)
-        self.positions = torch.arange(self.num_steps, device=device)
-        self.causal_mask = torch.tril(torch.ones(self.num_steps, self.num_steps, device=device))
-        self.input_buf = torch.zeros(1, 1, kv_heads * head_dim, device=device)
-        self.output_buf = torch.zeros(1, self.num_steps, device=device, dtype=torch.long)
-        # Warm up, then: self.graph = torch.cuda.CUDAGraph(); self.graph.capture(...)
-
-    def run_graph(self, initial_input: torch.Tensor) -> torch.Tensor:
-        self.input_buf.copy_(initial_input)
-        self.graph.replay()
-        return self.output_buf.clone()
-```
-
-### Performance Expectations
-
-Based on Qwen3-TTS code predictor experience:
-- **3-5x speedup** for the graphed component
-- Only effective for fixed batch sizes (typically batch_size=1)
-- Falls back to eager mode for unsupported configurations
+See [references/cuda-graph-example.md](references/cuda-graph-example.md) for
+a worked skeleton (Qwen3-TTS code predictor, 16-step AR loop), performance
+expectations (3–5× on the graphed component for fixed batch_size=1), and the
+graph-safety constraints you must honor inside the captured region.
 
 ### Deliverables
 
@@ -235,10 +465,33 @@ Based on Qwen3-TTS code predictor experience:
 - Benchmark script comparing eager vs graph performance
 - Documentation of constraints and fallback behavior
 
+## Phase 6: Pre-commit and DCO
+
+**Goal**: Every commit passes `pre-commit` lint and carries a DCO
+`Signed-off-by` line that matches the author email.
+
+- Install hooks once: `pre-commit install`.
+- Run `pre-commit run --files <changed-files>` before every push; accept any
+  auto-fixes, stage, re-commit.
+- Sign every commit with `git commit -s`. DCO checks that author email and
+  `Signed-off-by` email match — `git config user.email` must match your
+  GitHub account email.
+
+Common pre-commit failures, recovery commands for missing sign-off, and the
+full `pre-commit run` invocation for a TTS model:
+[references/precommit-dco.md](references/precommit-dco.md).
+
 ## Integration Checklist
 
 Use this checklist when integrating a new TTS model:
 
+### Cross-Cutting Invariants (verify at end of every phase)
+- [ ] I1: `forward()` docstring states cumulative vs delta; consolidation path audited end-to-end
+- [ ] I2: Tests / examples / benchmarks never use `dict.get(a) or dict.get(b)` on tensor values; list form handled
+- [ ] I3: No `.item()` / `.cpu()` / Python branch on tensor values inside per-step loops
+- [ ] I4: Offline RTF, browser streaming playback, and concurrent-request smoke test all pass
+- [ ] I5: Any cross-step cache keyed by `_omni_req_id`; entries freed when the request finishes
+
 ### Phase 1: HF Reference
 - [ ] Reference model runs and produces correct audio
 - [ ] Architecture documented (stages, codebooks, tokens, sample rate)
@@ -248,17 +501,27 @@ Use this checklist when integrating a new TTS model:
 - [ ] Model registered in `registry.py`
 - [ ] Config classes created with `model_type` registration
 - [ ] Stage 0 (AR) implemented and generates correct tokens
-- [ ] Stage 1 (Decoder) produces correct audio from tokens
+- [ ] Stage 1 (Decoder) produces correct audio from tokens — dtype float32 for codec decoder
+- [ ] Stage 1 `max_num_seqs` ≥ 4 in production config (default 1 causes gaps under concurrency)
+- [ ] Optional dependency fallbacks handled at `load_weights()` time (torchaudio/soundfile/etc.)
+- [ ] Streaming: codec codes accumulated across AR steps (not reset per step)
+- [ ] Streaming: delta audio emitted per chunk, not full re-decoded waveform
+- [ ] Streaming: all `forward()` return paths emit `model_outputs`
+- [ ] Streaming: per-request state keyed by request ID (not shared across requests)
+- [ ] Streaming: codec tensors moved to codec decoder device before decode
 - [ ] Stage config YAML created
 - [ ] `end2end.py` produces audio matching reference quality
 - [ ] README.md written
 
 ### Phase 3: Online Serving
-- [ ] Model added to `serving_speech.py`
+- [ ] All 5 `serving_speech.py` integration points added in one commit
+- [ ] Only extract params in `_build_*_params` that are forwarded to the model call (ruff F841)
 - [ ] Prompt builder handles text input correctly
 - [ ] Voice cloning works (if supported)
 - [ ] All response formats work (wav, mp3, flac, pcm)
 - [ ] Client scripts and server launcher created
+- [ ] E2E tests added per model priority tier (see **Test Case Writing (CI Levels)**)
+- [ ] Buildkite entries match level: `test-ready.yml` / `test-merge.yml` or nightly TTS job / `test-nightly.yml`
 - [ ] Gradio demo working
 - [ ] Documentation added (offline + online docs, nav, supported models)
 
@@ -276,9 +539,26 @@ Use this checklist when integrating a new TTS model:
 - [ ] Benchmark shows meaningful speedup
 - [ ] Fallback to eager works for unsupported configs
 
+### Phase 6: Pre-commit and DCO
+- [ ] `pre-commit run --files <changed>` passes before every push
+- [ ] Every commit has `Signed-off-by` matching the author email (`git commit -s`)
+- [ ] `git config user.email` matches the email registered on your GitHub account
+- [ ] Details and failure-recovery commands: [references/precommit-dco.md](references/precommit-dco.md)
+
 ## References
 
-- [TTS audio skill](../vllm-omni-audio-tts/SKILL.md) -- supported models and usage
-- [Fish Speech integration](../vllm-omni-audio-tts/references/fish-speech.md) -- complete example of Phases 1-3
-- [Qwen3-TTS reference](../vllm-omni-audio-tts/references/qwen-tts.md) -- complete example of all 5 phases
+In-skill references (details split out of the main body):
+
+- [references/single-stage-ar.md](references/single-stage-ar.md) — full `forward()` / generator skeleton for the MOSS-TTS-Nano-style pattern
+- [references/optional-deps.md](references/optional-deps.md) — torchaudio / torchcodec fallback pattern
+- [references/cuda-graph-example.md](references/cuda-graph-example.md) — Qwen3-TTS code-predictor CUDA graph skeleton
+- [references/precommit-dco.md](references/precommit-dco.md) — full pre-commit invocation, failure table, DCO recovery
+
+Project docs and adjacent skills:
+
+- [TTS audio skill](../vllm-omni-audio-tts/SKILL.md) — supported models and usage
+- [Fish Speech integration](../vllm-omni-audio-tts/references/fish-speech.md) — complete example of Phases 1–3
+- [Qwen3-TTS reference](../vllm-omni-audio-tts/references/qwen-tts.md) — complete example of all 5 phases
+- [vllm-omni-test skill](../vllm-omni-test/SKILL.md) — L1–L4 markers, naming, Buildkite wiring, run commands
 - [Adding a TTS model (developer guide)](https://github.com/vllm-project/vllm-omni/blob/main/docs/contributing/model/adding_tts_model.md)
+- `plan/voxcpm2_native_ar_design.md` — VoxCPM2's vLLM-native AR + side-computation pattern (distinct from the generator-based single-stage described above)

.claude/skills/add-tts-model/references/cuda-graph-example.md

@@ -0,0 +1,42 @@
+# CUDA Graph Example: Qwen3-TTS Code Predictor
+
+Reference sketch for capturing the 16-step code-predictor AR loop as a single
+CUDA graph. Adapt the shapes, number of steps, and KV-head layout to your
+model.
+
+```python
+import torch
+
+class CodePredictorGraph:
+    """Captures the 16-step code predictor AR loop as a single CUDA graph."""
+
+    def setup_graph(self, device: torch.device, kv_heads: int = 4, head_dim: int = 64):
+        self.num_steps = 16
+        self.kv_cache = torch.zeros(1, kv_heads, self.num_steps, head_dim, device=device)
+        self.positions = torch.arange(self.num_steps, device=device)
+        self.causal_mask = torch.tril(torch.ones(self.num_steps, self.num_steps, device=device))
+        self.input_buf = torch.zeros(1, 1, kv_heads * head_dim, device=device)
+        self.output_buf = torch.zeros(1, self.num_steps, device=device, dtype=torch.long)
+        # Warm up, then: self.graph = torch.cuda.CUDAGraph(); self.graph.capture(...)
+
+    def run_graph(self, initial_input: torch.Tensor) -> torch.Tensor:
+        self.input_buf.copy_(initial_input)
+        self.graph.replay()
+        return self.output_buf.clone()
+```
+
+## Performance expectations (Qwen3-TTS code predictor)
+
+- **3–5× speedup** on the graphed component.
+- Effective only for fixed batch sizes (typically `batch_size=1`).
+- Fall back to eager for any shape/config that wasn't captured — do not try
+  to recapture per request.
+
+## Graph-safety constraints
+
+- `torch.argmax` instead of `torch.multinomial`.
+- Fixed batch size.
+- No Python control flow that branches on tensor values inside the captured
+  region (use `torch.where` / masks).
+- No `.item()`, `.cpu()`, `.tolist()` — each would break the capture or
+  cause a GPU→CPU sync during replay.

.claude/skills/add-tts-model/references/optional-deps.md

@@ -0,0 +1,47 @@
+# Optional Dependency Handling
+
+Models that rely on `torchaudio`, `torchcodec`, `soundfile`, or other optional
+packages must handle the missing-package case at import time, not at call
+time. Failing to do this causes cryptic errors only on environments without
+the optional package — after the model is already deployed.
+
+## Pattern (used in MOSS-TTS-Nano)
+
+```python
+def _patch_torchaudio_load() -> None:
+    """Fallback torchaudio.load/save to soundfile if torchcodec is unavailable."""
+    try:
+        import torchcodec  # noqa: F401
+        return  # torchcodec present, torchaudio works as-is
+    except ImportError:
+        pass
+
+    import soundfile as sf
+    import torchaudio
+
+    def _sf_load(path, **kwargs):
+        data, sr = sf.read(str(path), dtype="float32", always_2d=True)
+        return torch.from_numpy(data).T, sr
+
+    torchaudio.load = _sf_load
+    # patch .save similarly if needed
+```
+
+## Rules
+
+- Mirror the full signature of the replaced function. `torchaudio.load`
+  accepts `frame_offset`, `num_frames`, `normalize`, `channels_first`,
+  `format` — missing any of them causes `TypeError` from calling code.
+- Catch `except Exception`, not just `ImportError`. `import torchaudio`
+  itself can fail with non-`ImportError` errors on broken installs. Log the
+  exception type and message (`logger.warning("torchaudio probe failed: %s: %s",
+  type(exc).__name__, exc)`) before falling back, so unrelated errors are not
+  silently swallowed.
+- Call the patch function at the top of `load_weights()` before loading any
+  audio assets. Do not call it at module import time.
+
+## Reference implementation
+
+Any in-tree model that patches `torchaudio.load` in its `load_weights()` —
+e.g. MOSS-TTS-Nano's `modeling_moss_tts_nano.py` once that integration
+lands.

.claude/skills/add-tts-model/references/precommit-dco.md

@@ -0,0 +1,54 @@
+# Pre-commit and DCO
+
+Every commit must pass `pre-commit` lint and carry a `Signed-off-by` line
+that matches the commit author email.
+
+## Pre-commit
+
+Install hooks once:
+
+```bash
+pre-commit install
+```
+
+Run before every push on the files you changed:
+
+```bash
+pre-commit run --files \
+  vllm_omni/model_executor/models/<model_name>/*.py \
+  vllm_omni/entrypoints/openai/serving_speech.py \
+  vllm_omni/model_executor/models/registry.py \
+  tests/e2e/offline_inference/test_<model_name>.py \
+  tests/e2e/online_serving/test_<model_name>.py
+```
+
+When pre-commit **modifies files** (ruff format auto-fix), it exits non-zero
+but the changes are correct — stage the modified files and re-commit.
+
+| Failure | Root cause | Fix |
+|---------|-----------|-----|
+| `ruff F841` | Variable extracted but never forwarded to model call | Remove the extraction or wire it through |
+| `ruff E402` | Import added below function definitions | Move to top-level import block |
+| `ruff format` | Line length, spacing, quote style | Accept auto-fix, stage, re-commit |
+
+## DCO sign-off
+
+Every commit must carry `Signed-off-by: Your Name <your@email.com>`. Use
+`-s`:
+
+```bash
+git commit -s -m "feat(<model>): add <Model> TTS support"
+```
+
+Or set it permanently: `git config format.signOff true`.
+
+The DCO check verifies that the commit author email matches the
+`Signed-off-by` line. Confirm `git config user.email` matches your GitHub
+account email before committing.
+
+Fix a missing or mismatched sign-off on the latest commit:
+
+```bash
+git commit --amend -s --no-edit
+git push origin <branch> --force-with-lease
+```

.claude/skills/add-tts-model/references/single-stage-ar.md

@@ -0,0 +1,108 @@
+# Single-Stage AR Pattern
+
+When the upstream model cannot be cleanly split into an AR stage and a separate
+decoder (e.g. MOSS-TTS-Nano, or any model that bundles AR + codec via an
+`inference_stream()` generator), run the whole pipeline inside a single AR
+worker that yields audio chunks per request.
+
+This is distinct from VoxCPM2's pattern, which also runs in a single stage but
+uses vLLM's native PagedAttention on the base language model with diffusion /
+VAE side-computation outside vLLM — see
+`plan/voxcpm2_native_ar_design.md` for that variant.
+
+## Implementation
+
+1. **Single model file** — load both AR LM and codec inside
+   `modeling_<model>.py`.
+2. **Load weights in `load_weights()`**, not `__init__()` — vLLM initializes
+   distributed state before any CUDA allocations.
+3. **Stream via a per-request generator** stored in `self._stream_gens`:
+
+```python
+class YourModelForCausalLM(nn.Module):
+    def __init__(self, *, vllm_config, prefix=""):
+        super().__init__()
+        self._lm = None                   # populated in load_weights()
+        self._stream_gens: dict = {}      # request_key → generator
+
+    def load_weights(self, weights):
+        # Load self._lm here, after vLLM distributed init
+        ...
+
+    def forward(
+        self,
+        input_ids,
+        positions,
+        intermediate_tensors=None,
+        inputs_embeds=None,
+        runtime_additional_information: list[dict] | None = None,  # one dict per request
+        **kwargs,
+    ) -> OmniOutput:
+        infos = runtime_additional_information or [{}]
+        # Skip dummy/profiling calls
+        if not runtime_additional_information or all(i.get("_is_dummy") for i in infos):
+            self._ar_emit_stop_token = True
+            return OmniOutput(...)  # return empty outputs
+
+        outputs, last_flags = [], []
+        for info in infos:
+            request_key = str(info.get("_omni_req_id", "0"))  # per-request ID from vLLM
+            if request_key not in self._stream_gens:
+                self._stream_gens[request_key] = self._create_stream_gen(info)
+            try:
+                chunk, is_last = next(self._stream_gens[request_key])
+            except StopIteration:
+                chunk, is_last = torch.zeros(0), True
+            if is_last:
+                del self._stream_gens[request_key]
+            outputs.append(chunk)
+            last_flags.append(is_last)
+
+        self._ar_emit_stop_token = all(last_flags)
+        return OmniOutput(multimodal_outputs={"model_outputs": outputs, ...})
+
+    def _create_stream_gen(self, info: dict):
+        """Yield (waveform_tensor, is_last) tuples from inference_stream()."""
+        for event in self._lm.inference_stream(...):
+            if event["type"] == "audio":
+                yield event["waveform"], False
+            elif event["type"] == "result":
+                # Fallback: some models emit a single "result" event instead of
+                # incremental "audio" events — handle both paths
+                yield event.get("waveform", torch.zeros(0)), True
+                return
+        yield torch.zeros(0), True
+
+    def compute_logits(self, hidden_states, sampling_metadata):
+        # Emit EOS only after the last chunk so the AR scheduler ends the request
+        ...
+```
+
+## Key points
+
+- `runtime_additional_information` is the correct parameter name (not
+  `**kwargs`) — it carries one dict per request in the batch.
+- The request ID is `info.get("_omni_req_id")` — set by vLLM, not by user code.
+- Handle both `"audio"` (incremental) and `"result"` (final combined) event
+  types from upstream models.
+
+## Stage config
+
+Single stage with `worker_type: ar`, `engine_output_type: audio`,
+`final_output: true`, `is_comprehension: true`, and `async_chunk: false` at
+the top level. Omitting any of these causes silent misclassification in the
+serving layer.
+
+## Lint discipline
+
+Only extract variables from `additional_information` that you actually
+forward to the model call — unused extractions trip `ruff F841` in
+pre-commit.
+
+## Reference implementation
+
+Look for any single-stage AR model under
+`vllm_omni/model_executor/models/` — e.g. `moss_tts_nano/` when its
+integration lands. If none is in tree yet, follow the skeleton above and
+cross-check against the `is_comprehension: true` / `async_chunk: false`
+dispatch in `vllm_omni/entrypoints/openai/serving_speech.py`.

.claude/skills/diffusion-perf-opt/SKILL.md

@@ -0,0 +1,510 @@
+---
+name: diffusion-perf-opt
+description: Diagnose and optimize vLLM Omni diffusion workloads, especially Wan/Qwen/Flux-style image and video generation. Use when Codex is asked to analyze profiling traces, choose parallel strategies, inspect torch profiler trace.json or trace.json.gz timelines, estimate optimization ROI, investigate GPU idle/free bubbles, compare USP/CFG/HSDP/VAE parallelism, or design operator/host/quantization optimizations for vLLM Omni.
+---
+
+# vLLM Omni Performance Optimization
+
+Use this skill to run a disciplined optimization loop for vLLM Omni diffusion workloads. Keep two ideas separate: real performance baselines are collected with low overhead, while torch profiler traces are diagnostic artifacts and may distort latency.
+
+## First Questions
+
+Before proposing changes, ask for the optimization scene if it is not already known:
+
+- GPU model, card count, topology, and whether NVLink is present.
+- Model and pipeline, for example Wan2.2 I2V A14B.
+- User workload: resolution, frames, steps, batch/concurrency, CFG scales, prompt/image inputs.
+- Three runnable commands for each model/workload, or scripts that generate them:
+  - **Server startup command**: exact `vllm serve` command, environment variables, model path, port, parallelism flags, profiler flags, and precision/compile settings.
+  - **User request command**: exact single-request client command (for example curl or Python client) with fixed prompt/media/seed/size/steps. Use this to validate correctness and collect per-request stage timings.
+  - **User benchmark command**: exact repeatable benchmark command or script with warmup count, measured iteration count, concurrency/batch policy, output directory, and summary format.
+- Current enabled strategies: USP/SP, CFG parallel, HSDP/FSDP, VAE patch parallel, torch.compile, profiler options.
+- Optimization target: latency, throughput, memory, cost, or quality-preserving speed.
+- Precision/quality tolerance: bf16/fp8/quantization/sparsity/approximate attention allowed or not.
+- Quality validation method: exact-output comparison when feasible, or metrics
+  such as SSIM, PSNR, LPIPS, cosine similarity, MAE, MSE, temporal flicker
+  checks, baseline self-run variance, and reviewable output artifacts.
+
+## Workflow
+
+1. **Freeze the measurement protocol and commands.**
+   - Before analyzing parallel strategies or traces, establish the three commands for the target model/workload: server startup, single user request, and user benchmark.
+   - Prefer checked-in scripts or one-off shell scripts over free-form commands in chat. The scripts should make workload variables explicit, including model path, port/base URL, prompt/media inputs, resolution, frames, steps, seed, warmup, measured iterations, concurrency, and output path.
+   - Measurement reports must preserve the concrete process, not just the final numbers. For each tested configuration, write down the server startup command, the user request command, the user benchmark command or polling loop, the final server response metrics, and the client-side timing/HTTP result.
+   - Present all duration and latency values in **milliseconds (ms)** in tables and summaries. If an API returns seconds, convert to ms before reporting. Keep raw units only inside quoted raw logs or source snippets.
+   - Keep profiler modes separate in the command set:
+     - Baseline/benchmark commands should avoid torch profiler and stack collection.
+     - Diagnostic commands may enable torch profiler after a baseline identifies a bottleneck.
+   - If any of the three commands is missing, ask for it or create a proposed script before proceeding.
+   - Make clear which command is authoritative for correctness validation and which command is authoritative for performance numbers.
+
+2. **Collect a real baseline.**
+   - Disable PyTorch profiler and stack collection.
+   - Disable or fix torch.compile state so A/B is fair.
+   - Avoid `--enforce-eager` for production-speed baselines unless eager is the target.
+   - Prefer `--log-stats` and `--enable-diffusion-pipeline-profiler` for low-overhead stage timing. PR 3069 has the relevant metrics/log-stats changes; if local code does not include them, fetch or cherry-pick the minimal metrics changes rather than merging unrelated PR drift.
+   - Run warmup requests and exclude first-request lazy init.
+
+3. **Model the parallel strategy before testing.**
+   - Estimate compute and communication for self-attention, cross-attention, FFN, CFG, VAE encode/decode, and HSDP.
+   - Select a small candidate matrix rather than exhaustively testing everything.
+   - Typical 2-card candidates: `USP=2`, `CFG=2`, `USP=1/HSDP on-off`, VAE parallel on-off if memory allows.
+   - Typical 4-card candidates: `USP=4`, `CFG=2 x USP=2`, `USP=2 x HSDP`, VAE parallel on-off.
+   - Typical 8-card candidates for official CFG-enabled video diffusion models:
+     - Primary candidate: `CFG=2 x USP=4` with VAE patch parallel across all 8 ranks.
+     - Compare against: `CFG=1 x USP=8` to test whether larger sequence parallel groups beat CFG branch parallelism.
+     - Isolate VAE parallelism: keep the DiT strategy fixed and compare VAE patch parallel on/off or different VAE patch sizes.
+     - If long-video `diffuse` remains dominant and Ulysses all-to-all is suspected, test a hybrid sequence strategy such as `CFG=2 x USP=2 x Ring=2`.
+     - Test HSDP off only if the model fits without it; treat HSDP primarily as a memory strategy until A/B proves otherwise.
+   - Prefer USP/SP for long video token sequences; prefer CFG parallel when CFG doubles transformer forwards and sequence length is modest.
+   - Convert workload shape to latent/patch token counts before choosing candidates. For video models, record frames, latent frames, latent HxW, patch size, approximate token count, and which stages should scale with the token count.
+
+4. **Search for the best parallel configuration.**
+   - Start with a small matrix that answers one question per comparison:
+     - `CFG parallel vs larger USP`: compare `CFG=2 x USP=world/2` against `CFG=1 x USP=world` for CFG-enabled workloads.
+     - `VAE patch parallel value`: compare the best DiT strategy with VAE patch parallel enabled and disabled.
+     - `HSDP cost`: compare HSDP on/off only when both configurations fit memory.
+     - `Ulysses vs Ring`: test a Ring/Ulysses hybrid only after long-sequence `diffuse` is confirmed dominant.
+   - For each configuration, create or record a stable config id, for example `A_cfg2_usp4_vaepp8_hsdp_tiling`.
+   - For each config id, capture:
+     - exact server command and environment variables,
+     - observed distributed setup from logs, such as SP groups, CFG groups, HSDP shard/replicate sizes, VAE patch size,
+     - exact request command for every scenario,
+     - final server response metrics and client-side elapsed time,
+     - output artifact paths and any failed/empty responses.
+   - Use one warmup request per scenario, then at least three measured repeats for the shortlisted config. For exploratory matrix pruning, one measured request is acceptable only if the margin is large; label it as one-shot.
+   - Compare configurations by stage, not only end-to-end latency. A configuration can improve `diffuse` while hurting `vae.decode`; record both effects.
+   - Select the best config only after checking the target metric, dominant stages, memory headroom, and output correctness.
+
+5. **Run targeted A/B tests.**
+   - Change one variable per test.
+   - Keep model, input, seed, request parameters, GPU placement, and warmup policy fixed.
+   - Record latency, stage timings, memory, output quality, and logs.
+   - Report comparison tables in ms. Include at least end-to-end client time, server `inference_time`, server stage generation time if available, `vae.encode`, `diffuse`, `vae.decode`, and peak memory.
+
+6. **Enforce the quality and precision gate for every optimization.**
+   - Do not mark an optimization ready until quality is checked on the same
+     workload: model, prompt/media, seed, shape, steps, scheduler, dtype,
+     backend, parallelism, and output encoding.
+   - For math-preserving changes, require exact/near-exact agreement or show
+     that differences are within baseline self-run variance.
+   - For precision, quantization, approximate attention, sparse/custom kernels,
+     or backend changes, include reviewable artifacts and metrics such as SSIM,
+     PSNR, LPIPS, cosine similarity, MAE, MSE, and temporal flicker checks.
+   - If quality tolerance is not stated, default to quality-preserving behavior.
+     Failed or inconclusive quality validation blocks a ready-to-merge claim.
+
+7. **Collect diagnostic trace only after narrowing hypotheses.**
+   - Use torch profiler for a small number of requests.
+   - Run two separate diagnostic traces instead of mixing concerns:
+     - **Operator/shape trace**: enable `torch_profiler_record_shapes=True` and keep stack collection disabled. Use this to rank CUDA kernels, NCCL collectives, attention/MLP/norm/RoPE work, and shape-specific hot operators.
+     - **Host-stack trace**: enable `torch_profiler_with_stack=True` and normally keep shape collection disabled. Use this to map CPU/Python host gaps, synchronization points, scheduler paths, and request handling overhead.
+   - Keep both trace commands and reports separate from baseline/benchmark commands. Torch profiler latency is diagnostic only and must not be used as the final latency claim.
+   - Prefer profiling only the narrowed dominant scenario, for example the highest-resolution/video-length case where `diffuse` dominates.
+   - If profiler endpoints are available, run one warmup request first, then call `/start_profile`, run one profiled request, and call `/stop_profile`. This keeps model initialization and warmup out of the diagnostic trace.
+   - Start by analyzing rank 0 only. Expand to more ranks only if rank 0 suggests imbalance, unclear GPU idle/free bubbles, high NCCL wait, server timing mismatch, CFG branch imbalance, or USP group stragglers.
+   - Diagnostic reports must be written to disk and preserve: server command, profiler config, warmup/request/polling commands, trace artifact paths, rank analyzed, analyzer output or summary, and the decision about whether additional ranks are necessary.
+   - Analyze both rank-level balance and device-level free bubbles when additional ranks are opened.
+
+8. **Analyze host, communication, and operators.**
+   - Find GPU idle/free intervals and map each large gap to the enclosing CPU/Python code.
+   - Separate real GPU idle from profiler overhead such as CUPTI `Command Buffer Full`.
+   - Compare NCCL kernel time to user annotations; annotations can overcount nested intervals.
+   - Rank operator work by total CUDA time and by repeated small-kernel launch count.
+
+9. **Produce an optimization plan.**
+   - Classify candidates as P0/P1/P2.
+   - For each candidate, state necessity, expected benefit, implementation path, validation plan, and quality risk.
+   - Include a concrete quality/precision validation gate for each candidate.
+   - Do not implement high-risk operator rewrites before proving the operator is a bottleneck for the target shapes.
+   - End the plan with a user-facing candidate selection table. The assistant
+     should not automatically choose a risky optimization just because it is
+     technically possible. Present the options clearly and let the user decide
+     which item is worth implementing based on latency target, engineering
+     budget, memory headroom, and quality tolerance.
+   - Organize the plan by optimization layer, not by a flat list of ideas:
+     - host/runtime optimization,
+     - measurement/benchmark reliability,
+     - parallelism and communication,
+     - VAE encode/decode and media pre/post processing,
+     - operator fusion and layout cleanup,
+     - attention main-path optimization,
+     - algorithmic, precision, or approximation changes.
+   - For each layer, tie every candidate back to evidence from baseline
+     metrics, diagnostic trace, source code, or output quality requirements.
+     Do not list generic optimizations without a trace or workload reason.
+
+## Priority Rules
+
+- **P0:** low risk, likely useful, or required for trustworthy measurement. Examples: real baseline, warmup, targeted parallel A/B, disabling avoidable `empty_cache`, scheduler coefficient caching.
+- **P1:** meaningful code changes with contained risk. Examples: cross-attention KV caching, VAE gather/broadcast reduction, AdaLayerNorm/RMSNorm/RoPE fusion after trace evidence.
+- **P2:** high implementation or quality risk. Examples: FA to LA replacement, custom Triton/CUDA fused kernels, FP8/quantization, sparsity/Rainfusion-style acceleration.
+
+Every implemented optimization needs A/B validation and a passing
+quality/precision gate, including math-preserving changes such as padding trim,
+layout cleanup, cache reuse, or host/runtime cleanup. Use objective metrics
+when available, keep reviewable artifacts, and compare against baseline
+self-run variance for precision, quantization, approximate kernels, or backend
+changes. If quality validation fails or is inconclusive, do not present the
+optimization as ready to merge.
+
+## Optimization Layers
+
+After baseline, parallel-search, and diagnostic traces, summarize optimization
+opportunities by layer. This is the core of the performance analysis: the goal
+is to connect evidence to a scoped implementation and a validation plan.
+
+### Host and Runtime Optimization
+
+Purpose: remove CPU/Python stalls, synchronization points, allocator overhead,
+and request-path overhead that leave GPU lanes empty.
+
+Evidence to look for:
+
+- High `idle_pct` or large `GAP` blocks in `trace_analyzer.py`.
+- Host-stack trace lines such as `torch.cuda.empty_cache`,
+  `cudaStreamSynchronize`, `cudaDeviceSynchronize`, Python locks, scheduler
+  waits, image/video preprocessing, or repeated small allocation paths.
+- Difference between client wall-clock time and server `inference_time_s`.
+
+Typical candidates:
+
+- Make avoidable `torch.cuda.empty_cache()` optional or guard it by memory
+  headroom.
+- Cache scheduler coefficients, timesteps, masks, or other tiny repeated CPU
+  computations when the request shape/steps are fixed.
+- Remove avoidable host-device synchronizations and blocking logging/stat calls.
+- Move expensive preprocessing out of the critical path or cache fixed prompt,
+  image, and transform work for benchmark scenarios.
+- Ensure benchmark scripts record client-side elapsed time, HTTP status, output
+  path, and server response metrics.
+
+Priority guidance:
+
+- Usually P0 when the change is measurement reliability or an obvious removable
+  synchronization.
+- Usually P1 when it changes scheduling, memory lifetime, or request execution
+  order.
+
+Validation:
+
+- Re-run non-profiler baseline with same workload and seed.
+- Confirm peak memory headroom if disabling cache cleanup.
+- Confirm generated output exists and quality/seed behavior is unchanged.
+
+### Parallelism and Communication Optimization
+
+Purpose: choose the right decomposition for CFG branches, sequence tokens, model
+weights, VAE tiles, and rank topology.
+
+Evidence to look for:
+
+- Baseline A/B across `CFG`, `USP/SP`, `Ring`, `HSDP/FSDP`, and VAE patch
+  parallelism.
+- Stage timing shifts: `diffuse`, `vae.encode`, `vae.decode`, and server
+  end-to-end.
+- NCCL kernel time from trace, not only `user_annotation` time.
+- Rank imbalance across SP group ranks or CFG branch ranks.
+- Memory headroom and OOM risk.
+
+Typical candidates:
+
+- `CFG=2 x USP=world/2` versus `CFG=1 x USP=world` for CFG-enabled models.
+- VAE patch parallel on/off or patch size tuning.
+- HSDP/FSDP on/off only if both configurations fit memory.
+- Ulysses versus Ulysses+Ring only after long-sequence `diffuse` is confirmed
+  dominant and all-to-all is suspected.
+- Rank mapping/topology changes if all-rank traces show stragglers or NCCL wait.
+- Buffer reuse or preallocation for FSDP/HSDP all-gather paths.
+
+Priority guidance:
+
+- P0/P1 for configuration-only changes with strong measured wins.
+- P1 for buffer reuse or rank mapping changes.
+- P2 for invasive distributed algorithm changes.
+
+Validation:
+
+- Measure by stage and memory, not only end-to-end.
+- Use one-variable A/B with identical prompt/media/seed/shape/steps.
+- When communication is suspected, compare rank0-3 in one USP group and rank0
+  versus rank4 across CFG branches for `CFG=2 x USP=4`.
+
+### VAE and Media Pipeline Optimization
+
+Purpose: reduce encode/decode, tiling, split/gather, and media conversion time.
+
+Evidence to look for:
+
+- Large `vae.encode` or `vae.decode` in low-overhead stage timings.
+- Host-stack gaps in VAE tile split, gather, merge, broadcast, or image/video
+  transforms.
+- VAE kernels or cuDNN convolution in operator trace.
+- Whether every rank needs the final decoded tensor.
+
+Typical candidates:
+
+- Keep VAE patch parallel enabled when it has clear measured benefit.
+- Reduce VAE gather/broadcast to only ranks that need the final media output.
+- Reuse tile metadata, split buffers, or gather buffers.
+- Evaluate bf16/autocast behavior for VAE only with visual quality checks.
+- Avoid redundant image conversion, resize, or tensor construction in repeated
+  benchmark runs.
+
+Priority guidance:
+
+- P0/P1 if VAE is a large share of the target workload or if a host gap is
+  obvious and low risk.
+- Lower priority when `diffuse` dominates and VAE is already patch-parallelized.
+
+Validation:
+
+- Compare `vae.encode`, `vae.decode`, server end-to-end, and peak memory.
+- Check output video integrity, artifacts, flicker, and seed stability.
+
+### Operator Fusion and Layout Cleanup
+
+Purpose: reduce high-frequency small kernels, memory bandwidth pressure, layout
+conversions, and launch overhead in transformer and VAE blocks.
+
+Evidence to look for:
+
+- Top operator tables showing many `aten::copy_`, `aten::cat`,
+  `split_with_sizes_copy`, `aten::add`, `aten::mul`, `aten::div`, norm,
+  activation, RoPE, or reshape/layout kernels.
+- `ops_rankN.xlsx` `by_shape` sheet showing repeated small shapes inside the
+  same block path.
+- Trace lanes showing many short kernels between larger GEMM/attention kernels.
+- Source code patterns with repeated elementwise chains or layout conversions.
+
+Typical fusion targets:
+
+- AdaLayerNorm / RMSNorm / LayerNorm plus scale/shift fusion.
+- RoPE fusion with Q/K layout preparation when shapes are stable.
+- Residual add, scale, gate, and elementwise chains.
+- MLP gate/up/down cleanup, such as fusing activation and multiply around
+  GEMM outputs when feasible.
+- QKV projection and reshape/split/cat path cleanup.
+- Attention pre/post layout cleanup to avoid unnecessary copies, cats, and
+  splits around sequence parallel all-to-all.
+
+Priority guidance:
+
+- P1 when implemented with existing PyTorch/Triton/local helper patterns and
+  validated against exact outputs or tolerances.
+- P2 when it requires custom CUDA/Triton kernels, changes numerics, or touches
+  attention math directly.
+
+Validation:
+
+- First prove the operator family is material for the target shape.
+- Use non-profiler A/B for latency and stage timing.
+- Use quality regression checks for generated video stability.
+- Check compile behavior and graph breaks if using `torch.compile`.
+
+### Attention Main-Path Optimization
+
+Purpose: address the dominant self-attention cost when FlashAttention or other
+attention kernels dominate CUDA time.
+
+Evidence to look for:
+
+- Operator trace where FlashAttention/SDPA kernels dominate total CUDA time.
+- Attention shape from `ops_rankN.xlsx` `by_shape`, model code, or trace
+  metadata.
+- Whether attention cost scales with latent frames, latent H/W, patch size, or
+  CFG duplication.
+- Layout/copy/all-to-all work around attention.
+
+Typical candidates:
+
+- Verify the attention backend and shape are on the intended fast kernel path.
+- Compare supported attention backends only with identical workload and quality
+  settings.
+- Reduce attention input size by safe model/config choices when allowed:
+  latent resolution, frame count, patching, boundary ratio, or windowing.
+- Remove avoidable layout conversions before/after attention.
+- Reuse condition-side KV or other static inputs if the model structure allows.
+- Consider custom kernels, sparse/window/linear attention, or approximation only
+  after quality risk is accepted.
+
+Priority guidance:
+
+- P1 for backend/config/layout changes with preserved math.
+- P2 for approximate attention, sparsity, custom kernels, or any change that can
+  alter quality/temporal consistency.
+
+Validation:
+
+- Always include output quality and seed behavior checks.
+- Compare `diffuse`, server end-to-end, peak memory, and attention kernel time
+  in diagnostic traces if needed.
+
+### Algorithmic, Precision, and Approximation Optimization
+
+Purpose: reduce mathematical work or precision cost beyond local code cleanup.
+
+Evidence to look for:
+
+- A single operator family dominates even after low-risk runtime, parallel, and
+  fusion work.
+- Memory bandwidth or compute utilization suggests precision or quantization
+  could matter.
+- The user explicitly allows quality-preserving or approximate methods.
+
+Typical candidates:
+
+- FP8/quantization for transformer or selected projections.
+- Sparsity or Rainfusion-style acceleration.
+- Reduced steps, scheduler changes, distillation, or caching across frames.
+- Approximate attention or linear attention.
+
+Priority guidance:
+
+- Usually P2 because quality, numerics, and implementation risk are high.
+
+Validation:
+
+- Requires strict A/B, visual quality review, temporal flicker checks, seed
+  stability, and possibly human evaluation.
+
+### Interpolation, Super-Resolution, and E2E Pipeline Optimization
+
+Purpose: optimize the whole user-visible video product, not only the base
+diffusion invocation. Some deployments trade base-model latency against
+post-processing, interpolation, or super-resolution stages.
+
+Evidence to look for:
+
+- E2E latency breakdown across base generation, interpolation, super-resolution,
+  encoding, storage, and response streaming.
+- Fast/slow GPU or fast/slow stage analysis across multiple cards and pipeline
+  stages.
+- User quality target: resolution, FPS, temporal smoothness, and acceptable
+  post-processing artifacts.
+
+Typical candidates:
+
+- Add or optimize a frame interpolation stage when it reduces required base
+  model frames for the same perceived FPS.
+- Add or optimize a super-resolution model when generating lower base
+  resolution plus SR is faster for the target quality.
+- Analyze E2E2 pipeline behavior: client request, service scheduling, diffusion,
+  VAE/media, post-process, file write, and response.
+- Identify fast/slow cards or stages and rebalance pipeline placement.
+
+Priority guidance:
+
+- P1 when using proven interpolation/SR components without changing diffusion
+  math.
+- P2 when quality risk is high or the pipeline adds significant operational
+  complexity.
+
+Validation:
+
+- Measure E2E wall-clock, per-stage server timings, output FPS/resolution,
+  artifacts, flicker, and user-visible quality.
+
+### Optimization Candidate Library
+
+Use this table as a compact menu, not as automatic recommendations. Pick items
+only when baseline metrics, trace evidence, source inspection, or quality
+tolerance supports them.
+
+| Layer | Candidate | Evidence | Priority | Validation focus |
+|---|---|---|---|---|
+| Measurement | Freeze server/request/benchmark commands | Missing or drifting commands | P0 | Repeatable non-profiler A/B |
+| Measurement | Separate baseline and diagnostic profiler runs | Profiler used for latency claims | P0 | Low-overhead stage timings |
+| Host/runtime | Guard or remove avoidable `empty_cache` | Host-stack gaps or sync stalls | P0/P1 | Latency, peak memory, OOM safety |
+| Host/runtime | Cache scheduler coefficients/timesteps | Repeated tiny CPU/GPU work | P0/P1 | Same seed/output, stage timing |
+| Host/runtime | Reduce framework scheduling overhead | Client time exceeds server time | P1 | E2E latency, throughput |
+| Parallel | `CFG=2 x USP=world/2` vs `USP=world` | CFG doubles forward work | P0/P1 | `diffuse`, NCCL, memory |
+| Parallel | Tune VAE patch parallelism | VAE encode/decode is material | P0/P1 | VAE time, output correctness |
+| Parallel | HSDP on/off or buffer reuse | HSDP affects memory/all-gather | P1 | Memory, latency, OOM risk |
+| Parallel | Ulysses vs Ulysses+Ring | Long sequence all-to-all suspected | P1/P2 | Rank balance, NCCL kernels |
+| Cross-attn | Disable SP for short condition tokens | Cross-attn comm exceeds compute | P1 | `diffuse`, correctness |
+| VAE/media | Reduce VAE gather/broadcast | Rank traces show VAE wait | P1 | Rank balance, output file |
+| VAE/media | Reuse tile metadata/buffers | Tile split/merge host gaps | P1 | `vae.encode/decode`, memory |
+| VAE/media | VAE bf16/autocast | VAE float kernels are slow | P1/P2 | Artifacts, flicker, seed stability |
+| Operator fusion | AdaLayerNorm/LayerNorm fusion | Norm plus scale/shift kernels | P1 | Numeric tolerance, latency |
+| Operator fusion | RMSNorm fusion | Many small RMSNorm kernels | P1 | Numeric tolerance, latency |
+| Operator fusion | RoPE cache/fuse/layout cleanup | RoPE copy/reshape kernels | P1/P2 | Kernel count, correctness |
+| Operator layout | QKV or attention layout cleanup | Copy/cat/split around attention | P1 | Copy kernels, compile behavior |
+| Attention | Verify backend fast path | FA/SDPA dominates trace | P1 | `diffuse`, attention kernels |
+| Attention | FA to LA or selected-head LA | Attention remains dominant | P2 | Quality, temporal stability |
+| Precision | Transformer FP8/quantization | Compute/bandwidth bound and allowed | P2 | Quality, speed, stability |
+| Sparsity | Rainfusion-style acceleration | DiT compute remains dominant | P2 | Prompt diversity, quality |
+| Pipeline | Frame interpolation | Fewer base frames can meet FPS | P1/P2 | E2E latency, motion artifacts |
+| Pipeline | Super-resolution | Lower base res plus SR may win | P1/P2 | Detail quality, artifacts |
+| E2E | Fast/slow-card analysis | Multi-card stragglers | P0/P1 | Per-rank/stage wall-clock |
+
+### Optimization Plan Template
+
+Use this table shape when reporting the next work items:
+
+| Priority | Layer | Candidate | Evidence | Expected benefit | Implementation path | Validation | Quality risk |
+|---|---|---|---|---|---|---|---|
+| P0 | Host/runtime | Guard `empty_cache` | Host-stack gap points to `torch.cuda.empty_cache` | Small latency reduction, less idle | Add config/env guard | Non-profiler A/B, memory check | Low |
+| P1 | Operator fusion | RMSNorm/AdaLayerNorm fusion | High-frequency norm/elementwise kernels | Lower launch/bandwidth overhead | Use existing fusion helper or targeted Triton | A/B + output check | Medium |
+| P1/P2 | Attention | Attention layout/backend investigation | FA kernel dominates CUDA time | Potentially large | Inspect shapes/backend and remove layout copies | A/B + trace + quality | Medium/high |
+
+Then present a short selection prompt using the same rows:
+
+```text
+Which candidate should we implement next?
+
+1. P0 Host/runtime: guard empty_cache
+   - Expected benefit: small but low-risk latency reduction.
+   - Risk: possible memory increase/OOM if memory headroom is insufficient.
+
+2. P1 Operator fusion: inspect by_shape and implement first norm/RoPE/layout fusion
+   - Expected benefit: medium if high-frequency small kernels are confirmed.
+   - Risk: numerical/compile/quality validation needed.
+
+3. P1/P2 Attention: FA/LA/backend/layout investigation
+   - Expected benefit: potentially large.
+   - Risk: high quality and implementation risk.
+```
+
+If the user has not chosen an item, default to explaining tradeoffs and asking
+which candidate to execute. Only proceed autonomously on low-risk P0 measurement
+or instrumentation fixes.
+
+## Analysis Helpers
+
+PyTorch profiler traces are Chrome/Perfetto-compatible JSON files, usually
+`trace_rankN.json` or `trace_rankN.json.gz`. They normally contain a top-level
+`traceEvents` list, though some exporters emit the raw event list directly.
+
+Use the checked-in analyzer from the repository root:
+
+```bash
+python3 .claude/skills/diffusion-perf-opt/scripts/trace_analyzer.py \
+  vllm_profile/.../trace_rank0.json.gz \
+  --min-gap-ms 5 \
+  --topn 20
+```
+
+For rank imbalance or communication questions, pass all relevant rank traces in
+one command. For host gaps, lower `--min-gap-ms` to `1` and use a host-stack
+trace. Read `gpu_span_s`, `busy_union_s`, `idle_union_s`, `idle_pct`, `GAP`
+blocks, `Top GPU/operator events by total duration`, and `Top NCCL-like events
+by category`. Treat `cat=user_annotation` NCCL ranges as enclosing annotations;
+prefer `cat=kernel` or `cat=gpu_user_annotation` for real device work.
+
+The analyzer summarizes timing only. It does not parse tensor shapes, attribute
+overlap to individual streams, prove quality, or provide final latency claims.
+Use `ops_rankN.xlsx` or PyTorch key averages for shape analysis, and re-test any
+optimization with non-profiler baseline commands.
+
+Read `references/optimization-playbook.md` when drafting the optimization table or comparing candidate techniques.
+
+## vLLM Omni Heuristics
+
+- Cross-attention usually should not use USP/SP when text/image condition token count is much smaller than latent video tokens. Confirm via trace; in Wan2.2 I2V, self-attention dominates cross-attention.
+- VAE bf16/autocast is often worthwhile but requires visual quality checks.
+- VAE patch parallel can help decode/encode but may add gather/merge/broadcast overhead. Check whether all ranks need the final decoded tensor.
+- HSDP/FSDP is primarily a memory strategy. If the model fits without it, run an on/off latency comparison.
+- Scheduler work can create small host/device gaps; cache tiny solve coefficients when timesteps/order are known.
+- `torch.cuda.empty_cache()` can prevent OOM but creates synchronization/idle. Make it optional if memory headroom is sufficient.
+- `Command Buffer Full` in profiler output is profiler overhead, not a model optimization target.

.claude/skills/diffusion-perf-opt/agents/openai.yaml

@@ -0,0 +1,4 @@
+interface:
+  display_name: "vLLM Omni Perf Opt"
+  short_description: "Profile and optimize vLLM Omni diffusion performance."
+  default_prompt: "Analyze and optimize a vLLM Omni diffusion workload using baselines, traces, parallel strategy modeling, and prioritized optimization plans."

.claude/skills/diffusion-perf-opt/references/optimization-playbook.md

@@ -0,0 +1,82 @@
+# Optimization Playbook
+
+Use this reference when turning trace evidence into prioritized vLLM Omni optimizations.
+
+## Parallel Strategy
+
+Start from workload shape:
+
+- Long video tokens, high resolution, many frames: self-attention and FFN dominate. USP/SP is usually valuable.
+- CFG enabled with positive and negative prompts: transformer forward count doubles. CFG parallel may beat USP on small/moderate sequence lengths.
+- Cross-attention: skip SP/USP when condition tokens are much smaller than latent tokens; extra all-to-all can exceed compute saved.
+- HSDP/FSDP: use to fit memory first. If memory allows, compare without HSDP because all-gathers and lazy init add overhead.
+- VAE parallel: useful when encode/decode is material. Check gather/merge/broadcast and whether every rank needs final decoded output.
+
+Candidate matrices:
+
+- 2 cards: `USP=2`, `CFG=2`, HSDP on/off, VAE PP on/off.
+- 4 cards: `USP=4`, `CFG=2 x USP=2`, `USP=2 x HSDP`, VAE PP on/off.
+- 8 cards: add pipeline/stage parallel only after measuring orchestration and inter-stage transfer overhead.
+
+## Host And Free Bubble Analysis
+
+Look for device intervals where no kernel, memcpy, or memset is active.
+
+Common causes:
+
+- First request FSDP/HSDP lazy init.
+- Python preprocessing: image load, resize, `to_tensor`, tokenizer, request marshaling.
+- `torch.cuda.empty_cache()` or explicit synchronizations.
+- Scheduler small linear algebra such as tiny `torch.linalg.solve`.
+- VAE tile split/merge Python loops.
+- Profiler overhead such as CUPTI command buffer flush/full.
+
+Also inspect trace-external time:
+
+- API request parsing and multipart upload.
+- Engine/orchestrator dispatch.
+- Worker RPC and queue wait.
+- Output serialization, video write, base64, HTTP response.
+
+## Operator Optimization Candidates
+
+Evaluate necessity before implementation:
+
+- **FA to LA / LA for head:** high potential only if attention is the main CUDA hotspot and target shapes fit LA well.
+- **Cross-attention BSND/no transpose:** do only if transpose/copy around cross-attn is measurable.
+- **AdaLayerNorm fusion:** fuse norm, scale/shift, gate, residual, and dtype conversion if repeated elementwise kernels are significant.
+- **RMSNorm fusion:** useful for Q/K norms and VAE norms when small-kernel launch count is high.
+- **FP32 LayerNorm to bf16:** quality-sensitive; run visual regression.
+- **LA preprocess:** integrate only if preprocess overhead is cached or amortized.
+- **RoPE move/cache/fuse:** useful when RoPE small kernels are visible around every self-attn block.
+- **VAE RMSNorm replacement:** only after VAE internal trace shows norm as a real component of decode/encode.
+- **Scheduler fusion/cache:** cache coefficient solves and fuse CFG combine with scheduler arithmetic when possible.
+
+Before writing a custom kernel, check:
+
+- Local vLLM Omni implementations.
+- Upstream vLLM kernels and attention backends.
+- SGLang, FlashInfer, FlashAttention, and Triton examples.
+- Shape constraints and layout expectations.
+
+## Quantization And Sparsity
+
+Treat these as high-risk optimizations:
+
+- Rainfusion/sparse diffusion can be high reward but needs model-specific quality validation.
+- VAE decode quantization can reduce decode time but usually has smaller end-to-end upside than transformer attention/FFN work.
+- FP8 or approximate kernels require seed-stable visual regression, temporal flicker checks, and prompt diversity.
+
+## Output Format For Recommendations
+
+Return a table with:
+
+- Priority: P0/P1/P2.
+- Optimization.
+- Evidence from trace or code.
+- Expected benefit.
+- Risk/quality impact.
+- Implementation path.
+- Validation command or A/B test.
+
+State clearly when a recommendation is a hypothesis requiring new trace with shapes.

.claude/skills/diffusion-perf-opt/scripts/trace_analyzer.py

@@ -0,0 +1,144 @@
+#!/usr/bin/env python3
+"""Summarize torch profiler Chrome traces for vLLM Omni diffusion optimization."""
+
+from __future__ import annotations
+
+import argparse
+import collections
+import gzip
+import json
+from pathlib import Path
+from typing import Any
+
+GPU_CATS = {"kernel", "gpu_memcpy", "gpu_memset"}
+CPU_CATS = {"python_function", "user_annotation", "cpu_op", "cuda_runtime", "cuda_driver"}
+
+
+def open_trace(path: Path):
+    if path.suffix == ".gz":
+        return gzip.open(path, "rt")
+    return path.open("rt")
+
+
+def event_name(event: dict[str, Any]) -> str:
+    return str(event.get("name", ""))
+
+
+def summarize_trace(path: Path, min_gap_us: float, topn: int) -> None:
+    with open_trace(path) as f:
+        data = json.load(f)
+
+    events = data if isinstance(data, list) else data.get("traceEvents", [])
+    gpu: list[tuple[float, float, float, str, str, Any, Any]] = []
+    cpu: list[tuple[float, float, float, str, str, Any, Any]] = []
+    by_gpu_name: dict[str, list[float]] = collections.defaultdict(lambda: [0, 0.0, 0.0])
+    by_name: dict[str, list[float]] = collections.defaultdict(lambda: [0, 0.0, 0.0])
+    nccl: dict[tuple[str, str], list[float]] = collections.defaultdict(lambda: [0, 0.0, 0.0])
+
+    for event in events:
+        dur = event.get("dur")
+        ts = event.get("ts")
+        if dur is None or ts is None or dur <= 0:
+            continue
+        cat = str(event.get("cat", ""))
+        name = event_name(event)
+        row = (float(ts), float(ts + dur), float(dur), name, cat, event.get("pid"), event.get("tid"))
+        stat = by_name[name]
+        stat[0] += 1
+        stat[1] += dur
+        stat[2] = max(stat[2], dur)
+        if cat in GPU_CATS:
+            gpu.append(row)
+            gstat = by_gpu_name[name]
+            gstat[0] += 1
+            gstat[1] += dur
+            gstat[2] = max(gstat[2], dur)
+        elif cat in CPU_CATS:
+            cpu.append(row)
+        if "nccl" in name.lower():
+            nstat = nccl[(cat, name)]
+            nstat[0] += 1
+            nstat[1] += dur
+            nstat[2] = max(nstat[2], dur)
+
+    print(f"\n== {path}")
+    print(f"events={len(events)} gpu_events={len(gpu)} cpu_events={len(cpu)}")
+    if not gpu:
+        print("No GPU events found.")
+        return
+
+    gpu.sort()
+    merged: list[list[Any]] = []
+    for start, end, dur, name, cat, pid, tid in gpu:
+        if not merged or start > merged[-1][1]:
+            merged.append([start, end, [(start, end, dur, name, cat, pid, tid)]])
+        else:
+            merged[-1][1] = max(merged[-1][1], end)
+            merged[-1][2].append((start, end, dur, name, cat, pid, tid))
+
+    span = max(end for _, end, *_ in gpu) - min(start for start, *_ in gpu)
+    busy = sum(end - start for start, end, _ in merged)
+    idle = span - busy
+    print(
+        f"gpu_span_s={span / 1e6:.3f} "
+        f"busy_union_s={busy / 1e6:.3f} "
+        f"idle_union_s={idle / 1e6:.3f} "
+        f"idle_pct={idle / span * 100:.2f}"
+    )
+
+    interesting_cpu = [
+        row
+        for row in cpu
+        if row[2] >= 1000
+        and (
+            row[4] in {"python_function", "user_annotation"}
+            or "cudaStreamSynchronize" in row[3]
+            or "cudaDeviceSynchronize" in row[3]
+            or "cudaLaunch" in row[3]
+            or "cudaMemcpy" in row[3]
+        )
+    ]
+
+    gaps = []
+    for idx in range(1, len(merged)):
+        gap_start = merged[idx - 1][1]
+        gap_end = merged[idx][0]
+        gap_dur = gap_end - gap_start
+        if gap_dur >= min_gap_us:
+            prev_event = max(merged[idx - 1][2], key=lambda x: x[1])
+            next_event = min(merged[idx][2], key=lambda x: x[0])
+            mid = (gap_start + gap_end) / 2
+            containers = [row for row in interesting_cpu if row[0] <= mid <= row[1]]
+            containers = sorted(containers, key=lambda x: x[2])[:8]
+            gaps.append((gap_dur, gap_start, gap_end, prev_event, next_event, containers))
+
+    print(f"gaps_ge_{min_gap_us / 1000:.3f}ms count={len(gaps)} sum_s={sum(g[0] for g in gaps) / 1e6:.3f}")
+    for gap_dur, gap_start, gap_end, prev_event, next_event, containers in sorted(gaps, reverse=True)[:topn]:
+        print(f"\nGAP {gap_dur / 1000:.3f} ms ts={gap_start:.0f}->{gap_end:.0f}")
+        print(f"  prev {prev_event[4]} {prev_event[2] / 1000:.3f} ms {prev_event[3][:160]}")
+        print(f"  next {next_event[4]} {next_event[2] / 1000:.3f} ms {next_event[3][:160]}")
+        for row in containers:
+            print(f"  in   {row[4]} {row[2] / 1000:.3f} ms {row[3][:180]}")
+
+    print("\nTop GPU/operator events by total duration:")
+    for name, (count, total, max_dur) in sorted(by_gpu_name.items(), key=lambda kv: kv[1][1], reverse=True)[:topn]:
+        print(f"  {int(count):8d} total={total / 1e6:9.3f}s max={max_dur / 1000:9.3f}ms {name[:180]}")
+
+    print("\nTop NCCL-like events by category:")
+    for (cat, name), (count, total, max_dur) in sorted(nccl.items(), key=lambda kv: kv[1][1], reverse=True)[:topn]:
+        print(f"  {int(count):8d} total={total / 1e6:9.3f}s max={max_dur / 1000:9.3f}ms cat={cat} {name[:160]}")
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument("traces", nargs="+", type=Path, help="trace.json or trace.json.gz files")
+    parser.add_argument("--min-gap-ms", type=float, default=5.0, help="minimum GPU idle gap to print")
+    parser.add_argument("--topn", type=int, default=20, help="number of gaps/hotspots to print")
+    args = parser.parse_args()
+
+    for trace in args.traces:
+        summarize_trace(trace, args.min_gap_ms * 1000.0, args.topn)
+
+
+if __name__ == "__main__":
+    main()

.claude/skills/precheck-pr/SKILL.md

@@ -0,0 +1,100 @@
+---
+name: precheck-pr
+description: Self-check your branch before creating a PR — catch dead code, verify accuracy/perf claims, validate PR title format, and confirm merge readiness. Use when the user says "precheck", "self review", "pre-submit check", or "check my PR before I open it." Never posts to GitHub.
+---
+
+# PR Pre-Check
+
+Self-review your branch before creating a PR against `vllm-project/vllm-omni`. Two modes: **quick** catches showstoppers, **full** does a thorough maintainer-grade review. Never posts to GitHub; the report is for the contributor's terminal only.
+
+## Mode Selection
+
+| Mode | When | Time |
+|------|------|------|
+| **Quick** | About to push, final sanity check | ~3 min |
+| **Full** | Ready for review, want maintainer-level scan | ~10 min |
+
+Default to quick if unsure. Run full before marking a PR "ready for review."
+
+## Workflow
+
+### Step 1: Detect Base Branch
+
+```bash
+BASE_SHA=$(git merge-base HEAD origin/main 2>/dev/null \
+         || git merge-base HEAD main 2>/dev/null \
+         || echo origin/main)
+echo "diffing against ${BASE_SHA}"
+git diff --name-only ${BASE_SHA}...HEAD
+```
+
+### Step 2: Validate PR Title
+
+Check the most recent commit message (or branch name if no commit yet) against the project convention. Valid prefixes:
+
+| Prefix | Applies to |
+|--------|-----------|
+| `[Bugfix]` | Bug fixes |
+| `[CI/Build]` | Build or CI improvements |
+| `[Doc]` | Documentation changes |
+| `[Model]` | New/improved models (include model name) |
+| `[Frontend]` | Frontend changes (API server, OmniLLM class, etc.) |
+| `[Kernel]` | CUDA/kernel changes |
+| `[Core]` | Core logic changes (OmniProcessor, OmniARScheduler, etc.) |
+| `[Hardware][Vendor]` | Hardware-specific (e.g., `[Hardware][Ascend]`) |
+| `[Misc]` | Other changes (use sparingly) |
+
+✗ if: missing prefix, wrong case (`[bugfix]`), or WIP/Draft in title.
+⚠ if: `[Model]` prefix without the model identifier (e.g., `[Model] Add new model` — should be `[Model] Add <ModelName> ...`).
+
+### Step 3: Categorize the PR
+
+| Diff contains | PR type |
+|---------------|---------|
+| New files under `vllm_omni/model_executor/models/<name>/` | **New Model** |
+| Changes to `vllm_omni/diffusion/` | **Diffusion Model** |
+| `[Bugfix]` prefix or single-file fix | **Bug Fix** |
+| Perf/benchmark/throughput claims in commit msg or diff | **Performance** |
+| Everything else | **General** |
+
+If multiple rows apply (e.g., a diffusion model is also a new model), union the checklists.
+
+### Step 4: Run Checklist
+
+Ask: "Quick mode or full mode?" Then walk the checklist for the detected PR type from [references/checklists.md](references/checklists.md). Each item produces ✓, ✗, or ⚠.
+
+**Also run the Code-Quality sweep on every PR**, regardless of type or mode: the five diff-scoped checks in [references/code-quality.md](references/code-quality.md) — kwargs fragility, broad-except swallow, `Any`/wrong type hints, hot-path `.clone()`/`deepcopy`, and event-loop blocking — plus the advisory conventions (log level, structured logging, synchronization, cleanup, dependencies, naming) in [checklists.md](references/checklists.md). These count only lines the PR *adds* — the pre-existing backlog across the repo is out of scope.
+
+### Step 5: Print Report
+
+```
+Pre-check report for <branch>
+
+  Mode: quick | full
+  Type: <new-model | diffusion-model | bug-fix | perf | general>
+
+  Dimension          Result
+  ─────────────────  ──────
+  PR title format    ✓
+  Code quality       ⚠ 1 broad except, 2 Any hints
+  PR desc integrity  ✓
+  Registry/config    ✓
+  Dead code          ⚠ 2 warnings
+  Accuracy           ✓
+  Benchmark          ✗ missing software versions
+
+  Verdict: 1 blocking | 2 warnings | recommend fixing ✗ before PR
+```
+
+**Severity:**
+
+| Mark | Meaning |
+|------|---------|
+| ✗ | Blocking — fix before opening PR |
+| ⚠ | Warning — consider fixing |
+| ✓ | Pass |
+| — | Skipped (not applicable) |
+
+### Stop Here
+
+Do not post comments, open PRs, or modify files. The report is for the contributor's terminal only.

.claude/skills/precheck-pr/references/checklists.md

@@ -0,0 +1,160 @@
+# Pre-Check Checklists
+
+Each item is a concrete check to run against the diff. ✗ = fix before PR. ⚠ = consider fixing.
+
+---
+
+## All PRs: Title Format (check first)
+
+PR title must follow the project convention documented in `docs/contributing/README.md`:
+
+| Prefix | Applies to |
+|--------|-----------|
+| `[Bugfix]` | Bug fixes |
+| `[CI/Build]` | Build or CI improvements |
+| `[Doc]` | Documentation changes |
+| `[Model]` | New/improved models (include model name) |
+| `[Frontend]` | Frontend changes (API server, OmniLLM class, etc.) |
+| `[Kernel]` | CUDA/kernel changes |
+| `[Core]` | Core logic changes (OmniProcessor, OmniARScheduler, etc.) |
+| `[Hardware][Vendor]` | Hardware-specific (e.g., `[Hardware][Ascend]`) |
+| `[Misc]` | Other changes (use sparingly) |
+
+**Examples:**
+```
+✓ [Model] Add MiniCPM-o 4.5 support
+✓ [Bugfix] Fix Ovis image text encoder dtype
+✓ [Core] Qwen3-Omni performance optimization
+✓ [Hardware][Ascend] Fix VoxCPM2 on Ascend
+✗ Fix Ovis image text encoder dtype              ← missing prefix
+✗ [bugfix] Fix Ovis image text encoder           ← wrong case
+✗ [WIP] Add model                                ← WIP in title
+```
+
+- [ ] **Title has valid prefix** — ✗ if missing prefix, wrong case, or WIP/Draft in title. For `[Model]` PRs, the title MUST also include the model identifier (e.g., `[Model] Add <ModelName> ...`). ⚠ if missing the model name.
+
+---
+
+## All PRs: Code Quality (run on every PR, quick & full)
+
+Diff-scoped sweep for five fragility patterns that are pervasive in the existing codebase — **only lines the PR adds count**, not the pre-existing backlog. Detection commands, codebase examples, and full severity rules live in [code-quality.md](code-quality.md).
+
+- [ ] **No new `**kwargs` string-lookup plumbing:** the diff adds no `**kwargs` + `kwargs["..."]` / `kwargs.get("...")` / `"..." in kwargs`. ⚠ any new instance; ✗ if a new string key is duplicated across ≥2 files without a shared constant, or unknown keys are silently dropped on a fail-fast path.
+- [ ] **No new broad exception swallow:** the diff adds no `except Exception:` / bare `except:`. ⚠ any new instance; ✗ for `except: pass` / `except Exception: return None` / `continue` on a fail-fast path (init, config validation, weight loading, request handling).
+- [ ] **No new `Any` / wrong type hints:** the diff adds no `: Any` / `-> Any` / untyped production signature, and no new `SimpleNamespace` in tests faking an object that already has a real typed stub. ⚠ any new instance; ✗ for a *wrong* (actively misleading) annotation or a `SimpleNamespace` test fake of a typed object.
+- [ ] **No new hot-path copy:** the diff adds no `.clone()` / `.copy_()` / `copy.deepcopy(...)` inside a per-step loop or per-request path. ⚠ any new instance; ✗ for `.clone()` of a full latent/activations tensor inside an AR or diffusion step loop without a comment, or `deepcopy` of a scheduler / sampling-config on the request path.
+- [ ] **No new event-loop blocking:** the diff adds no `time.sleep` / blocking HTTP (`requests`, `urllib`) / `await` inside a held lock in `async def` code. ⚠ any new instance; ✗ for blocking sleep or HTTP on the serving path (`engine/`, `entrypoints/`, async `connectors/`), or a lock held across an `await`. Blocking inside a dedicated worker thread is fine.
+
+Roll the ⚠/✗ counts into the report as a single **Code quality** dimension row.
+
+### Conventions (eyeball — no grep)
+
+Judgment calls that don't grep cleanly — apply them while reading the diff, not as a mechanical check:
+
+- **Right log level** — `info` is for end-user-visible events; routine internal events are `debug`; hot paths are `trace` or removed. Logging is not free.
+- **Structured fields over f-strings** — pass dynamic values as structured fields (`logger.info("msg", extra={...})`), not interpolated strings, so they stay greppable and parseable.
+- **No reflexive synchronization** — don't add `threading.Lock` / `asyncio.Lock` / `deepcopy` / re-wrap shared state unless there is real concurrent mutation; the GIL covers most Python code. Owners decide their own synchronization — don't pre-lock in a constructor.
+- **Context managers for cleanup** — use `with` / `contextlib` for files, locks, CUDA streams, and profilers; not raw `open()` + manual `.close()` or ad-hoc `finally`.
+- **No new dependencies without justification** — prefer the stdlib and existing vllm-omni deps; a new top-level dep needs a reason (check `pyproject.toml` / `requirements*.txt`).
+- **Naming / interface discipline** — don't rename or change interfaces just for taste in this PR, but do fix names that actively mislead (e.g. `serve` implying a long-running server, `Instance` where multiple instance kinds coexist).
+
+---
+
+## Bug Fix PRs
+
+### Quick
+
+- [ ] **PR body has 4 sections:** what broke (error message), repro steps, root cause, fix description
+- [ ] **Fix is minimal:** diff touches only files directly related to the bug. No unrelated refactors or style changes.
+- [ ] **Repro is runnable:** the PR body includes a copy-paste command that reproduces the bug
+- [ ] **Branch is rebased:** `git merge-base HEAD origin/main` is recent, no merge conflicts
+
+### Full (adds)
+
+- [ ] **Regression test exists:** `git diff --name-only` includes at least one `tests/` file
+- [ ] **Regression test reproduces the original error:** the test would fail on main, pass on this branch
+- [ ] **Fix matches root cause exactly:** no "fix the symptom + something else" — if the root cause is dtype, the fix is dtype, not dtype + reformatting
+- [ ] **No silent failure risk:** no bare `except: pass`, no `try: ... except: return None`, no empty fallback added. See the broad-except pattern in [code-quality.md](code-quality.md#2-broad-exception-swallow) for the full severity rule (⚠ any new broad catch; ✗ for a swallow on a fail-fast path).
+- [ ] **Upstream pattern match:** the fix follows the same pattern as existing code for similar cases (grep for analogous `from_pretrained` calls, etc.)
+- [ ] **Environment documented:** torch/transformers/vllm versions listed if the bug is version-dependent
+
+---
+
+## Performance PRs
+
+### Quick
+
+- [ ] **Before/after numbers in PR body:** at minimum, one metric (RTF, RPS, latency, VRAM) with main vs PR values
+- [ ] **Hardware specified:** GPU model, count, VRAM. "L4" or "H20" is sufficient. "tested on GPU" is not.
+- [ ] **Concurrency scaling:** if the optimization claims better throughput, numbers at 2+ concurrency levels
+- [ ] **No unexplained regressions:** if any metric goes down at any concurrency level, explain why
+- [ ] **Branch is rebased:** `git merge-base HEAD origin/main` is recent, no merge conflicts
+
+### Full (adds)
+
+- [ ] **Software versions:** torch, CUDA, vllm, vllm-omni versions in PR body
+- [ ] **Warmup stated:** "1 warmup + 3 measured" or equivalent
+- [ ] **Stddev or range:** not just a single best run. "average of 2 rounds" is minimum.
+- [ ] **VRAM measured:** peak GPU memory before and after
+- [ ] **Benchmark script checked in:** under `tests/` or `examples/`, runnable by reviewer
+- [ ] **Exact command line in PR:** not pseudocode, not "run the benchmark"
+- [ ] **Pytest summary included:** `36 passed, 3 skipped, 17 warnings in 2897.66s` — proves the whole suite passes
+- [ ] **Config changes documented:** if `max_num_seqs`, `gpu_memory_utilization`, or deploy YAML changed, old and new values stated with reasoning
+
+---
+
+## New Model PRs
+
+### Quick
+
+- [ ] **PR body file list matches `git diff --name-only`:** every file path in the description exists in the diff
+- [ ] **Registry entries match claimed count:** count `_OMNI_MODELS` or `_DIFFUSION_MODELS` entries added vs claimed in body
+- [ ] **Pipeline config `custom_process_input_func` paths resolve:** each referenced module.function exists
+- [ ] **`__init__.py` exports match registry:** each registered class appears in `__init__.py` and `__all__`
+- [ ] **Shell script modes match README:** `run_curl.sh` supports every mode the README documents
+- [ ] **Branch is rebased:** `git merge-base HEAD origin/main` is recent, no merge conflicts
+
+### Full (adds)
+
+- [ ] **Dead code scan (5 patterns):**
+  1. Dead `forward()` — defined but never called through the pipeline path
+  2. Dead factory/builder functions — instantiated nowhere
+  3. Dead wrappers — thin wrappers with no callers
+  4. Dead branch guards — `if version > X` where X is always satisfied
+  5. Unused parameters — `__init__` args never accessed
+- [ ] **Copy-paste detection:** no string constant defined in 3+ files, no cross-module validation duplication, no near-identical shape coercion fns
+- [ ] **Import hygiene:** no re-export-only imports, no module-level side effects without docs, no `import os` in function body when already at top. Function-body imports are acceptable when the import is only reachable from a closure or factory return value (i.e., `import` would otherwise execute at module load even though the user may never trigger that path). Either lift to top-of-file or annotate with `# noqa: PLC0415`.
+- [ ] **Accuracy:** at least one test compares output against reference; audio = valid WAV at correct sample rate; determinism verified if seed claimed. ⚠ if missing and not explicitly deferred to a follow-up PR or RFC section; — if explicitly deferred (the deferral target SHOULD be linked in the PR body).
+- [ ] **Performance:** hardware, software versions, warmup, and metrics (RTF, VRAM, RPS, latency) stated. ⚠ if missing and not explicitly deferred; — if explicitly deferred.
+- [ ] **Benchmark settings:** model config (TP, PP, max_model_len, enforce_eager, quant), runtime config (batch, input/output spec), environment versions
+- [ ] **Benchmark script checked in + exact command line + pytest summary**
+
+---
+
+## Diffusion Model PRs
+
+Diffusion models live under `vllm_omni/diffusion/`. In addition to the New Model checklist above, verify:
+
+- [ ] **Transformer adapter follows contract:** inherits from `nn.Module`, implements `load_weights()`, uses `vllm_omni.diffusion.attention.layer.Attention`
+- [ ] **Pipeline has both offline and online paths:** `forward()` for batch generation, streaming path for online serving
+- [ ] **Cache-DiT integration:** at least one acceleration (TeaCache, FBCache, etc.) is wired up. ✗ for new **top-level** pipelines. For **subclass** pipelines that inherit `__call__` from a parent where Cache-DiT is already absent, downgrade to ⚠ and recommend filing a follow-up issue against the parent pipeline family.
+- [ ] **Parallelism config:** TP/SP/USP/CFG-Parallel options exposed in pipeline config
+- [ ] **Docs table updated:** model added to `docs/models/supported_models.md` with correct metadata
+- [ ] **E2E test exists:** at least one test exercises the full generate path
+
+---
+
+## General PRs
+
+### Quick
+
+- [ ] **PR body matches diff:** description doesn't claim changes not in the diff
+- [ ] **Branch is rebased:** `git merge-base HEAD origin/main` is recent, no merge conflicts
+- [ ] **CI gates passing:** DCO, pre-commit, build — check `gh pr view --json statusCheckRollup` if a PR already exists
+
+### Full (adds)
+
+- [ ] **No new dead code in changed files:** scan for unreachable paths, unused imports
+- [ ] **Test coverage:** if core code changed (`engine/`, `stages/`, `connectors/`), tests are added or existing tests cover the path
+- [ ] **Import hygiene:** no redundant imports added
+- [ ] **No unrelated changes:** diff scope matches PR description

.claude/skills/precheck-pr/references/code-quality.md

@@ -0,0 +1,179 @@
+# Code-Quality Patterns
+
+Five fragility patterns to sweep on **every** PR, in both quick and full mode. These are pervasive in the existing codebase (hundreds of pre-existing sites) — so the checks are **diff-scoped**: only lines the PR *adds* count. Whole-repo greps light up every PR with the backlog and are useless.
+
+All commands assume:
+```bash
+BASE="$(git merge-base HEAD origin/main)"
+```
+
+**Severity:** ⚠ for any *new* instance introduced by the diff; ✗ for the acute sub-cases defined per pattern. Acute ✗ findings belong in the report's blocking count.
+
+---
+
+## 1. `**kwargs` string-lookup passthrough
+
+### Anti-pattern
+```python
+def forward(self, **kwargs):
+    # BLOCKER-prone: raw dict + string keys duplicated across files
+    if "runtime_additional_information" in kwargs and \
+       "model_intermediate_buffer" not in kwargs:
+        ...
+```
+Real instances of this exact guard are copy-pasted across `cosyvoice3.py`, `indextts2_talker.py` (×2), `indextts2_s2mel_decoder.py`, `qwen3_tts_talker.py`, `qwen3_omni.py`. A typo in any string key silently no-ops; there is no type check, no IDE completion, and unknown keys are silently dropped.
+
+### Why dangerous
+Raw `dict` + `**kwargs` + lists of param-name strings is **fragile by construction**: the compiler can't see a missing or misspelled key, so failures are silent. Each duplicated copy is a new place for the strings to drift.
+
+### Severity
+- ⚠ any *new* `**kwargs` signature whose body does `kwargs["..."]` / `kwargs.get("...")` / `"..." in kwargs`.
+- ✗ when the diff introduces a **new** string key that is duplicated across ≥2 files without a shared module constant, or when `**kwargs` plumbing **silently drops unknown keys** on a fail-fast path (init, config validation, weight loading).
+
+### Detect (diff-scoped)
+```bash
+git diff ${BASE}...HEAD -- "*.py" | grep "^+[^+]" \
+  | grep -E '\*\*kwargs|kwargs(\.get|\[)|"[a-z_]+"\s+in\s+kwargs'
+```
+
+### Fix
+Prefer explicit typed params. The repo already has the right primitive: `msgspec.Struct` with `forbid_unknown_fields=True` rejects unknown keys instead of silently dropping them — see `_StructBase` in `vllm_omni/data_entry_keys.py:105` and `MoriPullRequest` in `vllm_omni/distributed/omni_connectors/connectors/mori_transfer_engine_connector.py:81`. A `@dataclass` or `TypedDict` works too.
+
+If `**kwargs` is unavoidable (vLLM base-class compat), centralize the key strings as module-level constants behind **one** typed accessor, instead of repeating the literal across files.
+
+---
+
+## 2. Broad exception swallow
+
+### Anti-pattern
+```python
+# BLOCKER-prone: catches everything, returns None on any failure
+try:
+    evt = build_event(...)
+    return evt
+except Exception:
+    return None
+```
+Real clusters: `vllm_omni/metrics/stats.py:240,274,367,450` and `vllm_omni/metrics/{modality,utils}.py` — all `except Exception: return None`.
+
+### Why dangerous
+A generic `except Exception:` is **too flexible**. If we catch an exception type we don't expect, it is almost always an unhandled edge case that needs to be *fixed* — not swallowed into `None`/`pass`/`continue`. Broad catches turn fixable bugs into silent wrong behavior and make debugging impossible.
+
+The repo backs this with ruff `BLE001` (broad-exception catch); it is currently suppressed via `# noqa: BLE001` in `vllm_omni/patch.py:413`. Prefer fixing over suppressing.
+
+### Severity
+- ⚠ any *new* `except Exception:` / `except BaseException:` / bare `except:`.
+- ✗ for `except:` / `except Exception: pass|return None|continue` on a **fail-fast path** (init, config validation, weight loading, request handling, connector setup). This extends the Bug-Fix "No silent failure risk" ✗ in [checklists.md](checklists.md).
+
+### Detect (diff-scoped)
+```bash
+# Any new broad catch (⚠ baseline):
+git diff ${BASE}...HEAD -- "*.py" | grep "^+[^+]" \
+  | grep -E 'except\s*(Exception|BaseException)?\s*:'
+# Note: this matches only the immediate-colon swallow forms (`except :`,
+# `except Exception:`) — it deliberately excludes specific catches like
+# `except ValueError:` and the `except Exception as e:` variant. Eyeball
+# the latter; `as e` usually means it is logged, but still verify.
+```
+
+### Fix
+Catch the **specific** types you actually expect (`ValueError`, `KeyError`, `AttributeError`, `OSError`, `torch.cuda.OutOfMemoryError`, `msgspec.ValidationError`, …). At a genuine top-level / best-effort boundary (metrics, signal handlers), at minimum **log** the exception; never swallow into a bare `pass`/`return None` on a path that should fail loudly.
+
+---
+
+## 3. `Any` / wrong type annotations
+
+### Anti-pattern
+```python
+# ⚠ Any param + Any return — no contract, no narrowing
+def _extract_mm_output(engine_outputs: Any) -> dict[str, Any]:
+    ...
+```
+
+### Why dangerous (root cause)
+`Any` disables the type checker for that position, so typos and wrong shapes pass silently. The repo has 636 `: Any` params, 244 `-> Any` returns, and **1,173 `SimpleNamespace` in `tests/`**. The `Any` leak is largely test-driven: vibe-coded unit tests fake objects with `SimpleNamespace`, which forces the production code consuming them to be typed `Any`, and the `Any` then propagates outward into public signatures.
+
+### Severity
+- ⚠ any *new* `: Any` / `-> Any` / fully-untyped signature in production code.
+- ✗ for a **wrong** annotation (actively misleading — e.g. `-> bool` that returns `Optional[bool]`), or a *new* `SimpleNamespace` in a test that mimics an object which already has a real typed stub / `@dataclass` / `TypedDict` / `Protocol`.
+
+### Detect (diff-scoped)
+```bash
+# Any leaking into new signatures:
+git diff ${BASE}...HEAD -- "*.py" | grep "^+[^+]" \
+  | grep -E ':\s*Any\b|->\s*Any\b'
+# SimpleNamespace newly added in tests (the Any-leak source):
+git diff ${BASE}...HEAD -- "tests/" | grep "^+[^+]" | grep "SimpleNamespace"
+```
+
+### Fix
+Replace `Any` with the concrete type, a `Protocol`, or a `Union`. If a type is genuinely dynamic, prefer `object` + `isinstance` narrowing over `Any` — `object` keeps the checker engaged. In tests, replace `SimpleNamespace` with the real class, a small `@dataclass`/`TypedDict`, or a `Protocol` + fake, so the system under test stays fully typed.
+
+---
+
+## 4. Unnecessary copy — `.clone()` / `copy.deepcopy` in hot paths
+
+### Anti-pattern
+```python
+# Per-step in a diffusion/AR loop: clones the full latent every iteration
+for t in timesteps:
+    trajectory_latents.append(x_t.clone())   # O(N) GPU alloc + memcpy each step
+```
+Real clusters: `vllm_omni/diffusion/models/bagel/bagel_transformer.py` calls `x_t.clone()` ~9× across a trajectory loop (lines 1820–2186); the cache hooks `vllm_omni/diffusion/cache/teacache/hook.py:155` and `magcache/hook.py:182` clone `hidden_states` per step. Separately, `copy.deepcopy(self.scheduler)` / `copy.deepcopy(req.sampling_params)` is copy-pasted across diffusion pipelines — `pipeline_qwen_image.py:786`, `pipeline_hunyuan_image3.py:1904`, `pipeline_ltx2*.py` (×3), `pipeline_dreamzero.py:1085`, `diffusion_model_runner.py:398` — a full object-graph deepcopy on the request path. (Contrast `dreamzero/causal_wan_model.py:203`, which carries a `# No .clone() needed.` comment — the aware pattern.)
+
+### Why dangerous
+A `.clone()` of a GPU tensor or a `copy.deepcopy` of a config object inside a per-step / per-request path is a hidden allocator: it reads as one Python line but triggers a GPU memcpy or a deep object-graph walk every iteration. Across 50 diffusion steps or hundreds of AR steps it dominates latency and VRAM bandwidth. `deepcopy` additionally ignores any custom device / `__copy__` logic and can drag tensors through CPU.
+
+### Severity
+- ⚠ any *new* `.clone()` / `.copy_()` / `copy.deepcopy(...)` inside a per-step loop or per-request path.
+- ✗ for `.clone()` of a full latent/activations tensor inside an AR or diffusion step loop without an explanatory comment, or `copy.deepcopy` of a scheduler / sampling-config on the request hot path.
+
+### Detect (diff-scoped)
+```bash
+git diff ${BASE}...HEAD -- "*.py" | grep "^+[^+]" \
+  | grep -E '\.clone\(\)|\.copy_\(|copy\.deepcopy|deepcopy\('
+```
+
+### Fix
+Tensors: prefer a view (`x[..., :k]`, `torch.narrow`, `as_strided`) or write into a pre-allocated buffer with `dst.copy_(src)` instead of `.clone()`; if the caller no longer needs the original, `move`/`del` it rather than copy. Schedulers / configs: build a fresh lightweight instance per request, or split out the mutable per-request bits so the shared immutable part isn't deepcopied. If a copy is genuinely required (e.g. the source is mutated in place later), keep it and add a one-line comment saying why.
+
+---
+
+## 5. Blocking the asyncio event loop / lock held across `await`
+
+### Anti-pattern
+```python
+# BLOCKER-prone: blocking sleep / blocking HTTP / lock held across an await,
+# all inside an async (event-loop) function
+async def handle(self, req):
+    with self._lock:
+        resp = await self._client.call(req)   # lock held across await -> serializes all requests
+    time.sleep(0.1)                           # stalls the single loop for every concurrent request
+```
+
+### Why dangerous
+vLLM-Omni serves over a single asyncio event loop. Any synchronous blocking work on an `async def` path — `time.sleep`, blocking HTTP (`requests`/`urllib`), heavy CPU work, or a lock held across an `await` — stalls the loop for *every* concurrent request, not just the blocking one. A lock held across `await` silently serializes the whole pipeline; blocking I/O tanks throughput and inflates tail latency.
+
+### Severity
+- ⚠ any *new* `time.sleep(`, blocking HTTP call (`requests.get/post`, `urllib.request`, `urlopen`), or `await` inside a `with Lock` / `async with` critical section, in `async def` code.
+- ✗ for `time.sleep` or blocking HTTP on the serving path (`engine/`, `entrypoints/`, async `connectors/`), or a lock acquired and held across an `await` on the request path.
+
+### Detect (diff-scoped)
+```bash
+# New blocking primitives introduced by the diff — then eyeball whether each
+# sits in an async function on the loop vs a dedicated worker thread:
+git diff ${BASE}...HEAD -- "*.py" | grep "^+[^+]" \
+  | grep -E 'time\.sleep|requests\.(get|post|put|delete)|urllib\.request|urlopen'
+# Lock acquired with an await potentially inside the critical section (manual eyeball):
+git diff ${BASE}...HEAD -- "*.py" | grep "^+[^+]" | grep -E 'async with|\.acquire\(\)'
+```
+Note: blocking inside a **dedicated worker thread is correct** — the repo already does this (`diffusion/model_loader/hub_prefetch.py`, `distributed/omni_connectors/kv_transfer_manager.py`, the `mooncake`/`mori` connectors poll with `time.sleep` on their own threads). The check targets the *event-loop* path only.
+
+### Fix
+Move blocking work off the loop with `asyncio.to_thread(...)` / `run_in_executor`, or use a native async client (`httpx.AsyncClient` / `aiohttp`). Never `await` while holding a lock — drop the lock before the `await` and re-acquire after, or restructure so the critical section contains no `await`.
+
+---
+
+## Running the sweep
+
+Run all five detections against `${BASE}...HEAD`, then for each hit decide ⚠ vs ✗ using the severity rules above. Roll the results into the report as a single **Code quality** dimension row (count of ⚠ and ✗), alongside the type-specific checklist from [checklists.md](checklists.md).

.claude/skills/quantization/SKILL.md

@@ -0,0 +1,103 @@
+---
+name: quantization
+description: Work on vLLM-Omni quantization for diffusion, autoregressive, omni, or multi-stage models. Use when choosing or adding methods such as fp8, int8, gguf, mxfp8, mxfp4, mxfp4_dualscale, ModelOpt, AutoRound, INC, msModelSlim, awq, or gptq; debugging quantized loading; or validating memory, speed, and output quality.
+---
+
+# vLLM-Omni Quantization
+
+Use this skill for quantization work in `vllm-omni`. Start from the local
+code and docs, then pick the smallest validated path for the model, method,
+hardware, and quality target.
+
+## First Checks
+
+Before changing code or recommending a command, identify:
+
+- Model and task: AR text, omni/TTS, text-to-image, text-to-video, or multi-stage diffusion.
+- Hardware backend: CUDA generation, Ascend NPU, Intel XPU, or another platform.
+- Quantization mode: online from BF16/FP16 weights, pre-quantized checkpoint, runtime KV/FA quantization, or per-component routing.
+- Scope: transformer/DiT, thinker language model, text encoder, VAE, audio/vision encoder, talker, or a specific stage.
+- Validation target: memory reduction, latency/throughput, output quality, or all three.
+
+Do not assume a method is supported just because the CLI accepts the string.
+Check `docs/user_guide/quantization/` and the closest implementation first.
+
+## Quick Decision
+
+| Task | Start With |
+|------|------------|
+| Choose a method or command | `references/methods.md` and `references/modality-compat.md` |
+| Use `build_quant_config()` or per-component routing | `references/methods.md` |
+| Work on diffusion quantization | `references/diffusion.md` |
+| Add quantization to a new model | `references/adding-models.md` |
+| Convert or load ModelOpt FP8 checkpoints | `references/modelopt-fp8.md` |
+| Debug ModelOpt, GGUF, AutoRound, MXFP, or serialized Int8 loading | `references/diffusion.md` and method docs under `docs/user_guide/quantization/` |
+
+## Current Runtime Shape
+
+The unified entrypoint is:
+
+```python
+from vllm_omni.quantization import build_quant_config
+```
+
+It supports method strings, flat method dictionaries, per-component
+dictionaries, existing `QuantizationConfig` objects, and `None`. The factory
+delegates generic methods to upstream `vllm` and keeps vLLM-Omni overrides for
+diffusion or omni-specific routing:
+
+- `gguf`
+- `int8`
+- `mxfp8`
+- `mxfp4`
+- `mxfp4_dualscale`
+- `inc`, `auto-round`, `auto_round`
+- ModelOpt auto-detected configs: `modelopt`, `modelopt_fp4`, `modelopt_mixed`
+
+Use `ComponentQuantizationConfig` when only one stage or component should be
+quantized. Pre-quantized ModelOpt-style checkpoints should not spill into
+vision/audio encoders that have no corresponding scale tensors.
+
+## Ownership Boundary
+
+- Upstream `vllm` owns generic quantization configs, kernels, loader semantics,
+  hardware capability rules, and generic AR quantization methods.
+- `vllm-omni` owns unified config routing, diffusion-specific wrappers,
+  component scoping, GGUF or ModelOpt checkpoint adapters, model-specific
+  prefix mapping, docs, examples, and validation.
+
+If a new method needs missing generic kernels or loader behavior, fix upstream
+`vllm` first. In `vllm-omni`, add thin integration and model wiring.
+
+## Working Loop
+
+1. Read the local method doc under `docs/user_guide/quantization/`.
+2. Inspect the active factory and config class in `vllm_omni/quantization/`.
+3. Inspect the target model pipeline and transformer for stable prefixes and
+   whether every relevant vLLM linear layer receives `quant_config`.
+4. For pre-quantized checkpoints, inspect `transformer/config.json` and the
+   checkpoint tensor names before running full generation.
+5. Validate with the same prompt, seed, size, scheduler, steps, dtype, eager or
+   non-eager mode, and parallelism as the BF16 baseline.
+6. Report quality, latency, and memory separately. Do not call a method
+   supported until quality is checked.
+
+## Common Mistakes
+
+| Symptom | Likely Cause | Fix |
+|---------|--------------|-----|
+| `--quantization` has no visible effect | Wrong scope or unsupported model path | Check component routing and method docs |
+| Some layers stay BF16 unexpectedly | `quant_config` was not threaded into all vLLM linear layers | Audit transformer constructors and prefixes |
+| Quality collapses but loading succeeds | Too many sensitive layers were quantized | Add model-specific `ignored_layers` and compare to BF16 |
+| ModelOpt checkpoint loads but output is corrupted | Prefixes, packed-module mapping, scale routing, or BF16 fallback is wrong | Use `references/modelopt-fp8.md` |
+| GGUF shape or tensor mismatch | Missing architecture-specific adapter | Add explicit adapter mapping; avoid generic fallback |
+| Online and offline MXFP4 disagree | Wrong `mxfp4` vs `mxfp4_dualscale` mode | Check `docs/user_guide/quantization/mxfp4.md` |
+| Quantized path is slower than BF16 | Kernel path, compile mode, dtype, or shape mismatch | Compare eager-to-eager and non-eager-to-non-eager |
+
+## References
+
+- General methods and config routing: [references/methods.md](references/methods.md)
+- Modality and model scope: [references/modality-compat.md](references/modality-compat.md)
+- Diffusion implementation patterns: [references/diffusion.md](references/diffusion.md)
+- Adding model support: [references/adding-models.md](references/adding-models.md)
+- ModelOpt FP8 conversion and loading: [references/modelopt-fp8.md](references/modelopt-fp8.md)

.claude/skills/quantization/references/adding-models.md

@@ -0,0 +1,70 @@
+# Adding Quantization Support to a Model
+
+Use this checklist when extending quantization to a new architecture or stage.
+
+## Classify Ownership
+
+| Case | Owner |
+|------|-------|
+| New generic AR method, kernel, or loader semantics | upstream `vllm` first |
+| Existing upstream method needs diffusion wiring | `vllm-omni` model and config integration |
+| GGUF or ModelOpt tensor mapping for a diffusion architecture | `vllm-omni` adapter |
+| NPU MXFP or msModelSlim diffusion path | `vllm-omni` platform/method wrapper and docs |
+| Multi-stage omni checkpoint quantizes only the thinker LM | `vllm-omni` component scoping |
+
+Avoid building a private quantization stack in model code.
+
+## Model Wiring Checklist
+
+1. Find the closest supported model implementation.
+2. Confirm the model uses vLLM linear layers that accept `quant_config`.
+3. Add `quant_config` to the transformer constructor.
+4. Pass stable `prefix` values through every linear layer.
+5. Handle fused modules and packed mappings explicitly.
+6. Ensure `load_weights()` and any `WeightsMapper` remap `ignored_layers` or
+   `modules_to_not_convert`.
+7. Keep tokenizer, scheduler, VAE, text encoder, audio/vision encoders, and
+   talker BF16 unless method-specific docs validate them.
+8. Add or update a method doc under `docs/user_guide/quantization/`.
+
+## Checkpoint Adapter Checklist
+
+For pre-quantized checkpoints:
+
+- Inspect `transformer/config.json` first.
+- Check `quant_method`, `method`, `producer.name`, `quant_algo`, serialized
+  flags, `ignored_layers`, and any method-specific fields.
+- Verify tensor name mapping before generation.
+- Verify packed modules such as `to_qkv`, `add_kv_proj`, and `w13`.
+- Confirm scale tensor names and shapes.
+- Allow dequantization back to BF16 for intentionally skipped target layers.
+
+## Sensitive Layer Discovery
+
+Start conservative. Quantize the backbone first and keep output-side or routing
+layers BF16 until quality is proven.
+
+Frequently sensitive areas:
+
+- attention output projections such as `to_out`
+- FFN down projections such as `w2`
+- final projections such as `proj_out`, `final_layer`, and output heads
+- modulation, timestep, position, image, or text embedders
+- refiner blocks
+- MoE routers, experts selected by routing, and layernorm-adjacent modules
+- VAE, text/vision towers, audio encoders, and codec/talker stages
+
+Do not reuse one model family's ignore list blindly.
+
+## Tests and Evidence
+
+Minimum evidence for a PR:
+
+- loader/config test or smoke command
+- fixed-seed BF16 vs quantized output comparison
+- quality metrics and saved outputs for precision-changing work
+- memory and latency comparison under identical execution mode
+- failure behavior for unsupported models or missing checkpoint fields
+
+When the PR adds a method to a support table, include the exact model,
+hardware, command, quality threshold, and artifact path used for validation.

.claude/skills/quantization/references/diffusion.md

@@ -0,0 +1,113 @@
+# Diffusion Quantization Reference
+
+Use this file for DiT quantization, config routing, checkpoint adapters, and
+quality or loader debugging.
+
+## Supported Method Families
+
+The current local diffusion quantization families are:
+
+- CUDA/GPU online or checkpoint: `fp8`, `int8`, `gguf`, ModelOpt
+- Ascend NPU: `int8`, `mxfp8`, `mxfp4`, `mxfp4_dualscale`, msModelSlim paths
+- Intel XPU: AutoRound/INC and online MXFP8 paths where documented
+
+Always cross-check the method doc before claiming model support. The CLI
+accepting a method string is not enough.
+
+## Implementation Starting Points
+
+| Area | Files |
+|------|-------|
+| Unified factory | `vllm_omni/quantization/factory.py` |
+| Per-component routing | `vllm_omni/quantization/component_config.py` |
+| GGUF diffusion method | `vllm_omni/quantization/gguf_config.py` |
+| Int8 diffusion method | `vllm_omni/quantization/int8_config.py` |
+| MXFP8 diffusion method | `vllm_omni/quantization/mxfp8_config.py` |
+| MXFP4 and dual-scale method | `vllm_omni/quantization/mxfp4_config.py` |
+| AutoRound/INC omni extension | `vllm_omni/quantization/inc_config.py` |
+| ModelOpt checkpoint adapters | `vllm_omni/diffusion/model_loader/checkpoint_adapters/modelopt.py` |
+| FP8 scaled-mm runtime patches | `vllm_omni/patch.py` |
+| Similarity tool | `vllm_omni/quantization/tools/compare_diffusion_trajectory_similarity.py` |
+
+## Threading `quant_config`
+
+For a diffusion transformer:
+
+1. Accept `quant_config: QuantizationConfig | None` in the transformer.
+2. Pass `quant_config` and stable `prefix` values to every vLLM linear layer.
+3. Preserve packed-module mappings for fused QKV, FFN, or model-specific fused
+   projections.
+4. Call `apply_vllm_mapper()` on configs that need HF-to-vLLM name remapping.
+5. For component configs, keep prefixes specific enough to avoid quantizing VAE
+   or encoders by accident.
+
+Common skipped areas are `proj_out`, output projections, modulation layers,
+embedders, refiner blocks, MoE routers, and layernorm-adjacent paths.
+
+## Online vs Offline
+
+Online quantization:
+
+- starts from BF16/FP16 checkpoint weights
+- computes or stores quantized weights at load time
+- typically uses `--quantization <method>` or `quantization="<method>"`
+
+Offline or pre-quantized checkpoints:
+
+- include quantized weights and scales on disk
+- usually carry `quantization_config` in `transformer/config.json`
+- should auto-detect when the config is complete
+- may require model-specific adapter logic for tensor names and packed modules
+
+Do not mix online flags with incompatible offline checkpoint configs. If the
+active config and disk config differ, `resolve_quant_config_from_disk()` should
+raise or rebuild rather than silently loading corrupted weights.
+
+## GGUF Rules
+
+- Require `quantization_config.gguf_model`.
+- Use the normal base model for tokenizer, scheduler, text encoder, and VAE.
+- Add explicit architecture adapters for tensor-name mapping.
+- Use `named_parameters()` and `named_buffers()`, not `state_dict()`, to
+  discover loadable names.
+- Guard fused QKV/KV rewrites so tensors are not rewritten twice.
+- Flatten N-D activations around the GGUF matmul and restore shape.
+
+Unsupported architectures should fail clearly instead of falling back to a fake
+generic mapper.
+
+## MXFP Rules
+
+- `mxfp8` is W8A8 microscaling, currently centered on NPU and documented XPU
+  behavior.
+- `mxfp4` is online single-scale W4A4 on NPU.
+- `mxfp4_dualscale` is online/offline dual-scale W4A4 with BF16 fallback; the
+  offline path is the production path when calibrated `mul_scale` exists.
+- Offline dual-scale checkpoints must inject per-transformer `ignored_layers`
+  into config because cascade transformers can have different BF16 fallback
+  sets.
+
+Do not load an offline dual-scale checkpoint with the online single-scale path.
+
+## Common Failures
+
+| Symptom | Likely Cause | Fix |
+|---------|--------------|-----|
+| `Unknown quantization method` | Factory override or upstream registration is missing | Add the right override or upstream method |
+| Missing scale tensors | Wrong scope, wrong adapter, or BF16 layer not allowed to dequantize | Check config scope and adapter scale routing |
+| Output rank collapses from 3-D to 2-D | FP8 scaled-mm kernel ignored output shape | Check the patch in `vllm_omni/patch.py` |
+| Quality degrades silently | Sensitive layers quantized | Add `ignored_layers`; compare to BF16 |
+| Only one cascade transformer works | Disk config differs per transformer but active config is reused | Rebuild from each transformer's config |
+| Quantized path slower than BF16 | Kernel, dtype, compile, or benchmark mode mismatch | Compare eager-to-eager and non-eager-to-non-eager |
+
+## Verification
+
+For a new or changed diffusion quantization path, run:
+
+1. Config and loader smoke test.
+2. One fixed-seed generation against BF16.
+3. Numeric comparison with the similarity tool or equivalent metrics.
+4. Serving test if the path is user-facing online.
+5. Memory and latency check under the same execution mode as the baseline.
+
+Keep output artifacts when precision changes are part of the PR.

.claude/skills/quantization/references/methods.md

@@ -0,0 +1,82 @@
+# General Quantization Methods
+
+Use this file for method selection and unified `vllm-omni` config syntax. For
+diffusion-specific wiring, see `diffusion.md`. For ModelOpt checkpoint export
+and loading details, see `modelopt-fp8.md`.
+
+## Unified Entry Point
+
+`vllm_omni.quantization.build_quant_config()` accepts:
+
+- `None` or `"none"` to disable quantization
+- a method string such as `"fp8"`, `"int8"`, `"mxfp8"`, `"awq"`, or `"auto-round"`
+- a flat dict with `"method"` or `"quant_method"`
+- a per-component dict such as `{"transformer": {"method": "fp8"}, "vae": None}`
+- an already-built `QuantizationConfig`
+
+Per-component dicts are routed by `ComponentQuantizationConfig` using
+longest-prefix matching. Prefix stability matters: if a model remaps HF names
+to vLLM names, apply the mapper to `ignored_layers` or component prefixes.
+
+## Local Overrides
+
+The local factory delegates most methods to upstream `vllm`, but has vLLM-Omni
+overrides for methods that need diffusion, NPU, XPU, or component behavior:
+
+| Method | Local Config | Main Use |
+|--------|--------------|----------|
+| `gguf` | `DiffusionGGUFConfig` | Pre-quantized diffusion transformer weights |
+| `int8` | `DiffusionInt8Config` | Online or serialized W8A8 diffusion transformers |
+| `mxfp8` | `DiffusionMXFP8Config` | W8A8 MXFP8 diffusion on NPU and online XPU path |
+| `mxfp4` | `DiffusionMXFP4Config` | Online W4A4 MXFP4 diffusion on NPU |
+| `mxfp4_dualscale` | `DiffusionMXFP4DualScaleMixedConfig` | Online/offline W4A4 dual-scale plus BF16 fallback on NPU |
+| `inc`, `auto-round`, `auto_round` | `OmniINCConfig` | AutoRound/INC checkpoint loading, including MXFP8 checkpoint metadata |
+| ModelOpt config dicts | upstream ModelOpt configs through auto-detect | FP8, NVFP4, or mixed ModelOpt checkpoints |
+
+Check `SUPPORTED_QUANTIZATION_METHODS` in `vllm_omni/quantization/factory.py`
+before adding a new method name.
+
+## AR and Generic Quantization
+
+For AR-backed models, upstream `vllm` methods usually apply:
+
+- `awq` and `gptq`: 4-bit or 8-bit weight quantization with calibration,
+  commonly used for text and AR stages.
+- `fp8`: FP8 weight path or KV-cache FP8 depending on flags and hardware.
+- `inc` or `auto-round`: pre-quantized checkpoints with Intel Neural
+  Compressor / AutoRound metadata.
+
+Use these for the AR language model stage only when the checkpoint and model
+guide support it. Omni and TTS models often keep audio, vision, talker, and
+codec stages in BF16.
+
+## Diffusion and Offline Methods
+
+For diffusion transformers, prefer the method-specific docs:
+
+- `docs/user_guide/quantization/fp8.md`
+- `docs/user_guide/quantization/int8.md`
+- `docs/user_guide/quantization/gguf.md`
+- `docs/user_guide/quantization/modelopt.md`
+- `docs/user_guide/quantization/mxfp8.md`
+- `docs/user_guide/quantization/mxfp4.md`
+- `docs/user_guide/quantization/autoround.md`
+- `docs/user_guide/quantization/msmodelslim.md`
+
+Pre-quantized checkpoints should normally auto-detect from config. Runtime
+flags such as `--force-cutlass-fp8`, `--linear-backend cutlass`, and
+`--moe-backend cutlass` select kernels; they do not quantize a BF16 checkpoint.
+
+## Validation
+
+Use fixed-seed A/B comparisons and record:
+
+- model path and quantized checkpoint path
+- method config and `ignored_layers`
+- prompt, size, frame count, steps, scheduler, dtype, and seed
+- eager/non-eager mode, compile flags, and parallelism
+- output metrics such as PSNR, MAE, cosine similarity, SSIM, LPIPS, plus visual artifacts
+- latency and peak memory, reported separately from quality
+
+The helper `vllm_omni.quantization.tools.compare_diffusion_trajectory_similarity`
+is the default lightweight image/video comparison tool for diffusion paths.

.claude/skills/quantization/references/modality-compat.md

@@ -0,0 +1,47 @@
+# Quantization Compatibility by Modality
+
+Use this as a routing guide. The source of truth is the local method docs under
+`docs/user_guide/quantization/` and the current model implementation.
+
+## Summary
+
+- AR and omni language-model quantization mostly follows upstream `vllm`, with
+  vLLM-Omni scoping for multi-stage models.
+- Diffusion quantization is method-specific and model-specific. It usually
+  targets the diffusion transformer only.
+- Pre-quantized checkpoints should quantize only the component that has
+  quantized weights and scale tensors.
+
+## Model Type Matrix
+
+| Model Type | Typical Quantization Scope | Notes |
+|------------|----------------------------|-------|
+| AR text models | Language model | Upstream `vllm` methods usually apply |
+| Omni/TTS models | Thinker or AR language-model stage | Audio, vision, talker, and codec stages usually stay BF16 |
+| DiT image models | Diffusion transformer | Check FP8, Int8, GGUF, ModelOpt, AutoRound docs |
+| DiT video models | Diffusion transformer | HunyuanVideo FP8 and Wan2.2 MXFP paths are documented; validate other video paths |
+| Multi-stage diffusion | Selected stage only | Use per-component routing and per-stage validation |
+| Audio diffusion | Usually unsupported | Prefer dtype, offload, cache, or parallelism unless docs say otherwise |
+
+## Method Hints
+
+| Goal | Candidate |
+|------|-----------|
+| AR memory reduction | AWQ, GPTQ, AutoRound, ModelOpt where checkpoint supports it |
+| CUDA diffusion W8A8 | FP8 or Int8 |
+| Pre-quantized CUDA diffusion checkpoint | ModelOpt or GGUF |
+| Ascend NPU diffusion | Int8, MXFP8, MXFP4, MXFP4 DualScale, msModelSlim |
+| Intel XPU checkpoint | AutoRound/INC paths |
+| Conservative image quality | selective FP8/Int8 with `ignored_layers` |
+| Unsupported diffusion model | cache acceleration, CPU/layerwise offload, dtype, TP/SP/CFG parallelism |
+
+## Validation Rule
+
+Compatibility means more than successful loading. A method is compatible only
+after it has:
+
+- correct component scope
+- complete checkpoint/config metadata if offline
+- fixed-seed quality comparison against BF16
+- acceptable latency and memory behavior
+- docs or examples that state model, method, hardware, and caveats

.claude/skills/quantization/references/modelopt-fp8.md

@@ -0,0 +1,122 @@
+# ModelOpt FP8 Conversion and Loading
+
+Use this when exporting or loading ModelOpt FP8 checkpoints for diffusion
+transformers. This reference captures the practical conversion lessons from the
+2026-04-21 ModelOpt FP8 paste and aligns them with the current
+`vllm-omni` ModelOpt adapter.
+
+## Core Rule
+
+Treat ModelOpt FP8 as selective checkpoint quantization, not one-click
+whole-model FP8. Quantize the main transformer or DiT backbone, keep sensitive
+or peripheral modules BF16, serialize a complete ModelOpt checkpoint, and let
+vLLM-Omni auto-detect it from config.
+
+Do not pass `--quantization fp8` to force a ModelOpt checkpoint path. The
+checkpoint's `quantization_config` should select `modelopt`.
+
+## Conversion Strategy
+
+Prefer:
+
+- backbone attention Q/K/V and FFN expansion projections
+- conservative per-model ignore lists
+- MHA quantizers disabled until basic linear quantization is stable
+- full module-name prefixes in ignore/routing logic
+- BF16 fallback for output-side and routing-sensitive layers
+
+Avoid quantizing by default:
+
+- VAE
+- text, vision, or audio encoders
+- embedders and modulation layers
+- final heads and output projections
+- refiner modules
+- MoE routers and layernorm-sensitive blocks
+
+## Model-Specific Starting Points
+
+| Model Family | Quantize First | Keep BF16 First |
+|--------------|----------------|-----------------|
+| Z-Image | attention q/k/v, FFN w1/w3 | `to_out`, `w2`, `noise_refiner`, `context_refiner`, embedders, modulation, final/output |
+| Qwen-Image | transformer backbone linear layers | `proj_out`, `img_in`, `txt_in`, norms, modulation, position/time/embed layers |
+| FLUX2-dev / FLUX2-klein | backbone linear layers | `proj_out`, context/x embedders, `norm_out`, stream modulation, time embeddings |
+| HunyuanImage-3.0 | `model.model` backbone | VAE, vision tower, token/time/image embedders, final image head, LM head, routers, layernorm-sensitive blocks |
+
+These are starting points, not support claims. Validate with BF16 outputs.
+
+## Required Checkpoint Metadata
+
+`transformer/config.json` must include a complete `quantization_config`.
+At minimum it should identify ModelOpt and the algorithm:
+
+```json
+{
+  "quantization_config": {
+    "producer": {"name": "modelopt"},
+    "quant_method": "modelopt",
+    "quant_algo": "FP8"
+  }
+}
+```
+
+Current vLLM-Omni auto-detects:
+
+- `FP8` and `FP8_PER_CHANNEL_PER_TOKEN` -> `modelopt`
+- `NVFP4` -> `modelopt_fp4`
+- `MIXED_PRECISION` -> `modelopt_mixed`
+
+Packed modules and scales must map correctly:
+
+- `to_qkv` from `to_q`, `to_k`, `to_v`
+- `add_kv_proj` from `add_q_proj`, `add_k_proj`, `add_v_proj`
+- `w13` from `w1`, `w3`
+- `.input_scale`, `.weight_scale`, `.weight_scale_2`, `.weight_scale_inv`
+
+For selective FP8, target layers that remain BF16 must be allowed to
+dequantize from FP8 weights using the matching `weight_scale`, or the adapter
+must skip the quantized scale tensors for that BF16 target.
+
+## Runtime Signals
+
+Healthy loading should show ModelOpt auto-detection and a ModelOpt linear
+kernel path. Useful signals include:
+
+- auto-detected quantization from config
+- selected ModelOpt FP8 linear method
+- CUTLASS or FlashInfer FP8 scaled-mm backend when enabled and supported
+- non-eager path compiling the transformer when expected
+
+If online serving fails while offline inference works, inspect warmup and
+serving initialization before changing checkpoint conversion.
+
+## Validation Order
+
+1. Check config completeness.
+2. Run an export checker such as `examples/quantization/check_modelopt_fp8_export.py`.
+3. Confirm actual FP8 tensors and expected disk-size reduction.
+4. Run one offline generation with fixed prompt, seed, size, and steps.
+5. Run serving with `vllm serve --omni` and verify `/v1/images/generations`.
+6. Compare against BF16 using visual review plus metrics such as LPIPS, PSNR,
+   MAE, cosine similarity, and SSIM.
+7. Compare performance only under matched modes:
+   - BF16 eager vs FP8 eager
+   - BF16 non-eager vs FP8 non-eager
+
+FP8 is not automatically faster. Check whether execution reaches FP8 kernels,
+whether compile mode is comparable, and whether the workload shape fits the
+kernel path.
+
+## Triage
+
+| Symptom | First Check |
+|---------|-------------|
+| Startup failure | `config.json`, `model_index.json`, `quantization_config`, missing scales |
+| Loads but output is corrupted | Over-quantized sensitive layers or wrong full-name prefixes |
+| Offline works but online fails | Dummy warmup, serving init, compile path |
+| FP8 is not faster | Kernel selection, dtype, compile mode, workload shape |
+| BF16 fallback fails | Missing `weight_scale` for dequantization or wrong packed-module mapping |
+
+The success criterion is stable, correctly scoped checkpoint loading with
+quality preserved enough for the target use case, not maximizing the number of
+FP8 layers.

.claude/skills/readme.md

@@ -18,11 +18,27 @@ include:
 
 - `add-diffusion-model`: guides integration of a new diffusion model into
   `vllm-omni`
+- `diffusion-perf-opt`: guides diffusion model performance optimization,
+  including profiling traces, parallel strategies, stage timing analysis, and
+  benchmark-driven tuning
+- `quantization`: guides quantization method selection, model integration,
+  checkpoint loading, and quality/performance validation for vLLM-Omni
 - `add-omni-model`: covers addition of new omni-modality model support
 - `add-tts-model`: covers integration of new TTS models and related serving
   workflows
 - `generate-release-note`: helps prepare release notes for repository changes
+- `precheck-pr`: self-check a branch before creating a PR — validates PR title
+  format, catches dead code, verifies accuracy/perf claims, and confirms merge
+  readiness
+- `vllm-omni-test`: guides generation and execution of CI-aligned tests (L1–L4),
+  pytest marker selection (`core_model` / `advanced_model` / `full_model`,
+  `omni` / `tts` / `diffusion`), Buildkite wiring (`test-ready.yml`,
+  `test-merge.yml`, `test-nightly.yml`, `test-weekly.yml`), and copy-paste
+  local plus CI-like `pytest` commands; see `references/test-routing.md` for
+  level-to-command mapping
 - `review-pr`: provides a structured workflow for reviewing pull requests
+- `vllm-omni-npu-model-runner-upgrade`: upgrades NPU model runners to align with the
+  latest vllm-ascend NPUModelRunner
 
 ## Maintenance Guidelines
 

.claude/skills/vllm-omni-test/SKILL.md

@@ -0,0 +1,894 @@
+---
+name: vllm-omni-test
+description: Generate and run tests for vllm-project/vllm-omni with CI-aligned levels and markers; wire new tests into Buildkite (test-ready.yml for L1/L2, test-merge.yml for L3, test-nightly.yml for L4). On completion, always provide copy-paste local and CI-like pytest commands plus prerequisites. Use when creating regression tests, adding L1-L4 coverage, selecting pytest markers, or validating fixes from issues/PRs.
+---
+
+# vLLM-Omni Test Generator & Runner
+
+## Purpose
+
+Use this skill to generate minimal, stable test cases and run them with the correct marker/level strategy for [vllm-project/vllm-omni](https://github.com/vllm-project/vllm-omni).
+
+**Link convention:** Paths such as `.buildkite/` and `docs/contributing/` live at the **vllm-omni repo root**. Markdown links use repo-relative paths from this skill file (e.g. `../../../.buildkite/cuda/test-ready.yml`, `../../../docs/contributing/ci/test_system_overview.md`).
+
+Default priorities:
+
+1. Reproducible regression coverage for bug fixes
+2. Correct test level and marker selection
+3. Low flake, low dependency tests first
+4. CI-compatible run commands
+5. **Actionable run commands for the human**: whenever you add or change tests, always finish with **copy-paste-ready** `pytest` lines (local: single file and/or single test; CI-like: markers + `--run-level`), plus short **prerequisites** (GPU tier, HF cache, optional `model_prefix`). Do not assume the reader will infer commands from `test-routing.md` alone.
+
+## Inputs
+
+- Issue/PR link and summary
+- Changed files or suspected code path
+- Whether the user wants local quick validation or CI-equivalent validation
+- Hardware constraints (CPU only / CUDA / ROCm / NPU)
+
+## Workflow
+
+### Step 1: Classify Test Goal
+
+- **Bugfix regression**: start from a minimal failing scenario and add assertions that prevent recurrence. Before writing tests, output **`required` / `recommended` / `not_needed`**: **`required`** — stable logic/contract bug that should have been caught; **`recommended`** — environment-sensitive but a small regression still helps; **`not_needed`** — one-off external/config failure or existing tests already cover the path. Prefer the narrowest stable L1 (CPU) case; escalate to L2/L3 only when the bug needs real weights or serving.
+- **Feature coverage**: verify new behavior and one negative/boundary case.
+- **Perf/benchmark claim**: require benchmark-oriented tests and explicit metrics.
+
+### Step 2: Select Test Level
+
+- **L1**: unit/logic, deterministic, CPU-friendly, fastest feedback.
+- **L2**: basic e2e/integration and platform-dependent checks.
+- **L3/L4**: advanced model/integration/perf validation.
+
+Use [references/test-routing.md](references/test-routing.md) for level-to-marker and command mapping.
+
+### Step 3: Pick Markers
+
+Always attach markers deliberately:
+
+- **Level**: `core_model` (L1/L2) and/or `advanced_model` (L3) and/or `full_model` (L4 nightly)
+- **Model type** (required on model-centric e2e — pick exactly one):
+  - `omni` — end-to-end multimodal LLM pipelines (thinker/talker/stages; Qwen-Omni family)
+  - `tts` — speech synthesis / TTS-only models (`/v1/audio/speech`, voice clone, etc.)
+  - `diffusion` — generative diffusion models (image / audio / text / video from noise)
+- **Cross-cutting area** (when relevant): `parallel`, `cache`, `example`, `benchmark`
+- **Hardware**: `cpu`, `gpu`, `cuda`, `rocm`, `npu`, `L4`, `H100`, `distributed_cuda`, …
+- Optional: `slow`, distributed markers when multi-card is required
+
+**Baseline smoke (L2 + L3):** The simplest e2e case per model — default deploy, minimal request — should usually carry **both** `@pytest.mark.core_model` and `@pytest.mark.advanced_model` on the **same** test function so `test-ready.yml` and `test-merge.yml` share one test. `send_*_request` picks validation depth from `--run-level`. References: `test_voxcpm2_tts.py::test_text_to_audio_001`, `test_qwen3_tts_customvoice.py::test_text_to_audio_001`. Heavier scenarios use **`advanced_model` only**; L4 expansion uses **`full_model`**.
+
+For hardware-aware tests, prefer `@hardware_test(...)` or `hardware_marks(...)` in `tests/helpers/mark.py`.
+
+**Diffusion nightly split** (`test-nightly.yml`): all diffusion tests use `pytest.mark.diffusion`, but CI groups them by **output modality**, not by separate pytest markers:
+
+| Nightly group | Scope | Typical models / paths |
+|---------------|--------|-------------------------|
+| **Diffusion X2I(&A&T) Model Test** | x2**i** (image), x2**a** (audio), x2**t** (text) and other **non-video** diffusion | Qwen-Image*, BAGEL, FLUX, SD3, Z-Image, LongCat, DreamZero, … — `test_*_expansion.py` under X2I shards; L4 sweep uses `-k "not test_wan and not test_bagel_expansion and not hunyuan"` for L4 |
+| **Diffusion X2V Model Test** | x2**v** (video) only | Wan2.2, HunyuanVideo 1.5, Wan VACE, … — e.g. `test_wan22_expansion.py`, `test_hunyuan_video_15_expansion.py` |
+
+Wire new **video** diffusion expansion tests into the **X2V** group; wire **image/audio/text** diffusion expansion tests into **X2I(&A&T)**. Do not place x2v modules in X2I shards (see comment in `test-nightly.yml` above the X2I group).
+
+### Naming: generated test module files (L2–L4, model-centric e2e)
+
+When adding **new** pytest modules whose primary scope is a **specific model** (typical under `tests/e2e/offline_inference/` or `tests/e2e/online_serving/`), use this filename pattern:
+
+| Level | Filename pattern | Example (model `Qwen/Qwen2.5-Omni-7B`) |
+|-------|------------------|----------------------------------------|
+| **L2**, **L3** | `test_{lowercase_model_slug}.py` | `test_qwen2_5_omni.py` |
+| **L4** | `test_{lowercase_model_slug}_expansion.py` | `test_qwen2_5_omni_expansion.py` |
+
+**Slug rules for `{lowercase_model_slug}`:**
+
+1. Start from the HuggingFace-style id (e.g. `Qwen/Qwen2.5-Omni-7B`), but **do not** put the org into the filename: use the **repo segment only** (`Qwen2.5-Omni-7B`), not `Qwen_...` / `qwen_qwen2_5_...`.
+2. **Lowercase**; replace `.`, `-`, and whitespace with a single `_` (e.g. `Qwen2.5-Omni-7B` → `qwen2_5_omni`). **Omit** trailing size tokens such as `7b` / `30b` in the basename when a single file covers that model line in the directory (matches `test_qwen2_5_omni.py` in-tree).
+3. If two checkpoints in the same folder need separate modules, add a **minimal** disambiguator (e.g. `_7b` vs `_3b`) only then.
+4. **L1** unit tests are **not** bound to this pattern; use `tests/<area>/test_<feature>.py` as today.
+
+Routing tables and commands: [references/test-routing.md](references/test-routing.md) § *Model-centric e2e filename convention*.
+
+Existing references: `tests/e2e/offline_inference/test_qwen2_5_omni.py` (L2-style omni), `tests/e2e/offline_inference/test_qwen3_5_9b.py` (L2-style omni, single-stage VL), `tests/e2e/online_serving/test_qwen3_omni_expansion.py` (L4-style omni), `tests/e2e/online_serving/test_qwen_image_edit_expansion.py` / `test_qwen_image_expansion.py` (L4-style diffusion).
+
+### Step 4: Generate Test Case Skeleton
+
+**1. Pick the functional scenario** (then choose directory, fixtures, and markers):
+
+| Scenario | Typical location | Fixtures / runner pattern | Baseline markers & level |
+|----------|------------------|---------------------------|---------------------------|
+| **Offline inference e2e** | `tests/e2e/offline_inference/` | **Module (default):** `omni_runner` + `omni_runner_handler`. **Function (isolation only):** `omni_runner_function` + `omni_runner_handler_function`. Diffusion/TTS may use `Omni(...).generate` directly | L2: `core_model` + **one of** `omni` / `tts` / `diffusion`; `@hardware_test(...)` when GPU/NPU is required |
+| **Online serving e2e** | `tests/e2e/online_serving/` | **Module (default):** `omni_server` + `openai_client`. **Function (isolation only):** `omni_server_function` + `openai_client_function`. Clients: `send_omni_request` (omni), `send_audio_speech_request` (tts), `send_diffusion_request` / `send_video_diffusion_request` / `send_images_generations_request` (diffusion) | Baseline smoke: **`core_model` + `advanced_model`**; heavier paths: `advanced_model` only; L4 expansion: `full_model` |
+| **Documentation / runnable examples** | `tests/examples/offline_inference/`, `tests/examples/online_serving/` | **Offline docs (preferred):** extract Python/Bash blocks from the doc README (e.g. `ReadmeSnippet.extract_readme_snippets`), `pytest.mark.parametrize` each snippet, run via `example_runner.run` with a stable `output_subfolder`. **Online docs:** copy client/request scripts into dedicated tests and keep them in sync with the doc page. | Usually **L4**: `advanced_model`, often `example` plus hardware marks matching the nightly docs-example job (see `.buildkite/cuda/test-nightly.yml`). Full conventions: [docs/contributing/ci/test_examples/l4_doc_example_tests.inc.md](../../../docs/contributing/ci/test_examples/l4_doc_example_tests.inc.md) (introduced in [PR #1910](https://github.com/vllm-project/vllm-omni/pull/1910): naming, output directory layout, skip rules, avoid trimming `num_inference_steps` without a strong CI reason). |
+| **Performance / benchmark** | `tests/dfx/perf/tests/*.json` + `run_*_benchmark.py` | JSON or script-driven server + load config; assert explicit metrics / baselines | L4 Perf: JSON `mark` with `full_model` + `omni`/`tts`/`diffusion`; wire `test-nightly.yml` Perf steps |
+| **Invalid parameter / negative HTTP validation** | `tests/dfx/reliability/invalid_param_test/` | Live `omni_server` + low-level `send_*_http_request` with `err_code` / `err_message` | `pytest.mark.slow` + `omni` / `tts` / `diffusion` + `@hardware_marks` (`H100` or `L4`); CI in **`test-weekly.yml`** (not ready/merge/nightly) |
+
+**If the user’s test plan includes invalid parameter validation / invalid params / negative HTTP / 400 validation:** do **not** add those `test_*` functions to `tests/e2e/online_serving/test_*.py` or `*_expansion.py`. **Move or author them** under `tests/dfx/reliability/invalid_param_test/` in the **endpoint-matching script** (see **Invalid parameter validation** below). Success-path e2e and invalid-param dfx tests must stay in separate modules.
+
+**1b. Model type — `omni` vs `tts` vs `diffusion`**
+
+After choosing offline/online/docs/perf, classify the **product under test** and attach **exactly one** model-type marker. All three can live under the same `tests/e2e/` trees; conventions diverge:
+
+| Dimension | **Omni** (`pytest.mark.omni`) | **TTS** (`pytest.mark.tts`) | **Diffusion** (`pytest.mark.diffusion`) |
+|-----------|------------------------------|----------------------------|----------------------------------------|
+| **What it is** | Multimodal LLM pipeline (thinker/talker/stages; text + vision + audio I/O) | Speech synthesis / voice models | Generative diffusion (noise → image, audio, text, or video) |
+| **Examples** | Qwen2.5-Omni, Qwen3-Omni | Qwen3-TTS, VoxCPM2, Higgs-Audio, Step-Audio2 | Qwen-Image, BAGEL, Wan2.2, HunyuanVideo |
+| **Offline runner** | `omni_runner` + `omni_runner_handler`, `generate_multimodal` | `Omni(...).generate` or stage YAML + TTS params | `Omni(...).generate` + `OmniDiffusionSamplingParams` |
+| **Online client** | `openai_client.send_omni_request` (chat completions, modalities) | `openai_client.send_audio_speech_request` (`/v1/audio/speech`) | `send_diffusion_request` (chat/T2I), `send_video_diffusion_request` (`/v1/videos`, X2V), or `send_images_generations_http_request` / `send_images_edits_http_request` (DALL-E routes) |
+| **Typical assertions** | Stage outputs, text/audio keywords via `OmniRunnerHandler` / response handler | WAV bytes, stream chunks, speech endpoint contract | Image/video dimensions, `final_output_type`, `assert_diffusion_response` |
+| **Stage / deploy YAML** | Per-model omni stage configs (`ci/qwen3_omni_moe.yaml`, …) | `qwen3_tts.yaml`, `voxcpm2.yaml`, … | Often default serve; parallel/offload YAML for heavy DiT |
+| **Nightly group** (`test-nightly.yml`) | **Omni Model Test** — `-m "full_model and omni"` | **TTS Model Test** — `-m "full_model and tts"` | **Diffusion X2I(&A&T)** *or* **Diffusion X2V** (see Step 3 table; same `diffusion` marker, different YAML group / file shard) |
+| **L4 pressure** | Expansion per modality/model as needed | Expansion + accuracy/perf in TTS group | X2I: merge feature combos per [#1832](https://github.com/vllm-project/vllm-omni/issues/1832); X2V: separate nightly group |
+
+Do **not** mix fixtures across types (e.g. do not use `omni_runner` layout for a pure diffusion or TTS model without mirroring an in-tree test in that family).
+
+**Diffusion only — X2I(&A&T) vs X2V (nightly routing, not extra markers):**
+
+- **X2I(&A&T)**: image / audio / text generation — Qwen-Image*, FLUX, SD3, Z-Image, BAGEL (expansion), LongCat, audio diffusion, etc.
+- **X2V**: video generation only — Wan2.2, HunyuanVideo 1.5, Wan VACE, LTX video similarity paths.
+
+When adding a new `test_*_expansion.py`, place it in the matching **nightly group** step (explicit file list in `test-nightly.yml`), not only by marker expression.
+
+**2. Use the narrowest deterministic skeleton for the scenario**
+
+*L1 unit / logic (CPU-first):*
+
+**Mocking rule (L1 only):** use **pytest** integration — **`mocker`** (`pytest-mock`) or **`monkeypatch`** (built-in). **Do not** import or call **`unittest.mock`** (`patch`, `MagicMock`, `@patch`, `with patch(...)`, etc.) in L1 tests; patches must auto-revert with the test lifecycle.
+
+```python
+import pytest
+
+pytestmark = [pytest.mark.core_model, pytest.mark.cpu]
+
+
+def test_<scenario_name>(mocker):
+    # Prefer mocker.patch / mocker.spy / mocker.Mock — not unittest.mock.patch
+    fake_fn = mocker.patch("vllm_omni.some.module.expensive_call", return_value=...)
+    # Act
+    # Assert
+    fake_fn.assert_called_once()
+
+
+def test_<env_or_attr>(monkeypatch):
+    # For simple env / attribute substitution without a Mock object
+    monkeypatch.setenv("SOME_FLAG", "1")
+    monkeypatch.setattr("vllm_omni.some.module.CONST", 42)
+    ...
+```
+
+**Avoid in L1:**
+
+```python
+# BAD — do not use in L1 unit tests
+from unittest.mock import patch, MagicMock
+
+@patch("vllm_omni.some.module.fn")
+def test_bad(mock_fn): ...
+
+def test_also_bad():
+    with patch("...") as m: ...
+```
+
+See **L1 unit test constraints (mocking)** below for the full do/don't list.
+
+*Offline multimodal e2e — **Omni** (representative):*
+
+```python
+@pytest.mark.core_model
+@pytest.mark.omni
+@hardware_test(...)
+@pytest.mark.parametrize("omni_runner", test_params, indirect=True)
+def test_<scenario>(omni_runner, omni_runner_handler) -> None:
+    request_config = {"prompts": ..., "modalities": [...]}  # optional: images, videos, audios
+    omni_runner_handler.send_omni_request(request_config)
+```
+
+*Offline generative e2e — **Diffusion** (representative):*
+
+```python
+@pytest.mark.core_model
+@pytest.mark.diffusion
+@hardware_test(...)
+@pytest.mark.parametrize("omni_runner", test_params, indirect=True)
+def test_text_to_image_001(omni_runner_handler) -> None:
+    omni_runner_handler.send_diffusion_request({"prompt": "...", "extra_body": {"num_inference_steps": 4, ...}})
+```
+
+*Offline TTS e2e — **Qwen3-TTS** (two-stage; representative):*
+
+```python
+@pytest.mark.advanced_model
+@pytest.mark.tts
+@hardware_test(...)
+@pytest.mark.parametrize("omni_runner", tts_server_params, indirect=True)
+def test_text_to_audio_001(omni_runner, omni_runner_handler) -> None:
+    omni_runner_handler.send_audio_speech_request({
+        "input": "...",
+        "task_type": "Base",
+        "ref_audio": REF_AUDIO_URL,
+        "ref_text": REF_TEXT,
+    })
+```
+
+*Offline TTS e2e — **single-stage** (Coqui XTTS, MOSS-TTS-Nano; representative):*
+
+```python
+@pytest.mark.advanced_model
+@pytest.mark.tts
+@hardware_test(...)
+@pytest.mark.parametrize("omni_runner", tts_server_params, indirect=True)
+def test_voice_clone_001(omni_runner_handler) -> None:
+    omni_runner_handler.send_single_stage_tts_request({
+        "input": "...",
+        "language": "en",
+        "prompt_audio_path": REF_AUDIO_PATH,
+        "response_format": "wav",
+        "run_level": "advanced_model",
+    })
+```
+
+*Online serving e2e — **Omni** (representative):*
+
+```python
+@pytest.mark.core_model
+@pytest.mark.omni
+@hardware_test(...)
+@pytest.mark.parametrize("omni_server", test_params, indirect=True)
+def test_text_to_text_001(omni_server, openai_client) -> None:
+    request_config = {"model": omni_server.model, "messages": ..., "modalities": ["text"]}
+    openai_client.send_omni_request(request_config)
+```
+
+*Online serving e2e — **TTS** (representative):*
+
+```python
+@pytest.mark.core_model
+@pytest.mark.advanced_model
+@pytest.mark.tts
+@hardware_test(...)
+@pytest.mark.parametrize("omni_server", tts_server_params, indirect=True)
+def test_text_to_audio_001(omni_server, openai_client) -> None:
+    request_config = {"model": omni_server.model, "input": "...", "response_format": "wav", ...}
+    openai_client.send_audio_speech_request(request_config)
+```
+
+*Online serving e2e — **Diffusion** X2I (representative):*
+
+```python
+@pytest.mark.core_model
+@pytest.mark.diffusion
+@pytest.mark.parametrize("omni_server", _get_default_case(MODEL), indirect=True)
+def test_text_to_image_001(omni_server, openai_client) -> None:
+    openai_client.send_diffusion_request({...})  # chat completions + extra_body
+```
+
+*Online serving e2e — **Diffusion** X2V (representative):*
+
+```python
+@pytest.mark.core_model
+@pytest.mark.diffusion
+def test_text_to_video_001(omni_server, openai_client) -> None:
+    openai_client.send_video_diffusion_request({"model": ..., "form_data": {...}})  # /v1/videos
+```
+
+*Documentation example tests:* follow the **Preferred Test Strategy** in [l4_doc_example_tests.inc.md](../../../docs/contributing/ci/test_examples/l4_doc_example_tests.inc.md): dynamic extraction for offline READMEs; explicit copied client code for online pages until extraction is justified; use the documented **naming**, **output directory** (page folder + case id), and **skipping** rules (e.g. Gradio-only scripts).
+
+*Performance tests:* add or extend entries under `tests/dfx/perf/tests/` (and JSON configs where the project uses them), with **explicit baselines**, **`mark`** on each case (`hardware_marks` + `full_model` + type marker), and the same nightly Perf step pattern as in-tree configs.
+
+**3. Cross-cutting rules**
+
+- Reuse existing fixtures for the chosen scenario; do not mix “online client” assumptions into offline `OmniRunner` tests without a clear reason.
+- Avoid external network dependency in assertions unless the scenario is explicitly “online serving” or doc examples that require a model hub (then align with CI secrets/cache).
+- Keep **one test function = one intent** (one modality combo, one endpoint contract, or one acceleration combo).
+- **E2E test function layout**: **one case → one `test_<scenario>` function** with a name that states what is validated (endpoint, size/`n`, server flag, or route). **Do not** merge multiple cases into a single test that branches on `request.node.callspec.id`, `param.id`, or `if case_id == ...`. Use `@pytest.mark.parametrize("omni_server", [...], indirect=True)` per function (usually **one** `OmniServerParams` per test). A loop inside one test is OK only when it serves **that function’s single intent** (e.g. three standard sizes in `test_*_sizes_256_512_1024`).
+- **Runtime fixture scope** (`tests/helpers/fixtures/runtime.py`): default **`omni_server` / `omni_runner`** (module) + matching client/handler; use **`omni_server_function` / `omni_runner_function`** only when each `test_*` must start a fresh instance (see below).
+- **L1 mocks**: never `unittest.mock`; use `mocker` or `monkeypatch` only (see below).
+- **API calls (L2+ e2e, online and offline)**: **reuse** `send_*_request` in `tests/helpers/runtime.py` when it exists; **otherwise add** the helper in `runtime.py` first, then call it from the test. General `assert_*` inside `send_*_request`; special `assert_*` only in the test. See **Runtime send helpers** below — do not call `omni.generate`, raw HTTP, or SDK clients from `test_*.py`.
+- **Response assertions**: reusable checks on API bodies / decoded media belong in **`tests/helpers/assertions.py`** — not as private `_assert_*` helpers inside `test_*.py` (see below).
+- **Model-specific payloads stay in test modules** — per-model `MODEL`, deploy path, `REF_AUDIO_URL`, `get_prompt()`, `_build_request()`, and inline `request_config` dicts live in `test_{slug}.py` / `test_{slug}_expansion.py` (and offline/L1 siblings). **Do not** create `tests/helpers/{slug}.py` to deduplicate them; a little copy across files is preferred (see `test_glm_tts.py`, `test_cosyvoice3_tts_expansion.py`). `tests/helpers/` is repo-wide harness only (`mark`, `media`, `runtime`, `stage_config`, `assertions`, `fixtures/`).
+
+#### L1 unit test constraints (mocking)
+
+L1 tests (`core_model and cpu`, under `tests/diffusion/`, `tests/engine/`, `tests/model_executor/`, etc.) must follow the repo’s **pytest-mock** convention (`pytest-mock>=3.10.0` in `pyproject.toml` `[project.optional-dependencies] dev`).
+
+| Do | Don't |
+|----|-------|
+| `def test_foo(mocker):` + `mocker.patch(...)`, `mocker.spy(...)`, `mocker.Mock()`, `mocker.MagicMock()`, `mocker.AsyncMock()` | `from unittest.mock import patch, MagicMock, Mock, AsyncMock` |
+| `def test_bar(monkeypatch):` + `monkeypatch.setattr` / `setenv` / `delenv` / `setitem` | `@patch(...)` decorator |
+| Let `mocker` auto-stop patches after the test | `with patch(...):` / `patch.object(...)` context managers |
+| Mirror neighboring L1 tests in the same directory | `unittest.mock.create_autospec` unless an existing file already documents an exception |
+
+**Rationale:** `mocker` ties patch lifecycle to pytest fixtures (no leaked patches across tests). `unittest.mock` decorators/context managers are easy to compose incorrectly with parametrized or async tests and are inconsistent with in-tree L1 style.
+
+**Minimal patterns:**
+
+```python
+def test_returns_cached_config(mocker):
+    loader = mocker.patch(
+        "vllm_omni.foo.load_yaml",
+        return_value={"stages": []},
+    )
+    result = get_deploy_config("ci/foo.yaml")
+    assert result["stages"] == []
+    loader.assert_called_once_with("ci/foo.yaml")
+
+
+def test_skips_when_env_unset(monkeypatch):
+    monkeypatch.delenv("VLLM_OMNI_FEATURE", raising=False)
+    assert should_enable_feature() is False
+```
+
+E2E levels (L2+) generally **avoid mocks**; if a rare L2 stub is unavoidable, still prefer `mocker` over `unittest.mock` for consistency.
+
+#### Runtime fixtures — scope (`tests/helpers/fixtures/runtime.py`)
+
+vLLM-Omni e2e tests start a real **OmniServer** (online) or **OmniRunner** (offline). Pick scope by **how often the process must be recreated**, not by test level alone.
+
+| Scope | Online fixtures | Offline fixtures | When to use |
+|-------|-----------------|------------------|-------------|
+| **Module** (default) | `omni_server` → `openai_client` | `omni_runner` → `omni_runner_handler` | **Default for L2/L3/L4 expansion** — amortize model/server init across `test_*` in the same module. Same `OmniServerParams` / runner config can be reused by multiple tests. |
+| **Function** | `omni_server_function` → `openai_client_function` | `omni_runner_function` → `omni_runner_handler_function` | **Only when required** — each `test_*` must get a **clean** server/runner (no shared engine/GPU state). Typical: `tests/dfx/reliability/`, sleep/wakeup, crash/restart, tests that mutate global server state. |
+
+**Rules:**
+
+1. **Default to module scope** (`omni_server` / `omni_runner`) unless the scenario **explicitly** needs a fresh instance per test function.
+2. **Indirect parametrize name must match the fixture name**: `@pytest.mark.parametrize("omni_server", ...)` with `omni_server` + `openai_client`; `@pytest.mark.parametrize("omni_server_function", ...)` with `omni_server_function` + `openai_client_function`. Do not mix module fixture with function client (or vice versa).
+3. Different `OmniServerParams` per `test_*` (e.g. default vs `--enable-cpu-offload`) is still OK with **module** `omni_server` — pytest parametrizes per test node; only switch to `_function` when isolation between tests matters, not merely because `server_args` differ.
+4. One `test_*` with **many** server configs (expansion matrix) → single function + `@pytest.mark.parametrize("omni_server", [...], indirect=True)` + module `omni_server` (see in-tree `test_qwen_image_expansion.py`).
+
+```python
+# Default — module-scoped server (L4 expansion)
+@pytest.mark.parametrize("omni_server", [pytest.param(OmniServerParams(model=MODEL), marks=H100)], indirect=True)
+def test_foo_images_generations_default_1024(omni_server, openai_client) -> None:
+    openai_client.send_images_generations_request({...})
+
+# Function-scoped — reliability / per-test clean state only
+@pytest.mark.parametrize(
+    "omni_server_function",
+    [pytest.param(OmniServerParams(model=MODEL), marks=H100)],
+    indirect=True,
+)
+def test_foo_sleep_wakeup_cycle(omni_server_function, openai_client_function) -> None:
+    openai_client_function.send_omni_sleep_http_request({...})
+    openai_client_function.send_omni_wakeup_http_request({...})
+```
+
+#### Runtime send helpers — online **and** offline (`tests/helpers/runtime.py`)
+
+**L2+ e2e (online serving and offline inference) must call APIs through `tests/helpers/runtime.py`.** Fixtures live in `tests/helpers/fixtures/runtime.py`; **send/assert implementation** lives in `tests/helpers/runtime.py` + `tests/helpers/assertions.py`.
+
+| Principle | Action |
+|-----------|--------|
+| **Reuse first** | Grep `runtime.py` for an existing **`send_*_request`** on `OpenAIClientHandler` (online) or `OmniRunnerHandler` (offline) that matches the endpoint / pipeline shape. |
+| **Extend when close** | If an existing `send_*_request` almost fits (e.g. missing one optional field), extend it in `runtime.py` — do not fork logic in the test file. |
+| **Add when missing** | No suitable helper → add **`send_<feature>_request`** (high-level: call + general `assert_*`) or **`send_<route>_<verb>_http_request`** (low-level HTTP for negative/dfx) in **`runtime.py` first**, then call it from tests. |
+| **Test module owns payload only** | In `test_*.py`: `MODEL`, deploy path, vendored media, `get_prompt()`, and inline **`request_config` dicts** only. **No** `omni.generate(...)`, raw `requests.post`, OpenAI SDK calls, or `_collect_audio()` / `_process_output()` in e2e tests. |
+
+**Online** (`openai_client` from `omni_server`): `OpenAIClientHandler.send_*_request`.
+
+**Offline** (`omni_runner_handler` from `omni_runner`): `OmniRunnerHandler.send_*_request`.
+
+| Do | Don't |
+|----|-------|
+| Online: `openai_client.send_omni_request`, `send_diffusion_request`, `send_audio_speech_request`, `send_video_diffusion_request`, `send_images_generations_request`, … | `requests.post(f"{base_url}/v1/...", json=…)` or `client.chat.completions.create(...)` inside a test |
+| Offline: `omni_runner_handler.send_omni_request`, `send_diffusion_request`, `send_audio_speech_request`, `send_single_stage_tts_request`, `send_single_stage_tts_batch_request`, … | `omni_runner.omni.generate(...)` + hand-rolled tensor/WAV extraction in `test_*.py` |
+| Add missing `send_*` to **`runtime.py`** first; bundle general `assert_*` inside the send helper | A one-off `def _post_*` / `def _collect_audio` at the bottom of a test module |
+| Mirror naming/style of neighboring `send_*` (docstring, `request_config` dict, optional `run_level`, `err_code` / `err_message` for negative cases) | Different parameter shapes per test file for the same endpoint |
+
+**Workflow when generating online/offline e2e tests:**
+
+1. Decide whether the needed check is **general** (every success call of this `send_*`) or **special** (one case / one parameter combo only). See **General vs special assert placement** below.
+2. **Search `runtime.py`** for a matching **`send_*_request`**; if missing, add it (with general `assert_*` inside) before writing the test body.
+3. In the test module: build `request_config` → call **`send_*_request` only** for the general contract; call **`assert_*` in the test file only** for special, case-specific checks (import from `assertions.py`).
+4. Reserve **low-level** `send_*_http_request` for negative/dfx tests (`err_code` / `err_message`) — not for ordinary L2+ success-path e2e.
+5. When wiring Buildkite `source_file_dependencies`, include **`tests/helpers/runtime.py`** and/or **`tests/helpers/assertions.py`** when new helpers were added.
+
+**Exceptions (document in the test docstring why):** models whose offline prompt path cannot go through existing handlers yet (e.g. `test_higgs_audio_v2.py`, `test_voxtral_tts.py` with custom tokenizer compose) may call `omni.generate` directly until a `send_*_request` is added to `runtime.py` — treat as **debt**, not the default for new TTS/diffusion/omni e2e.
+
+**General vs special assert placement:**
+
+| Kind | Definition | Where it lives | Test file calls |
+|------|------------|----------------|-----------------|
+| **General** | Default success contract for this endpoint/client method — e.g. HTTP 200, `data[]` shape, `n` count, `size` dimensions, decodable image/audio, bundled omni/diffusion fields | Implement in `assertions.py`, invoke **inside** `send_*_request` in `runtime.py` (after low-level send / SDK call, when `err_code` is not set) | **`send_*_request` only** — do **not** repeat the same `assert_*` |
+| **Special** | Extra check tied to **this test case only** — e.g. seed byte-identical replay, accuracy/CLIP threshold, perf ceiling, model-specific optional field | Add or reuse `assert_*` in `assertions.py` (never inline in test) | `send_*_request` **then** `assert_<special>(...)` **once** for that case |
+
+Changing `request_config` fields (`size`, `n`, `seed`, server flags) is **not** special validation — the general `assert_*` should read those from `request_config` inside `send_*`.
+
+**`send_*` ↔ `assert_*` pairing (online `OpenAIClientHandler`):**
+
+| High-level `send_*` (prefer in L2+ e2e) | Assert already invoked inside `runtime.py` | Low-level HTTP-only sibling |
+|----------------------------------------|--------------------------------------------|-----------------------------|
+| `send_omni_request` | `assert_omni_response` | `send_chat_completions_http_request` → `assert_http_error` only |
+| `send_diffusion_request` | `assert_diffusion_response` | — |
+| `send_audio_speech_request` | `assert_audio_speech_response` | `send_audio_speech_http_request` → `assert_http_error` only |
+| `send_video_diffusion_request` | `assert_diffusion_response` | `send_videos_*_http_request` → `assert_http_error` only |
+| `send_images_generations_request` *(add when needed)* | **`assert_images_generations_response`** *(general — inside send)* | `send_images_generations_http_request` → `assert_http_error` only |
+| `send_images_edits_request` *(add when needed)* | **`assert_images_edits_response`** *(general — inside send)* | `send_images_edits_http_request` → `assert_http_error` only |
+
+**Rule:** Tests call high-level `send_*_request` for the general contract. **Never** call the same bundled `assert_*` again. Call an extra `assert_*` in the test **only** for special, case-specific validation.
+
+**Common `OpenAIClientHandler` entry points** (non-exhaustive — grep `runtime.py` before adding):
+
+| Area | High-level (SDK + assert) | Low-level HTTP (`*_http_request`) |
+|------|---------------------------|-----------------------------------|
+| Omni chat | `send_omni_request` | `send_chat_completions_http_request` |
+| Diffusion T2I (chat route) | `send_diffusion_request` | — |
+| Diffusion X2V | `send_video_diffusion_request` | `send_videos_create_http_request`, `send_video_content_http_request`, … |
+| DALL-E T2I / edit | `send_images_generations_request`, `send_images_edits_request` | `send_images_generations_http_request`, `send_images_edits_http_request` |
+| TTS (online) | `send_audio_speech_request` | `send_audio_speech_http_request`, `send_audio_generate_http_request`, … |
+| Ops / meta | — | `send_health_http_request`, `send_models_http_request`, `send_omni_sleep_http_request`, … |
+
+**Common `OmniRunnerHandler` entry points** (offline — grep `runtime.py` before adding):
+
+| Area | High-level (`send_*_request` + assert) | Notes |
+|------|----------------------------------------|-------|
+| Omni multimodal | `send_omni_request` | `generate_multimodal` path |
+| Diffusion offline | `send_diffusion_request` | chat-route / `OmniTextPrompt` offline |
+| Qwen-style TTS | `send_audio_speech_request` | two-stage, `generate_multimodal` + `mm_processor_kwargs` |
+| Single-stage TTS (Coqui XTTS, MOSS-TTS-Nano, …) | `send_single_stage_tts_request`, `send_single_stage_tts_batch_request` | `prompt` + `additional_information` + `omni.generate` — **do not** duplicate in tests |
+| New model family | **Add** `send_<family>_request` here first | Then call from `tests/e2e/offline_inference/test_*.py` |
+
+L1 tests under `tests/entrypoints/` may use **FastAPI `TestClient`** or direct handler calls with mocks; they do **not** need `OpenAIClientHandler` / `OmniRunnerHandler`. **L2+ online and offline e2e** must use `runtime.py` `send_*` helpers.
+
+#### Invalid parameter validation (`tests/dfx/reliability/invalid_param_test/`)
+
+When the user asks for **invalid parameter validation**, **invalid request bodies**, **HTTP 4xx contract tests**, or any case that sends **malformed / out-of-range / mismatched** API payloads against a **live** server:
+
+1. **Do not** place these in `tests/e2e/online_serving/` or `*_expansion.py`. If already drafted there, **move** the `test_*` into the correct `invalid_param_test` script and delete the duplicate from e2e.
+2. **Pick the script by HTTP route** (extend an existing file; add a new `test_invalid_<area>.py` only when no in-tree script covers that route family):
+
+| API / area | Script |
+|------------|--------|
+| Omni chat completions, WebSocket video/realtime paths | `test_invalid_omni_chat.py` |
+| `POST /v1/audio/speech`, stream, batch, voices | `test_invalid_audio_speech.py` |
+| Audio diffusion endpoints | `test_invalid_audio_diffusion.py` |
+| `POST /v1/images/generations` | `test_invalid_image_generation.py` |
+| `POST /v1/images/edits` | `test_invalid_image_editing.py` |
+| `POST/GET/DELETE /v1/videos*` | `test_invalid_video_generation.py` |
+| Sleep / wakeup / server control | `test_invalid_server_control.py` |
+
+3. **Match in-tree style** in the chosen script:
+
+| Element | Convention |
+|---------|------------|
+| **Module markers** | `pytestmark = [pytest.mark.slow, pytest.mark.<omni or tts or diffusion>]` |
+| **Server fixture** | `_PARAMS` / `_QWEN3_TTS_SPEECH`-style list of `pytest.param(OmniServerParams(...), id="...", marks=hardware_marks(...))`; `@pytest.mark.parametrize("omni_server", _PARAMS, indirect=True)` |
+| **Hardware** | `hardware_marks(res={"cuda": "H100"})` for heavy diffusion/omni/video; `hardware_marks(res={"cuda": "L4"})` for smaller TTS models (must match weekly `-m "slow and L4"` step) |
+| **HTTP client** | **Low-level** `openai_client.send_*_http_request({..., "err_code": 400, "err_message": (...)} )` — **not** `send_*_request` (success path) |
+| **Case shape** | Prefer **one `test_*` per route family** + `@pytest.mark.parametrize("body_spec, err_message", [...])` with stable `id=` per case; or dedicated `test_<route>_malformed_json` when not parametrized |
+| **Body helpers** | `_minimal_<endpoint>_json(omni_server)` / `_minimal_*_form_data()`; merge overrides with `body.update(body_spec)` |
+| **Known gaps** | `pytest.mark.skip(reason="…#3649")` as `_SKIP_ISSUE_3649` when server validation is not yet strict (mirror neighboring cases) |
+| **Sections** | Route banner comments (`# ─── POST /v1/images/generations ───`) like existing files |
+| **Shared fixtures** | Reuse `tests/dfx/reliability/invalid_param_test/conftest.py` (`tiny_png_bytes`, env defaults) |
+
+4. **Adding a new model** to invalid-param coverage: append a `pytest.param(OmniServerParams(model="...", stage_config_path=..., server_args=...), ...)` entry to the script’s `_PARAMS` list — do **not** create `test_invalid_<model>.py` unless the route is new.
+
+5. **CI — `.buildkite/cuda/test-weekly.yml` only** (not `test-ready.yml` / `test-merge.yml` / `test-nightly.yml`):
+
+| Weekly step | Command | When your cases run |
+|-------------|---------|---------------------|
+| **Invalid parameters Test · H100** | `pytest -s -v tests/dfx/reliability/invalid_param_test/ -m "slow and H100"` | Diffusion / omni / video invalid-param tests with `H100` hardware mark |
+| **Invalid parameters Test · L4** | `pytest -s -v tests/dfx/reliability/invalid_param_test/ -m "slow and L4"` | TTS / lighter models marked `L4` |
+
+- **Trigger:** `build.env("WEEKLY") == "1"` or PR label `weekly-test`.
+- **Default:** extending an existing `invalid_param_test` script needs **no YAML edit** — the weekly steps already sweep the whole directory.
+- **Edit YAML only** when adding a **new hardware queue**, a **new top-level script** that must run in isolation, or a **model-specific weekly shard** (mirror neighboring reliability steps).
+- Weekly steps **do not** use `source_file_dependencies`.
+
+**Example — append to `test_invalid_image_generation.py`:**
+
+```python
+@pytest.mark.parametrize(
+    "body_spec, err_message",
+    [
+        pytest.param({"seed": -1}, ("seed", "greater_than_equal", "0"), id="seed_negative"),
+    ],
+)
+@pytest.mark.parametrize("omni_server", _PARAMS, indirect=True)
+def test_images_generations_invalid_requests(
+    omni_server: OmniServer,
+    openai_client: OpenAIClientHandler,
+    body_spec: dict[str, object],
+    err_message: str | tuple[str, ...],
+) -> None:
+    body = _minimal_images_gen_json(omni_server)
+    body.update(body_spec)
+    openai_client.send_images_generations_http_request(
+        {"json": body, "timeout": 300, "err_code": 400, "err_message": err_message}
+    )
+```
+
+See [references/test-routing.md](references/test-routing.md) **Invalid parameter / weekly CI**.
+
+#### Assertion helpers (`tests/helpers/assertions.py`)
+
+**Do not** add module-local helpers such as `_assert_images_generations_payload` or `_send_and_assert_*` in e2e test files. Response/media validation belongs in **`tests/helpers/assertions.py`**, grouped by **category**:
+
+| Category | Existing anchors | When to extend vs add |
+|----------|------------------|------------------------|
+| **Image** (chat diffusion + DALL-E JSON) | `assert_image_diffusion_response`, `assert_image_valid` | DALL-E `/v1/images/generations` JSON → add **`assert_images_generations_response`** beside image helpers (reuses `assert_image_valid`); do not fork decode logic into tests |
+| **Video** | `assert_video_diffusion_response`, `assert_video_valid` | Extend these for new video contracts |
+| **Audio** | `assert_audio_diffusion_response`, `assert_audio_speech_response`, `assert_audio_valid` | Extend for new TTS/audio endpoints |
+| **Omni multimodal** | `assert_omni_response` | Extend for new modality combos |
+| **HTTP errors** | `assert_http_error`, `assert_err_message_in_text` | Used inside low-level `send_*_http_request` and negative tests |
+
+| Do | Don't |
+|----|-------|
+| Put **general** `assert_*` inside `send_*_request` in `runtime.py`; tests call `send_*_request` only | Call `assert_diffusion_response` after `send_diffusion_request` (general assert already bundled) |
+| Put **special** `assert_*` in `assertions.py` and call it **in the test** after `send_*_request` when that case needs extra checks | Put general decode/count/size logic in the test because “this case uses `n=4`” (that belongs in general assert reading `request_config`) |
+| Extend the matching category function, or add `assert_<endpoint>_response` next to its category | `_send_and_assert_*` or PIL/base64 loops in `test_*.py` |
+| Use low-level `send_*_http_request` in dfx/negative tests (`err_code`) | Use low-level HTTP send + manual general assert in every L2+ expansion file |
+
+**Workflow when generating tests:**
+
+1. Implement or reuse **general** `assert_*` in `assertions.py` (by category).
+2. Wire it into **`send_*_request`** in `runtime.py` if not already bundled.
+3. In the test: `request_config` → `send_*_request`. Add a separate `assert_*` import/call **only** for special case validation.
+4. Update `source_file_dependencies` when `runtime.py` or `assertions.py` changes.
+
+**Example — DALL-E `/v1/images/generations`:**
+
+```python
+# tests/helpers/assertions.py — general contract (image category)
+def assert_images_generations_response(resp_body: dict, request_config: dict, *, run_level: str | None = None) -> None:
+    ...  # data[], n, size → assert_image_valid
+
+# tests/helpers/runtime.py — general assert INSIDE send
+def send_images_generations_request(self, request_config: dict[str, Any], ...) -> list[HttpResponse]:
+    responses = self.send_images_generations_http_request(request_config, ...)
+    if request_config.get("err_code") is None:
+        body = responses[0].json_body
+        assert isinstance(body, dict)
+        assert_images_generations_response(body, request_config, run_level=self.run_level)
+    return responses
+
+# tests/e2e/.../test_foo_expansion.py — one case per test_* (no case_id branching)
+@pytest.mark.parametrize("omni_server", [pytest.param(OmniServerParams(model=MODEL), marks=SINGLE_L4)], indirect=True)
+def test_foo_images_generations_default_1024(omni_server, openai_client) -> None:
+    openai_client.send_images_generations_request({"json": body, "timeout": 300})
+
+@pytest.mark.parametrize("omni_server", [pytest.param(OmniServerParams(model=MODEL), marks=SINGLE_L4)], indirect=True)
+def test_foo_images_generations_sizes_256_512_1024(omni_server, openai_client) -> None:
+    for size in ("256x256", "512x512", "1024x1024"):
+        openai_client.send_images_generations_request({"json": {**body, "size": size}, "timeout": 300})
+
+# chat diffusion — separate test; send_diffusion_request bundles assert_diffusion_response
+@pytest.mark.parametrize("omni_server", [pytest.param(OmniServerParams(model=MODEL), marks=SINGLE_L4)], indirect=True)
+def test_foo_chat_completions_t2i_fallback(omni_server, openai_client) -> None:
+    openai_client.send_diffusion_request({...})
+```
+
+L1 tests under `tests/entrypoints/` may keep **minimal** asserts next to the handler under test when validating a single branch; still prefer `assertions.py` when the same JSON/media check appears in more than one file or level (e.g. L1 protocol test + L4 e2e).
+
+#### Omni Test Writing Guidance (L1-L4 Layering)
+
+When the goal is a general Omni/multimodal test case, prioritize mapping the test to the correct purpose, directory, and resource assumptions aligned with the layers (see [test_system_overview.md](../../../docs/contributing/ci/test_system_overview.md) and [test_writing_guide.md](../../../docs/contributing/ci/test_writing_guide.md)):
+
+- **L1**: Unit/logic validation on CPU (`core_model and cpu`). Cover input validation, branches, and exception paths (`tests/<component>/test_*.py`). **Mock with `mocker` / `monkeypatch` only — not `unittest.mock`.**
+- **L2**: Basic e2e (online/offline basic scenarios). Prefer dummy/lightweight models to validate the end-to-end request-to-output-structure/streaming chain (typically `tests/e2e/online_serving/` and `tests/e2e/offline_inference/`).
+- **L3/L4**: Important integration, performance, and accuracy validation. L4 emphasizes “full functional scenarios + performance/stress + runnable doc examples” (typically `*_expansion.py` plus related expansion cases).
+
+Also keep markers consistent with the run level: use `core_model` for L1/L2 and `advanced_model` for L3/L4, and pair with `--run-level` to select the intended CI strategy.
+
+#### Diffusion Test Writing Guidance (L4 Coverage Combinations)
+
+When the task involves diffusion models/features, organize L4 test cases following [`#1832`](https://github.com/vllm-project/vllm-omni/issues/1832): combine multiple diffusion features into as few test cases as possible to fit limited CI GPU resources.
+
+Implementation strategy:
+
+1. If “full L4 is too heavy”: first provide a **reduced local validation case in L1/L2** (so the key assertion and the contract fix point are covered deterministically).
+2. Then provide **CI/nightly-ready L3/L4 high-marked cases** (e.g. `advanced_model`) to broaden coverage under resource constraints.
+
+#### L4 nightly sub-pillars — Function vs Accuracy vs Perf
+
+**L4 is not only `test_*_expansion.py`.** In `test-nightly.yml`, each model-type group typically runs **separate jobs**:
+
+| Sub-pillar | Nightly step label pattern | What you add | Typical paths |
+|------------|----------------------------|--------------|---------------|
+| **Function** | `· Function Test with …` | E2e expansion / feature matrix | `tests/e2e/online_serving/test_<model>_expansion.py` (or offline) |
+| **Accuracy** | `· Accuracy Test` | Quality / similarity vs baseline | `tests/e2e/accuracy/test_<model>*.py` |
+| **Perf** | `· Perf Test · <Model>` | Throughput / latency / memory benchmarks | `tests/dfx/perf/tests/test_<model>_vllm_omni.json` + runner script |
+| **Doc** (optional) | `· Doc Test` | Runnable doc examples | `tests/examples/*/test_text_to_image.py`, … |
+
+**Default when the user asks for “L4 functional cases”:** deliver **Function** pillar only (`*_expansion.py` + Function Test shard in `test-nightly.yml`). **Do not** silently add Perf or Accuracy unless the user also asks for **performance** / **benchmark** / **accuracy** / **full L4** / **full L4 coverage**.
+
+**When the user explicitly asks for L4 perf (or full L4 including perf):**
+
+1. **Do not** put throughput/latency assertions inside `test_*_expansion.py` — perf uses the **dfx benchmark harness**, not `omni_server` e2e fixtures.
+2. Add or extend a **JSON config** under `tests/dfx/perf/tests/` (mirror in-tree names: `test_qwen_image_vllm_omni.json`, `test_cosmos3_vllm_omni.json`, `test_qwen3_omni_async_chunk.json`).
+3. Each JSON **case**: `test_name`, optional **`mark`** (`hardware_marks` required when present; `marks`: `full_model` + `omni`/`tts`/`diffusion`), `server_params` (`model`, `serve_args` or `stage_config_name`), `benchmark_params[]` with **`name`** and explicit **`baseline`** (`throughput_qps`, `latency_mean`, `peak_memory_mb_mean`, …). Diffusion cases also set `server_type` (e.g. `"vllm-omni"`) and usually `benchmark_endpoint`.
+4. Run locally via the matching script (model type → runner):
+
+| Model type | Runner script | Example config |
+|------------|---------------|----------------|
+| **Diffusion X2I/X2V** | `tests/dfx/perf/scripts/run_diffusion_benchmark.py` | `tests/dfx/perf/tests/test_qwen_image_vllm_omni.json` |
+| **TTS** | `tests/dfx/perf/scripts/run_benchmark.py` | `tests/dfx/perf/tests/test_tts.json` (shared) or `test_{slug}.json` (dedicated; use when the model must not join the shared nightly matrix before integration — e.g. `test_voxcpm2.json`, `test_coqui_tts.json`) |
+| **Omni** | `tests/dfx/perf/scripts/run_benchmark.py` | `test_qwen3_omni_no_async_chunk.json`, `test_qwen3_omni_async_chunk.json`, `test_qwen3_omni_vllm_text.json`, `test_qwen3_omni_multi_replicas.json` |
+
+5. Wire **`test-nightly.yml` · Perf Test** step for that model (separate from Function Test): export `DIFFUSION_BENCHMARK_DIR` / `BENCHMARK_DIR`, run pytest on the script with **`--test-config-file`** (nightly perf steps do not use `-m`), **upload artifacts** (`buildkite-agent artifact upload`), often **multi-GPU H100** for diffusion perf.
+6. **One benchmark scenario → one `test_name` block** in JSON (same spirit as one `test_*` per function case). Combine server `serve_args` + workload in one entry; use multiple `benchmark_params` rows for size/step sweeps under the same server config.
+
+**Perf local commands:**
+
+```bash
+cd tests
+export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
+export DIFFUSION_ATTENTION_BACKEND=FLASH_ATTN
+# Single file (CI-like)
+pytest -s -v dfx/perf/scripts/run_diffusion_benchmark.py \
+  --test-config-file dfx/perf/tests/test_<model>_vllm_omni.json
+# Bulk load + filter by JSON mark
+pytest -sv dfx/perf/scripts/run_diffusion_benchmark.py -m "full_model and diffusion and H100"
+pytest -sv dfx/perf/scripts/run_benchmark.py -m "full_model and omni and H100"
+```
+
+**Clarify in the test plan** which L4 pillars you deliver: `Function only` | `Function + Perf` | `Function + Accuracy + Perf`.
+
+See [references/test-routing.md](references/test-routing.md) **L4 nightly pillars** for Buildkite step patterns.
+
+### Step 5: Wire Buildkite (when CI must run the new test)
+
+If the test is not already collected by an existing pipeline command (for example, L1 tests marked `core_model and cpu` are already covered by the **Simple Unit Test** step in ready/merge), update the appropriate pipeline under `.buildkite/`:
+
+| Test level | Edit this file | Typical trigger / intent |
+|------------|----------------|---------------------------|
+| **L1** and **L2** | [`.buildkite/cuda/test-ready.yml`](../../../.buildkite/cuda/test-ready.yml) | PR **ready** label; L1 CPU + L2 GPU/basic e2e (steps labeled **Omni ·**, **TTS ·**, **Diffusion ·**) |
+| **L3** | [`.buildkite/cuda/test-merge.yml`](../../../.buildkite/cuda/test-merge.yml) | Post-merge; `advanced_model` integration per model type |
+| **L4** | [`.buildkite/cuda/test-nightly.yml`](../../../.buildkite/cuda/test-nightly.yml) | Nightly; grouped by model type (see below) |
+| **Invalid param / reliability (weekly)** | [`.buildkite/cuda/test-weekly.yml`](../../../.buildkite/cuda/test-weekly.yml) | Weekly; `tests/dfx/reliability/invalid_param_test/` — **not** L1–L4 e2e pipelines |
+
+**Level-specific Buildkite delivery (follow the user’s requested level):**
+
+| User asks for | Document / edit **only** | Do **not** add by default |
+|---------------|---------------------------|---------------------------|
+| **L4** (`*_expansion.py`, `full_model`) | `test-nightly.yml` — append file to the matching nightly shard (X2I / X2V / Omni / TTS) **or** note it is already collected by an existing `-m` / `-k` sweep | `test-merge.yml` (L3 / `advanced_model`) or `test-ready.yml` (L2) |
+| **L3** | `test-merge.yml` with `advanced_model` + `source_file_dependencies` | `test-nightly.yml` unless user also wants nightly |
+| **L2** | `test-ready.yml` with `core_model` + `source_file_dependencies` | merge / nightly pipelines |
+| **Invalid param** | `test-weekly.yml` — **Invalid parameters Test** group (`-m "slow and H100"` / `-m "slow and L4"`). Usually **no YAML edit** when appending cases to existing `invalid_param_test` scripts | `test-ready.yml`, `test-merge.yml`, `test-nightly.yml` |
+
+`test-nightly.yml` shards use **explicit pytest file paths** in `commands` (and PR labels like `diffusion-x2iat-test`); they generally **do not** use `source_file_dependencies`. Only suggest merge/ready YAML when the requested level is L3/L2 (or the user explicitly asks for multi-level CI).
+
+**`test-nightly.yml` top-level groups** (each uses `full_model` + type marker in pytest `-m`):
+
+| Group | Type marker | Notes |
+|-------|-------------|--------|
+| **Omni Model Test** | `omni` | Function / doc / accuracy / **perf** for Qwen-Omni family |
+| **TTS Model Test** | `tts` | `-m "full_model and L4 and tts"` (and H100 variants when enabled); **Perf Test** uses `run_benchmark.py` + `test_tts.json` |
+| **Diffusion X2I(&A&T) Model Test** | `diffusion` | Image/audio/text diffusion; explicit file shards for **Function**; **Perf Test · &lt;Model&gt;** uses `run_diffusion_benchmark.py` + JSON config |
+| **Diffusion X2V Model Test** | `diffusion` | Video-only (Wan, HunyuanVideo, …); PR label `diffusion-x2v-test`; separate **Perf** steps when in-tree |
+
+Guidelines when editing YAML:
+
+- **Match markers and `--run-level`** to the test: L1/L2 in `test-ready.yml` use `core_model` + `omni` / `tts` / `diffusion` as appropriate; L3 merge uses `advanced_model`; L4 nightly uses `full_model` with the same type marker.
+- **Diffusion L4**: add x2i/x2a/x2t expansion files to an **X2I(&A&T)** step; add x2v expansion files to **X2V** — never rely on a broad `tests/e2e/` sweep alone (shards exist to save GPU and avoid wrong queue).
+- **`source_file_dependencies` (required for new E2E jobs in `test-ready.yml` and `test-merge.yml`)**: every step under the **`:card_index_dividers: E2E Test`** group **must** declare `source_file_dependencies` so Buildkite only schedules the job when relevant paths change. Omitting it breaks path-based CI skipping for GPU e2e jobs.
+- **Reuse or extend an existing step** when the new test shares the same marker expression, queue, and timeout; otherwise add a new `steps:` entry with the correct `agents.queue`, `timeout_in_minutes`, and docker/kubernetes plugin blocks consistent with neighboring jobs.
+- **Job timeout — use `timeout_in_minutes` on the step, not `timeout` before pytest**: when adding or documenting a new E2E job in `test-ready.yml` / `test-merge.yml`, set `timeout_in_minutes` on the step and run `pytest` directly in `commands`. **Do not** wrap pytest in `timeout 40m bash -c "..."` / `timeout 20m bash -c "..."`; Buildkite already enforces the step deadline via `timeout_in_minutes`.
+- **Never omit `plugins` in skill examples or PR YAML snippets**: H100 E2E steps on `mithril-h100-pool` use the full `kubernetes` block (`resources`, `volumeMounts`, `env` with `HF_TOKEN` secret, `nodeSelector: gpu-h100-sxm`, `devshm` + `hf-cache` volumes). L4/docker steps on `gpu_1_queue` use the full `docker#v5.2.0` block (`shm-size`, `HF_HOME`, volumes). Do not write `# ... plugins unchanged` — paste the complete block from a sibling step.
+- **Platform forks** (e.g. AMD-ready / AMD-merge) live alongside these files; apply the same level → file mapping for those pipelines when the test is platform-specific.
+
+#### `source_file_dependencies` for E2E Test jobs (`test-ready.yml`, `test-merge.yml`)
+
+When adding or splitting a step inside the **E2E Test** group, **always** include `source_file_dependencies` on that step. List paths the job actually depends on — at minimum:
+
+1. **Test module(s)** — every `tests/e2e/...` file the `pytest` command runs (online and offline if both are in one step).
+2. **Model implementation** — `vllm_omni/model_executor/models/<family>/` and/or `vllm_omni/diffusion/models/<family>/` as applicable.
+3. **Stage / input plumbing** (omni & multi-stage TTS) — `vllm_omni/model_executor/stage_input_processors/<family>.py` when present.
+4. **Deploy / stage YAML** — `vllm_omni/deploy/<config>.yaml` or `vllm_omni/deploy/ci/<config>.yaml` referenced by the test’s `OmniServerParams` / `get_deploy_config_path(...)`.
+5. **Test helpers** (when changed in the same PR) — `tests/helpers/runtime.py` for new `send_*` client methods; `tests/helpers/assertions.py` for new shared `assert_*` helpers.
+
+Use **repo-relative paths** (no leading `./`). Prefer **directory** entries (`.../models/bagel/`) when the whole package matters; use **file** entries for single test modules and YAML configs. Mirror a neighboring step for the same model family (e.g. **Diffusion · Bagel Test** in `test-ready.yml`).
+
+**Example** (L2 ready — diffusion online smoke, H100 kubernetes):
+
+```yaml
+- label: "Diffusion · Bagel Test"
+  source_file_dependencies:
+    - tests/e2e/online_serving/test_bagel.py
+    - vllm_omni/model_executor/models/bagel/
+    - vllm_omni/diffusion/models/bagel/
+    - vllm_omni/model_executor/stage_input_processors/bagel.py
+    - vllm_omni/deploy/bagel.yaml
+  timeout_in_minutes: 40
+  commands:
+    - |
+      export VLLM_IMAGE_FETCH_TIMEOUT=60
+      pytest -s -v tests/e2e/online_serving/test_bagel.py -m 'core_model' --run-level 'core_model'
+  agents:
+    queue: "mithril-h100-pool"
+  plugins:
+    - kubernetes:
+        podSpec:
+          containers:
+            - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
+              resources:
+                limits:
+                  nvidia.com/gpu: 1
+              volumeMounts:
+                - name: devshm
+                  mountPath: /dev/shm
+                - name: hf-cache
+                  mountPath: /root/.cache/huggingface
+              env:
+                - name: HF_HOME
+                  value: /root/.cache/huggingface
+                - name: HF_TOKEN
+                  valueFrom:
+                    secretKeyRef:
+                      name: hf-token-secret
+                      key: token
+          nodeSelector:
+            node.kubernetes.io/instance-type: gpu-h100-sxm
+          volumes:
+            - name: devshm
+              emptyDir:
+                medium: Memory
+            - name: hf-cache
+              hostPath:
+                path: /mnt/hf-cache
+                type: DirectoryOrCreate
+```
+
+**Example** (L3 merge — TTS online + offline, L4 docker):
+
+```yaml
+- label: "TTS · Qwen3-TTS Base Test"
+  source_file_dependencies:
+    - tests/e2e/online_serving/test_qwen3_tts_base.py
+    - tests/e2e/offline_inference/test_qwen3_tts_base.py
+    - vllm_omni/model_executor/models/qwen3_tts/
+    - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
+    - vllm_omni/deploy/qwen3_tts.yaml
+  timeout_in_minutes: 20
+  commands:
+    - |
+      export VLLM_LOGGING_LEVEL=DEBUG
+      export VLLM_ALLOW_LONG_MAX_MODEL_LEN=1
+      pytest -s -v tests/e2e/online_serving/test_qwen3_tts_base.py tests/e2e/offline_inference/test_qwen3_tts_base.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+  agents:
+    queue: "gpu_1_queue" # g6.4xlarge instance on AWS, has 1 L4 GPU
+  plugins:
+    - docker#v5.2.0:
+        image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
+        always-pull: true
+        propagate-environment: true
+        shm-size: "8gb"
+        environment:
+          - "HF_HOME=/fsx/hf_cache"
+        volumes:
+          - "/fsx/hf_cache:/fsx/hf_cache"
+```
+
+**Example** (L3 merge — diffusion online + offline, H100 kubernetes):
+
+```yaml
+- label: "Diffusion · Qwen Image Test"
+  source_file_dependencies:
+    - tests/e2e/online_serving/test_qwen_image.py
+    - tests/e2e/offline_inference/test_qwen_image.py
+    - vllm_omni/diffusion/models/qwen_image/
+  timeout_in_minutes: 40
+  commands:
+    - |
+      export VLLM_IMAGE_FETCH_TIMEOUT=60
+      pytest -s -v \
+        tests/e2e/online_serving/test_qwen_image.py \
+        tests/e2e/offline_inference/test_qwen_image.py \
+        -m 'advanced_model and cuda' --run-level 'advanced_model'
+  agents:
+    queue: "mithril-h100-pool"
+  plugins:
+    - kubernetes:
+        podSpec:
+          containers:
+            - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
+              resources:
+                limits:
+                  nvidia.com/gpu: 1
+              volumeMounts:
+                - name: devshm
+                  mountPath: /dev/shm
+                - name: hf-cache
+                  mountPath: /root/.cache/huggingface
+              env:
+                - name: HF_HOME
+                  value: /root/.cache/huggingface
+                - name: HF_TOKEN
+                  valueFrom:
+                    secretKeyRef:
+                      name: hf-token-secret
+                      key: token
+          nodeSelector:
+            node.kubernetes.io/instance-type: gpu-h100-sxm
+          volumes:
+            - name: devshm
+              emptyDir:
+                medium: Memory
+            - name: hf-cache
+              hostPath:
+                path: /mnt/hf-cache
+                type: DirectoryOrCreate
+```
+
+When extending an **existing** E2E step to run an additional test file, **append** that file (and any new model/deploy paths) to `source_file_dependencies` on the same step. When authoring a PR, mention the dependency list in the **Buildkite change** section of your test plan.
+
+#### L3 / L4: validating Buildkite from a feature branch (`cuda/pipeline.yml`)
+
+Production behavior is driven by [`.buildkite/cuda/pipeline.yml`](../../../.buildkite/cuda/pipeline.yml): it builds the CI image, then uploads **L2** (`test-ready.yml` on non-`main`), **L3** (`test-merge.yml` when `build.branch == "main"`), and **L4** (`test-nightly.yml` when `build.env("NIGHTLY") == "1"`).
+
+When you add or debug **L3** or **L4** tests and need those child pipelines to run **off `main`** or **without** `NIGHTLY=1`, apply **temporary** edits (revert before merge):
+
+1. **Point the upload at the right file (optional shortcut)**  
+   In the **Upload Ready Pipeline** step, the command is `buildkite-agent pipeline upload .buildkite/cuda/test-ready.yml`. For a one-off run you may change that path to `.buildkite/cuda/test-merge.yml` (L3) or `.buildkite/cuda/test-nightly.yml` (L4), and adjust the step’s `if:` so it does not conflict with the other upload steps (avoid double-uploading unless intentional).
+
+2. **L3 — merge pipeline**  
+   The gate `if: build.branch == "main"` lives on the **Upload Merge Pipeline** step in `cuda/pipeline.yml` (not inside `test-merge.yml`). **Comment out that `if` line** so `test-merge.yml` is uploaded on your feature branch. Alternatively rely on step (1) instead of the dedicated merge upload step.
+
+3. **L4 — nightly pipeline**  
+   **Comment out** `if: build.env("NIGHTLY") == "1"` on the **Upload Nightly Pipeline** step in `cuda/pipeline.yml` so the nightly definition is uploaded without setting the env var.  
+   Child steps in [`test-nightly.yml`](../../../.buildkite/cuda/test-nightly.yml) often repeat `if: build.env("NIGHTLY") == "1"`; **comment out those lines on the steps you need to run**, otherwise they will still be skipped after upload.
+
+### Step 6: Run Tests
+
+Pick one command path:
+
+- **Quick local regression (preferred first)**: single file or `file.py::test_name`
+- **CI-like level run**: marker expression + `--run-level`
+
+Full templates live in [references/test-routing.md](references/test-routing.md). **After authoring tests, always emit concrete commands** (see **Output Format**).
+
+**Typical copy-paste examples** (run from repo `tests/` directory; requires matching vLLM/vllm-omni install and hardware):
+
+| Area | Example |
+|------|---------|
+| **Omni offline L2** | `pytest -s -v e2e/offline_inference/test_qwen2_5_omni.py -m "core_model and omni and not cpu" --run-level=core_model` |
+| **Omni online L2** | `pytest -s -v e2e/online_serving/test_qwen3_omni.py -m "core_model and omni" --run-level=core_model` |
+| **TTS online L2** | `pytest -s -v e2e/online_serving/test_qwen3_tts_base.py -m "core_model and tts" --run-level=core_model` |
+| **Diffusion X2I L2 online** | `pytest -s -v e2e/online_serving/test_qwen_image.py -m "core_model and diffusion" --run-level=core_model` |
+| **Diffusion X2V L2 online** | `pytest -s -v e2e/online_serving/test_wan22_t2v.py -m "core_model and diffusion" --run-level=core_model` |
+| **Diffusion X2I L4 nightly** | `pytest -s -v e2e/online_serving/test_qwen_image_expansion.py -m "full_model and diffusion and H100" --run-level=full_model` |
+| **Diffusion X2I L4 perf (nightly)** | `pytest -s -v dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file dfx/perf/tests/test_qwen_image_vllm_omni.json` |
+| **Diffusion X2V L4 nightly** | `pytest -s -v e2e/online_serving/test_wan22_expansion.py -m "full_model and cuda" --run-level=full_model` |
+| **Invalid param (weekly H100)** | `pytest -s -v dfx/reliability/invalid_param_test/ -m "slow and H100"` |
+| **Invalid param (weekly L4)** | `pytest -s -v dfx/reliability/invalid_param_test/ -m "slow and L4"` |
+| **L1 CPU** | `pytest -s -v -m "core_model and cpu"` |
+
+**Prerequisites to mention when relevant**: GPU model (e.g. L4 vs H100), `HF_HOME` / token for hub weights, module-level `skipif` (NPU/XPU-only gaps), and whether CI already collects the path (e.g. `test_*_expansion.py` glob in `test-nightly.yml`).
+
+### Step 7: Validate Result Quality
+
+Before finishing:
+
+- Is the new assertion directly tied to the bug/feature contract?
+- Are API calls made via **`runtime.py` `send_*_request`** (not inline HTTP/SDK in the test file)?
+- Is the **general** `assert_*` bundled inside `send_*_request` (not duplicated in the test)?
+- Is each scenario a **separate `test_*` function** (no `case_id` / `if param.id` branching across cases)?
+- **Fixture scope**: default **`omni_server` / `omni_runner`** (module); **`_function`** variants only when each test must restart the instance?
+- Is the test deterministic (no fragile timing/network coupling)?
+- Is runtime appropriate for the selected level?
+- Are markers and `--run-level` consistent?
+- **Invalid-param cases** in `tests/dfx/reliability/invalid_param_test/` (not e2e), using `send_*_http_request` + `err_code`, with `pytest.mark.slow` and correct `H100` / `L4` hardware mark?
+
+## Output Format
+
+When completing a request, return:
+
+1. **Test plan** (level, markers, file target, **L4 pillar(s)** if applicable: Function / Accuracy / Perf / Doc; **or** **Invalid param** pillar with target `invalid_param_test/test_invalid_<area>.py`; module basename for e2e: `test_{slug}.py` / `test_{slug}_expansion.py`)
+2. **Generated/updated test file(s)** — model-specific constants and request payloads stay **inside** those test modules (**never** `tests/helpers/{slug}.py`). Extend **`tests/helpers/runtime.py`** / **`tests/helpers/assertions.py`** only when new **repo-wide** `send_*` or shared assert helpers are required (do not leave ad-hoc HTTP or `_assert_*` in test modules). **If L4 Perf was requested**, also list **`tests/dfx/perf/tests/*.json`**. **If invalid-param cases were requested**, list the **`invalid_param_test/` script** and parametrized `test_*` / new `id=` rows — **not** e2e paths.
+3. **Buildkite change** — **match the requested level and pillar**:
+   - **L4 Function**: `test-nightly.yml` **Function Test** shard (file list / note existing sweep). **No** `test-merge.yml` unless user also asked for L3.
+   - **L4 Perf**: `test-nightly.yml` **Perf Test · &lt;Model&gt;** step (new step or extend commands); include env exports + artifact upload pattern from a sibling perf job.
+   - **L4 Accuracy**: `test-nightly.yml` **Accuracy Test** step under the same model-type group.
+   - **L3**: `test-merge.yml` + full `source_file_dependencies` + `agents` + `plugins`.
+   - **L2**: `test-ready.yml` + same E2E block requirements.
+   - **Invalid param**: `test-weekly.yml` — note **Invalid parameters Test · H100/L4** group; usually **no YAML change** when only extending existing scripts.
+   If the user asked only for **L4 functional cases**, state explicitly that **Perf / Accuracy / Invalid param** were not included (offer to add if needed). If the plan mixes **success e2e** and **invalid param**, split deliverables across e2e vs `invalid_param_test/` and call out both CI files.
+4. **Run commands (required)** — always include, in fenced `bash` blocks:
+   - **Local — whole file**: `cd tests` then `pytest -s -v <path> …`
+   - **Local — single test** (optional but preferred when the change is one function): `pytest -s -v path::test_func …`
+   - **CI-like** (when not L1 CPU): the same **marker + `--run-level`** pairing the level uses (see [references/test-routing.md](references/test-routing.md) and **Step 6** table above)
+   - **Prerequisites** (one line): e.g. “Requires CUDA L4 + weights cached locally”; “Requires H100 + `HF_TOKEN`”; “Syntax-only validation when vllm is not installed locally”
+5. **Result summary** (pass/fail if you executed tests; if not executed, state that explicitly and what the user should run)
+
+## Additional Resources
+
+- Marker and command routing: [references/test-routing.md](references/test-routing.md)
+- CI pipelines (vllm-omni): [test-ready.yml](../../../.buildkite/cuda/test-ready.yml) (L1/L2), [test-merge.yml](../../../.buildkite/cuda/test-merge.yml) (L3), [test-nightly.yml](../../../.buildkite/cuda/test-nightly.yml) (L4), [test-weekly.yml](../../../.buildkite/cuda/test-weekly.yml) (invalid param / reliability)
+- L4 documentation example tests (naming, extraction vs copied scripts, output dirs, skips): [docs/contributing/ci/test_examples/l4_doc_example_tests.inc.md](../../../docs/contributing/ci/test_examples/l4_doc_example_tests.inc.md) — see also [PR #1910](https://github.com/vllm-project/vllm-omni/pull/1910)

.claude/skills/vllm-omni-test/references/test-routing.md

@@ -0,0 +1,395 @@
+# Test Routing Reference
+
+Use this reference to map testing goals to levels, markers, and runnable commands.
+
+**Repo paths** (`.buildkite/`, `docs/contributing/…`): link with repo-relative paths from this file (e.g. `../../../../.buildkite/cuda/test-ready.yml`, `../../../../docs/contributing/ci/test_system_overview.md`).
+
+## Model-centric e2e filename convention (L2–L4)
+
+When generating a **new** test module tied to a **specific model** under `tests/e2e/offline_inference/` or `tests/e2e/online_serving/`:
+
+| Level | Pattern | Example (`Qwen/Qwen2.5-Omni-7B`) |
+|-------|---------|----------------------------------|
+| **L2**, **L3** | `test_{lowercase_model_slug}.py` | `test_qwen2_5_omni.py` |
+| **L4** | `test_{lowercase_model_slug}_expansion.py` | `test_qwen2_5_omni_expansion.py` |
+
+**Slug rules for `{lowercase_model_slug}`** (canonical: [SKILL.md](../SKILL.md) § *Naming: generated test module files*):
+
+1. Start from the HuggingFace-style id (e.g. `Qwen/Qwen2.5-Omni-7B`), but **do not** put the org into the filename: use the **repo segment only** (`Qwen2.5-Omni-7B`), not `Qwen_...` / `qwen_qwen2_5_...`.
+2. **Lowercase**; replace `.`, `-`, and whitespace with a single `_` (e.g. `Qwen2.5-Omni-7B` → `qwen2_5_omni`). **Omit** trailing size tokens such as `7b` / `30b` in the basename when a single file covers that model line in the directory (matches `test_qwen2_5_omni.py` in-tree).
+3. If two checkpoints in the same folder need separate modules, add a **minimal** disambiguator (e.g. `_7b` vs `_3b`) **only then**.
+4. **L1** unit tests are **not** bound to this pattern; use `tests/<area>/test_<feature>.py` as today.
+
+## Model type markers (`omni` / `tts` / `diffusion`)
+
+Every model-centric e2e test must declare **exactly one** type marker:
+
+| Marker | Model family | Typical APIs |
+|--------|--------------|--------------|
+| `pytest.mark.omni` | Multimodal LLM (Qwen-Omni, …) | `send_omni_request`, `omni_runner`, stage YAML |
+| `pytest.mark.tts` | TTS / speech synthesis | `send_audio_speech_request`, `/v1/audio/speech` |
+| `pytest.mark.diffusion` | Diffusion generative models | `send_diffusion_request`, `send_video_diffusion_request`, `send_images_generations_http_request`, `OmniDiffusionSamplingParams` |
+
+**Diffusion** uses one marker for all output modalities. Nightly CI splits diffusion by **group**, not by extra markers:
+
+| Nightly group (`test-nightly.yml`) | Diffusion scope | Examples |
+|-----------------------------------|-----------------|----------|
+| **Diffusion X2I(&A&T) Model Test** | x2**i** / x2**a** / x2**t** — image, audio, text (non-video) | Qwen-Image*, BAGEL, FLUX, SD3, Z-Image, LongCat, DreamZero |
+| **Diffusion X2V Model Test** | x2**v** — video only | Wan2.2, HunyuanVideo 1.5, Wan VACE |
+
+PR labels for selective nightly runs: `diffusion-x2iat-test`, `diffusion-x2v-test` (plus `nightly-test` / `NIGHTLY=1`).
+
+## Level and Marker Mapping
+
+| Goal | Suggested Level | Marker baseline | Typical location |
+|------|------------------|-----------------|------------------|
+| Unit logic, regression on pure code path | L1 | `core_model and cpu` | `tests/<component>/test_*.py` |
+
+### L1 unit tests — mocking (`mocker`, not `unittest.mock`)
+
+L1 modules run in **Simple Test** (`test-ready.yml` / `test-merge.yml`: `-m 'core_model and cpu'`). When isolation requires doubles:
+
+- **Use** the **`mocker`** fixture (`pytest-mock`): `mocker.patch`, `mocker.spy`, `mocker.Mock`, `mocker.MagicMock`, `mocker.AsyncMock`.
+- **Use** **`monkeypatch`** for env vars and simple `setattr` / `delattr` without mock objects.
+- **Do not use** `unittest.mock` — no `from unittest.mock import ...`, no `@patch`, no `with patch(...)`.
+
+```python
+# Good (L1)
+def test_handler(mocker):
+    mocker.patch("vllm_omni.pkg.fn", return_value=0)
+
+# Bad (L1)
+from unittest.mock import patch
+
+@patch("vllm_omni.pkg.fn")
+def test_handler(mock_fn): ...
+```
+
+E2E (L2+) should not rely on mocks unless documenting a rare exception; prefer real (or lightweight) execution paths.
+| Basic integration/e2e | L2 | `core_model` + **one of** `omni` / `tts` / `diffusion` | `tests/e2e/...` |
+| Advanced integration | L3 | `advanced_model` + type marker | `tests/e2e/...` |
+| Full function / nightly | L4 | `full_model` (nightly) + type marker | **Function:** `tests/e2e/*_expansion.py`; **Perf:** `tests/dfx/perf/tests/*.json`; **Accuracy:** `tests/e2e/accuracy/` |
+| Invalid HTTP / param validation | Weekly (dfx) | `pytest.mark.slow` + type marker + `H100` or `L4` | `tests/dfx/reliability/invalid_param_test/test_invalid_*.py` |
+
+## Marker Selection Rules
+
+1. **Level** (pick one):
+   - `core_model` — L1/L2 (`test-ready.yml`)
+   - `advanced_model` — L3 (`test-merge.yml`)
+   - `full_model` — L4 nightly (`test-nightly.yml`); some expansion tests still carry both `advanced_model` and `full_model` during migration
+2. **Model type** (pick one): `omni`, `tts`, or `diffusion`
+3. **Cross-cutting** when relevant: `parallel`, `cache`, `example`, `benchmark`
+4. **Hardware**: `cpu`, `cuda`, `rocm`, `npu`, `L4`, `H100`, `distributed_cuda`, …
+5. Multi-card: `@hardware_test(...)` in `tests/helpers/mark.py`
+
+## Command Templates
+
+### Quick local checks
+
+```bash
+cd tests
+pytest -s -v test_xxxx.py
+```
+
+### L1
+
+```bash
+cd tests
+pytest -s -v -m "core_model and cpu"
+```
+
+### L2
+
+```bash
+cd tests
+pytest -s -v -m "core_model and not cpu" --run-level=core_model
+```
+
+### L3 (merge)
+
+```bash
+cd tests
+pytest -s -v -m "advanced_model" --run-level=advanced_model
+```
+
+### L4 (nightly)
+
+**Function** (e2e expansion — default when user asks for “L4 functional cases”):
+
+```bash
+cd tests
+pytest -s -v e2e/online_serving/test_qwen_image_expansion.py -m "full_model and diffusion and H100" --run-level=full_model
+```
+
+**Perf** (dfx benchmark harness — separate pillar; not `omni_server` fixtures):
+
+```bash
+cd tests
+export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
+export DIFFUSION_ATTENTION_BACKEND=FLASH_ATTN
+# CI-like: single JSON file (nightly Perf Test steps)
+pytest -s -v dfx/perf/scripts/run_diffusion_benchmark.py \
+  --test-config-file dfx/perf/tests/test_qwen_image_vllm_omni.json
+# Local bulk load: all *.json under tests/dfx/perf/tests/, filter by JSON mark
+pytest -sv dfx/perf/scripts/run_diffusion_benchmark.py -m "full_model and diffusion and H100"
+pytest -sv dfx/perf/scripts/run_benchmark.py -m "full_model and omni and H100"
+```
+
+Broad marker sweep (when no explicit file shard):
+
+```bash
+cd tests
+pytest -s -v -m "full_model" --run-level=full_model
+```
+
+## L4 nightly pillars (Function / Accuracy / Perf)
+
+Nightly **L4** for a model is often **multiple jobs** in `test-nightly.yml`, not a single pytest module:
+
+| Pillar | User says | Deliver | Nightly step |
+|--------|-----------|---------|--------------|
+| **Function** | “L4 functional cases” / “functional” / `*_expansion.py` | `tests/e2e/.../test_<model>_expansion.py` | `· Function Test with H100/L4` |
+| **Perf** | “performance” / “perf” / “benchmark” / “full L4” | `tests/dfx/perf/tests/test_<model>_vllm_omni.json` + runner | `· Perf Test · <Model>` |
+| **Accuracy** | “accuracy” / “similarity” | `tests/e2e/accuracy/test_<model>*.py` | `· Accuracy Test` |
+
+**Rule:** “L4 functional cases” → **Function only** by default. State that Perf/Accuracy are separate pillars; add them only when requested.
+
+**Diffusion perf JSON** (`tests/dfx/perf/tests/test_<slug>_vllm_omni.json`): array of **case objects**, each with:
+
+| Field | Required | Notes |
+|-------|----------|-------|
+| `test_name` | Yes | Unique server/workload id |
+| `mark` | Recommended | When present: **`hardware_marks` required** + optional `marks` (`full_model`, `diffusion`, `local_model`, …). Parsed by `resolve_pytest_marks`; enables local `-m` filtering. |
+| `server_type` | Diffusion | e.g. `"vllm-omni"` — routes case to `run_diffusion_benchmark.py` |
+| `benchmark_endpoint` | Diffusion | e.g. `/v1/videos`, `/v1/images/generations` |
+| `server_params` | Yes | `model`, `serve_args`, … |
+| `benchmark_params[]` | Yes | Each row: **`name`** (pytest id suffix), workload fields, **`baseline`** |
+
+Copy structure from `test_qwen_image_vllm_omni.json` or `test_cosmos3_vllm_omni.json` (2-GPU mark example).
+
+**Omni perf JSON**: `test_qwen3_omni_*.json` — same `mark` shape with `"marks": ["full_model", "omni"]`; run via `run_benchmark.py`.
+
+**Runner split**: `is_diffusion_perf_config()` → diffusion when `server_type` is set or `"diffusion"` ∈ `mark.marks`; otherwise omni/tts → `run_benchmark.py`.
+
+**Param ids**: `{test_name}-{benchmark_params.name}` (e.g. `test_tts-p0`, not `test_tts-0`).
+
+**Result filenames**: runtime hardware from `get_runtime_resource_label()`; `H100` omitted on default CI pool.
+
+**Diffusion perf nightly step** (under **Diffusion X2I(&A&T)**; copy full `kubernetes` plugins from `· Perf Test · Qwen-Image`):
+
+```yaml
+      - label: ":full_moon: Diffusion X2I(&A&T) · Perf Test · <Model-Display-Name>"
+        key: nightly-diffusion-x2iat-performance-<slug>
+        timeout_in_minutes: 180
+        commands:
+          - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
+          - export DIFFUSION_ATTENTION_BACKEND=FLASH_ATTN
+          - export CACHE_DIT_VERSION=1.3.0
+          - |
+            set +e
+            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_<slug>_vllm_omni.json
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/dfx/perf/results/diffusion_result_*.json"
+            buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
+            exit $$EXIT
+        agents:
+          queue: "mithril-h100-pool"
+        plugins:
+          # … full kubernetes block (often multi-GPU) — copy from Perf Test · Qwen-Image
+```
+
+| Model type | Perf runner | Config example |
+|------------|-------------|----------------|
+| Diffusion X2I/X2V | `dfx/perf/scripts/run_diffusion_benchmark.py` | `test_qwen_image_vllm_omni.json`, `test_cosmos3_vllm_omni.json` |
+| TTS | `dfx/perf/scripts/run_benchmark.py` | `test_tts.json`, `test_voxcpm2.json`, `test_higgs_audio_v3.json` |
+| Omni | `dfx/perf/scripts/run_benchmark.py` | `test_qwen3_omni_async_chunk.json`, `test_qwen3_omni_no_async_chunk.json`, … |
+
+Do **not** put throughput/latency baselines inside `test_*_expansion.py` — that belongs in the dfx perf JSON + nightly Perf job.
+
+## Invalid parameter validation (weekly / dfx)
+
+When the user’s test plan includes **invalid parameter validation**, **invalid request bodies**, or **HTTP 4xx** validation against a live server:
+
+1. **Do not** author these in `tests/e2e/online_serving/test_*.py` or `*_expansion.py`. **Move** any drafted `test_*` into `tests/dfx/reliability/invalid_param_test/`.
+2. **Pick script by route** (extend in-tree file):
+
+| Route family | Script |
+|--------------|--------|
+| Omni chat / WS video·realtime | `test_invalid_omni_chat.py` |
+| `/v1/audio/speech` (+ stream / batch / voices) | `test_invalid_audio_speech.py` |
+| Audio diffusion | `test_invalid_audio_diffusion.py` |
+| `/v1/images/generations` | `test_invalid_image_generation.py` |
+| `/v1/images/edits` | `test_invalid_image_editing.py` |
+| `/v1/videos*` | `test_invalid_video_generation.py` |
+| Sleep / wakeup / server control | `test_invalid_server_control.py` |
+
+3. **Style:** `pytestmark = [pytest.mark.slow, pytest.mark.<type>]`; `_PARAMS` + `hardware_marks`; `send_*_http_request` with `err_code` + `err_message`; parametrized `body_spec` rows with `id=`; `_minimal_*_json()` helpers; `_SKIP_ISSUE_3649` when tracked in [#3649](https://github.com/vllm-project/vllm-omni/issues/3649).
+4. **CI:** [`.buildkite/cuda/test-weekly.yml`](../../../../.buildkite/cuda/test-weekly.yml) group **Reliability Test - Invalid parameters Test** — **not** ready/merge/nightly.
+
+```bash
+# Weekly H100 (diffusion / omni / video invalid-param)
+cd tests
+pytest -s -v dfx/reliability/invalid_param_test/ -m "slow and H100"
+
+# Weekly L4 (e.g. Qwen3-TTS speech invalid-param)
+pytest -s -v dfx/reliability/invalid_param_test/ -m "slow and L4"
+
+# Single script / case
+pytest -s -v dfx/reliability/invalid_param_test/test_invalid_image_generation.py::test_images_generations_invalid_requests -m "slow and H100"
+```
+
+**Trigger:** `WEEKLY=1` or PR label `weekly-test`. Appending cases to existing scripts usually needs **no YAML edit** (directory sweep). No `source_file_dependencies` on weekly steps.
+
+### Platform-targeted examples
+
+```bash
+cd tests
+pytest -s -v -m "core_model and distributed_cuda and L4" --run-level=core_model
+```
+
+### Concrete e2e paths (common in-tree)
+
+Paths are relative to `tests/` after `cd tests`.
+
+| Scenario | Example command |
+|----------|------------------|
+| **Omni** offline L2 | `pytest -s -v e2e/offline_inference/test_qwen2_5_omni.py -m "core_model and omni and not cpu" --run-level=core_model` |
+| **Omni** offline L2 — one test | `pytest -s -v e2e/offline_inference/test_qwen2_5_omni.py::test_text_to_text -m "core_model and omni and not cpu" --run-level=core_model` |
+| **Omni** online L2 | `pytest -s -v e2e/online_serving/test_qwen3_omni.py -m "core_model and omni" --run-level=core_model` |
+| **Omni** offline L2 (Qwen3.5-9B VL) | `pytest -s -v e2e/offline_inference/test_qwen3_5_9b.py -m "core_model and omni and not cpu" --run-level=core_model` |
+| **TTS** online L2 | `pytest -s -v e2e/online_serving/test_qwen3_tts_base.py -m "core_model and tts" --run-level=core_model` |
+| **TTS** online L2 — one test | `pytest -s -v e2e/online_serving/test_qwen3_tts_base.py::test_text_to_audio_001 -m "core_model and tts" --run-level=core_model` |
+| **Diffusion X2I** offline L2 | `pytest -s -v e2e/offline_inference/test_t2i_model.py -m "core_model and diffusion and not cpu" --run-level=core_model` |
+| **Diffusion X2I** L2 online smoke | `pytest -s -v e2e/online_serving/test_qwen_image.py e2e/online_serving/test_bagel.py -m "core_model and diffusion" --run-level=core_model` |
+| **Diffusion X2V** L2 online smoke | `pytest -s -v e2e/online_serving/test_wan22_t2v.py -m "core_model and diffusion" --run-level=core_model` |
+| **Diffusion X2I** L4 online expansion (H100) | `pytest -s -v e2e/online_serving/test_qwen_image_expansion.py -m "full_model and diffusion and H100" --run-level=full_model` |
+| **Diffusion X2I** L4 perf (nightly) | `pytest -s -v dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file dfx/perf/tests/test_qwen_image_vllm_omni.json` |
+| **Diffusion X2I** L4 online expansion (Edit) | `pytest -s -v e2e/online_serving/test_qwen_image_edit_expansion.py -m "full_model and diffusion and H100" --run-level=full_model` |
+| **Diffusion X2V** L4 nightly | `pytest -s -v e2e/online_serving/test_wan22_expansion.py -m "full_model and cuda" --run-level=full_model` |
+| **Invalid param** weekly H100 | `pytest -s -v dfx/reliability/invalid_param_test/ -m "slow and H100"` |
+| **Invalid param** weekly L4 | `pytest -s -v dfx/reliability/invalid_param_test/ -m "slow and L4"` |
+
+### Agent / author completion checklist
+
+When adding or modifying tests, do not stop at “where the file lives” — also deliver:
+
+1. **Local**: `cd tests` + `pytest -s -v <path>` (and `path::test_func` when a single case is enough).
+2. **CI-like**: marker string + `--run-level` matching the pipeline (`core_model` / `advanced_model` / `full_model`).
+3. **L1 mocks**: `mocker` / `monkeypatch` only; no `unittest.mock`.
+4. **API client + assert placement**: **General** validation → implement in `assertions.py`, call **inside** `send_*_request` in `runtime.py`; tests call **`send_*_request` only**. **Special** case validation → `assert_*` in `assertions.py`, called **in the test** after `send_*_request`. Low-level `send_*_http_request` is for negative/dfx tests (`err_code`), not ordinary L2+ success e2e.
+5. **Shared assertions**: logic in **`tests/helpers/assertions.py`**; general checks inside `send_*_request`; special checks only in the test. No `_assert_*` in `test_*.py`.
+6. **One case → one `test_*`**: function name reflects what is validated; no `if case_id == ...` mega-test merging multiple scenarios.
+7. **Fixture scope**: default **`omni_server` + `openai_client`** / **`omni_runner` + `omni_runner_handler`** (module). Use **`omni_server_function` + `openai_client_function`** / **`omni_runner_function` + `omni_runner_handler_function`** only when each `test_*` must spawn a fresh instance. Parametrize name must match fixture (`omni_server` vs `omni_server_function`).
+8. **Type marker**: `omni`, `tts`, or `diffusion` on every model e2e module.
+9. **Diffusion L4 Function**: wire `*_expansion.py` into **X2I(&A&T)** or **X2V** **Function Test** in **`test-nightly.yml` only** — do not add `test-merge.yml` unless the user also requested L3.
+10. **Diffusion L4 Perf** (only when requested): add `tests/dfx/perf/tests/test_<slug>_vllm_omni.json` + **Perf Test · &lt;Model&gt;** step (artifact upload); not part of “L4 functional cases” by default.
+11. **E2E Buildkite (L2/L3 only)**: `test-ready.yml` / `test-merge.yml` steps need **`source_file_dependencies`** + full **`agents` + `plugins`**. **L4 nightly** uses explicit file lists or perf scripts in `test-nightly.yml` (no merge job).
+12. **Invalid param**: cases in **`tests/dfx/reliability/invalid_param_test/`** (route-matching script), `send_*_http_request` + `err_code`, `pytest.mark.slow` + `H100`/`L4`; CI = **`test-weekly.yml`** only — do not put in e2e or nightly.
+13. **Prerequisites**: GPU tier, HF cache/token, and any module `skipif` / platform-only YAML.
+
+## Buildkite pipeline mapping
+
+| Level | Repo file | Model-type grouping |
+|-------|-----------|---------------------|
+| L1, L2 | [`.buildkite/cuda/test-ready.yml`](../../../../.buildkite/cuda/test-ready.yml) | Steps prefixed **Omni ·**, **TTS ·**, **Diffusion ·** under **E2E Test** — **`source_file_dependencies` required** |
+| L3 | [`.buildkite/cuda/test-merge.yml`](../../../../.buildkite/cuda/test-merge.yml) | Per-model E2E steps; `-m "advanced_model and …"` — **`source_file_dependencies` required** |
+| L4 | [`.buildkite/cuda/test-nightly.yml`](../../../../.buildkite/cuda/test-nightly.yml) | **Omni / TTS / Diffusion X2I(&A&T) / X2V** — each group may have **Function**, **Accuracy**, **Perf**, **Doc** steps |
+| Invalid param (weekly) | [`.buildkite/cuda/test-weekly.yml`](../../../../.buildkite/cuda/test-weekly.yml) | **Invalid parameters Test · H100** / **· L4** — sweeps `tests/dfx/reliability/invalid_param_test/` |
+
+### `source_file_dependencies` (E2E Test only — ready & merge)
+
+Required on each step inside the **E2E Test** group in `test-ready.yml` and `test-merge.yml`. Typical entries:
+
+| Category | Path pattern |
+|----------|----------------|
+| Test module | `tests/e2e/online_serving/test_<slug>.py`, `tests/e2e/offline_inference/test_<slug>.py` |
+| API client helpers | `tests/helpers/runtime.py` (when the e2e job uses newly added `send_*` methods) |
+| Shared assert helpers | `tests/helpers/assertions.py` (when the e2e job uses newly added `assert_*` helpers) |
+| AR / omni model | `vllm_omni/model_executor/models/<family>/` |
+| Diffusion model | `vllm_omni/diffusion/models/<family>/` |
+| Stage processor | `vllm_omni/model_executor/stage_input_processors/<family>.py` |
+| Deploy config | `vllm_omni/deploy/<name>.yaml` or `vllm_omni/deploy/ci/<name>.yaml` |
+
+Copy the dependency block from the closest in-tree E2E step for the same model family. Update the list whenever the pytest command or server YAML changes.
+
+### E2E `agents` + `plugins` blocks (required — do not truncate)
+
+When generating or documenting E2E steps, **always include the complete `plugins` section**. Set **`timeout_in_minutes`** on the step for the job deadline; run **`pytest` directly** in `commands` (no `timeout 40m bash -c` wrapper around pytest).
+
+Common patterns in `test-ready.yml` / `test-merge.yml`:
+
+| Queue | Plugin | Required fields |
+|-------|--------|-----------------|
+| `mithril-h100-pool` | `kubernetes` | `resources.limits.nvidia.com/gpu`, `volumeMounts` (`devshm`, `hf-cache`), `env` (`HF_HOME`, `HF_TOKEN` via `secretKeyRef`), `nodeSelector: gpu-h100-sxm`, `volumes` (`devshm` emptyDir, `hf-cache` hostPath `/mnt/hf-cache`) |
+| `gpu_1_queue` / `gpu_4_queue` | `docker#v5.2.0` | `image`, `always-pull`, `propagate-environment`, `shm-size: "8gb"`, `environment: HF_HOME`, `volumes: /fsx/hf_cache` (add `HF_TOKEN` when the step needs hub access) |
+
+**H100 kubernetes template** (copy from **Diffusion · Bagel Test** / **Diffusion · Qwen Image Test** in-tree):
+
+```yaml
+  agents:
+    queue: "mithril-h100-pool"
+  plugins:
+    - kubernetes:
+        podSpec:
+          containers:
+            - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
+              resources:
+                limits:
+                  nvidia.com/gpu: 1
+              volumeMounts:
+                - name: devshm
+                  mountPath: /dev/shm
+                - name: hf-cache
+                  mountPath: /root/.cache/huggingface
+              env:
+                - name: HF_HOME
+                  value: /root/.cache/huggingface
+                - name: HF_TOKEN
+                  valueFrom:
+                    secretKeyRef:
+                      name: hf-token-secret
+                      key: token
+          nodeSelector:
+            node.kubernetes.io/instance-type: gpu-h100-sxm
+          volumes:
+            - name: devshm
+              emptyDir:
+                medium: Memory
+            - name: hf-cache
+              hostPath:
+                path: /mnt/hf-cache
+                type: DirectoryOrCreate
+```
+
+**L4 docker template** (copy from **TTS · Qwen3-TTS Base Test** in `test-merge.yml`):
+
+```yaml
+  agents:
+    queue: "gpu_1_queue"
+  plugins:
+    - docker#v5.2.0:
+        image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
+        always-pull: true
+        propagate-environment: true
+        shm-size: "8gb"
+        environment:
+          - "HF_HOME=/fsx/hf_cache"
+        volumes:
+          - "/fsx/hf_cache:/fsx/hf_cache"
+```
+
+When extending an existing step (e.g. add `tests/e2e/offline_inference/test_qwen_image.py` to **Diffusion · Qwen Image Test**), update `source_file_dependencies` and `commands` only; **keep the existing `agents` + `plugins` block unchanged** unless the hardware tier changes.
+
+[`.buildkite/cuda/pipeline.yml`](../../../../.buildkite/cuda/pipeline.yml) decides **which** child file is uploaded. To run **L3** on a feature branch, comment out `if: build.branch == "main"` on the merge upload step. To run **L4** without `NIGHTLY=1`, comment out the nightly upload step’s `if: build.env("NIGHTLY") == "1"` and the same `if` on relevant steps inside `test-nightly.yml`. Revert such edits before merging.
+
+Platform-specific pipelines (e.g. AMD) follow the same level → file pairing under `.buildkite/`.
+
+## Diffusion RFC (#1832) Alignment Tips
+
+For **X2I(&A&T)** coverage planning:
+
+- Prioritize high-value feature combinations with minimal case count.
+- Split into lightweight L1/L2 validation plus L4 `*_expansion.py` under the **X2I** nightly shards.
+- **X2V** models (Wan, HunyuanVideo) stay in the **X2V** group — do not merge into X2I sweeps.
+
+If hardware is insufficient, provide an executable reduced case plus a deferred full CI/nightly plan.

.gitattributes

@@ -27,6 +27,7 @@
 *.mov binary
 *.mp4 binary
 *.mp3 binary
+*.wav binary
 *.flv binary
 *.fla binary
 *.swf binary

.github/CODEOWNERS

@@ -3,15 +3,64 @@
 # More info: https://docs.github.com/en/repositories/managing-your-repositorys-settings-and-features/customizing-your-repository/about-code-owners
 
 # Default owners (fallback reviewers for all files)
-# later, we can add more reviewers here
-* @hsliuustc0106
+* @Gaohan123 @hsliuustc0106 @ywang96
 
-# Assign reviewers by directory
-/.github/ @owner1
-/tests/ @owner1
-/docs/ @owner1
+# Repository operations
+/.github/ @hsliuustc0106
 
-# Example: reviewers for specific directories
-# /examples/offline_inference/ @offline-inference-team
-# /vllm_omni/core/ @core-team
-# /vllm_omni/model_executor/ @model-team
+# Documentation and Recipes
+/docs/ @hsliuustc0106 @Gaohan123 @david6666666
+/recipes @hsliuustc0106 @Gaohan123 @david6666666
+
+# Skills
+/.claude/ @hsliuustc0106
+
+# Tests
+/tests/ @yenuo26
+
+# CI
+/.buildkite/ @yenuo26 @congw729
+
+# Engine and model_executor
+/vllm_omni/core/ @tzhouam @yinpeiqi @fake0fan @Sy0307
+/vllm_omni/engine/ @tzhouam @yinpeiqi @fake0fan @Sy0307
+/vllm_omni/entrypoints/ @tzhouam @yinpeiqi @fake0fan @alex-jw-brooks
+/vllm_omni/model_executor/ @tzhouam @gcanlin
+/vllm_omni/worker/ @tzhouam @gcanlin
+
+# Diffusion
+/vllm_omni/diffusion/ @Isotr0py @princepride @SamitHuang @wtomin @ZJY0516 @RuixiangMa @david6666666 @xuechendi
+/vllm_omni/lora/ @princepride @SamitHuang @wtomin @ZJY0516 @RuixiangMa @yuanheng-zhao
+
+# Diffusion CPU offloading
+/vllm_omni/diffusion/offloader/ @Isotr0py @princepride @SamitHuang @wtomin @ZJY0516 @RuixiangMa @yuanheng-zhao @xuechendi
+
+# Omni and TTS support
+/vllm_omni/transformers_utils/ @tzhouam @yuanheng-zhao
+/vllm_omni/model_executor/models/ @linyueqian @ZeldaHuang @princepride @yuanheng-zhao @amy-why-3459 @Bounty-hunter @Sy0307 @gcanlin
+
+# Benchmarks
+/benchmarks/accuracy/ @david6666666
+/benchmarks/build_dataset/ @linyueqian @yuanheng-zhao
+/benchmarks/diffusion/ @Isotr0py @princepride @SamitHuang @wtomin @ZJY0516 @RuixiangMa @xuechendi
+/benchmarks/distributed/ @princepride
+/benchmarks/glm_image/ @princepride @JaredforReal
+/benchmarks/tts/ @linyueqian @ZeldaHuang @princepride @Sy0307 @gcanlin
+
+# Usability and Observability
+/vllm_omni/deploy/ @tzhouam @lishunyang12
+/vllm_omni/metrics/ @lishunyang12
+/vllm_omni/profiler/ @gcanlin
+
+# ComfyUI
+/apps/ @fhfuih @alex-jw-brooks
+
+# Large Scale Serving
+/vllm_omni/distributed/ @princepride @yuanheng-zhao @xuechendi @natureofnature @fake0fan
+
+# Quantization and configuration
+/vllm_omni/config/ @lishunyang12 @alex-jw-brooks
+/vllm_omni/quantization/ @david6666666 @Isotr0py @lishunyang12
+
+# Hardware plugins and accelerators
+/vllm_omni/platforms/ @gcanlin @tjtanaa @xuechendi

.github/ISSUE_TEMPLATE/400-bug-report.yml

@@ -19,6 +19,7 @@ body:
 
       Consider redacting or replacing sensitive values with placeholders like `<YOUR_TOKEN_HERE>` when sharing configuration or code examples.
 - type: textarea
+  id: current-environment
   attributes:
     label: Your current environment
     description: |
@@ -42,6 +43,7 @@ body:
     required: true
 
 - type: textarea
+  id: code-version
   attributes:
     label: Your code version
     description: |
@@ -66,6 +68,7 @@ body:
     required: true
 
 - type: textarea
+  id: bug-description
   attributes:
     label: 🐛 Describe the bug
     description: |

.github/PULL_REQUEST_TEMPLATE.md

@@ -1,21 +1,15 @@
 <!-- markdownlint-disable -->
-PLEASE FILL IN THE PR DESCRIPTION HERE ENSURING ALL CHECKLIST ITEMS (AT THE BOTTOM) HAVE BEEN CONSIDERED.
+PLEASE FILL IN THE PR DESCRIPTION HERE.
 
 ## Purpose
 
 ## Test Plan
 
+**vLLM Version:**
+
+**vLLM-Omni Commit:**
+
 ## Test Result
 
----
-<details>
-<summary> Essential Elements of an Effective PR Description Checklist </summary>
-
-- [ ] The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
-- [ ] The test plan. Please provide the test scripts & test commands. Please state the reasons if your codes don't require additional test scripts. For test file guidelines, please check the [test style doc](https://docs.vllm.ai/projects/vllm-omni/en/latest/contributing/ci/tests_style/)
-- [ ] The test results. Please paste the results comparison before and after, or the e2e results.
-- [ ] (Optional) The necessary documentation update, such as updating `supported_models.md` and `examples` for a new model. **Please run `mkdocs serve` to sync the documentation editions to `./docs`.**
-- [ ] (Optional) Release notes update. If your change is user-facing, please update the release notes draft.
-</details>
-
-**BEFORE SUBMITTING, PLEASE READ <https://github.com/vllm-project/vllm-omni/blob/main/CONTRIBUTING.md>** (anything written below this line will be removed by GitHub Actions)
+**BEFORE SUBMITTING:** read [CONTRIBUTING.md](https://github.com/vllm-project/vllm-omni/blob/main/CONTRIBUTING.md) and run the [precheck-pr skill](https://github.com/vllm-project/vllm-omni/blob/main/.claude/skills/precheck-pr/SKILL.md) with the code agent for a self-check against project conventions.
+(anything written below this line will be removed by GitHub Actions)

.github/mergify.yml

@@ -0,0 +1,223 @@
+pull_request_rules:
+- name: label-documentation
+  description: Automatically apply documentation label
+  conditions:
+    - label != stale
+    - or:
+      - files~=^[^/]+\.md$
+      - files~=^docs/
+      - files~=^examples/
+  actions:
+    label:
+      add:
+        - documentation
+    comment:
+      message: "Documentation preview: https://vllm--{{number}}.org.readthedocs.build/en/{{number}}/"
+
+- name: comment-pre-commit-failure
+  description: Comment on PR when pre-commit check fails
+  conditions:
+    - status-failure=pre-commit
+    - -closed
+    - -draft
+  actions:
+    comment:
+      message: |
+        Hi @{{author}}, the pre-commit checks have failed. Please run:
+
+        ```bash
+        uv pip install pre-commit
+        pre-commit install
+        pre-commit run --all-files
+        ```
+
+        Then, commit the changes and push to your branch.
+
+        For future commits, `pre-commit` will run automatically on changed files before each commit.
+
+        > [!TIP]
+        > <details>
+        > <summary>Is <code>mypy</code> or <code>markdownlint</code> failing?</summary>
+        > <br/>
+        > <code>mypy</code> and <code>markdownlint</code> are run differently in CI. If the failure is related to either of these checks, please use the following commands to run them locally:
+        >
+        > ```bash
+        > # For mypy (substitute "3.10" with the failing version if needed)
+        > pre-commit run --hook-stage manual mypy-3.10
+        > # For markdownlint
+        > pre-commit run --hook-stage manual markdownlint
+        > ```
+        > </details>
+
+- name: comment-dco-failure
+  description: Comment on PR when DCO check fails
+  conditions:
+    - status-failure=dco
+    - -closed
+    - -draft
+  actions:
+    comment:
+      message: |
+        Hi @{{author}}, the DCO check has failed. Please click on DCO in the Checks section for instructions on how to resolve this.
+
+- name: label-ci-build
+  description: Automatically apply ci/build label
+  conditions:
+    - label != stale
+    - or:
+      - files~=^\.github/
+      - files~=\.buildkite/
+      - files~=^cmake/
+      - files=CMakeLists.txt
+      - files~=^docker/Dockerfile
+      - files~=^requirements.*\.txt
+      - files=setup.py
+  actions:
+    label:
+      add:
+        - ci/build
+
+- name: label-frontend
+  description: Automatically apply frontend label
+  conditions:
+    - label != stale
+    - files~=^vllm_omni/entrypoints/
+  actions:
+    label:
+      add:
+        - frontend
+
+- name: label-diffusion
+  description: Automatically apply diffusion label
+  conditions:
+    - label != stale
+    - files~=^vllm_omni/diffusion/
+  actions:
+    label:
+      add:
+        - diffusion
+
+- name: label-new-model
+  description: Automatically apply new-model label
+  conditions:
+    - label != stale
+    - or:
+        - files=vllm_omni/model_executor/models/registry.py
+        - files=vllm_omni/diffusion/registry.py
+  actions:
+    label:
+      add:
+        - new-model
+
+- name: label-npu
+  description: Automatically apply npu label
+  conditions:
+    - label != stale
+    - or:
+      - files~=^vllm_omni/platforms/npu/
+      - title~=(?i)NPU
+      - title~=(?i)Ascend
+  actions:
+    label:
+      add:
+        - npu
+    assign:
+      users:
+        - "gcanlin"
+
+- name: label-rocm
+  description: Automatically apply rocm label
+  conditions:
+    - label != stale
+    - or:
+      - files~=^vllm_omni/platforms/rocm/
+      - title~=(?i)AMD
+      - title~=(?i)ROCm
+  actions:
+    label:
+      add:
+        - rocm
+    assign:
+      users:
+        - "tjtanaa"
+
+- name: label-xpu
+  description: Automatically apply xpu label
+  conditions:
+    - label != stale
+    - or:
+      - files~=^vllm_omni/platforms/xpu/
+      - title~=(?i)Intel
+      - title~=(?i)XPU
+  actions:
+    label:
+      add:
+        - xpu
+    assign:
+      users:
+        - "yma11"
+
+- name: label-modelrunner
+  description: Automatically apply modelrunner label
+  conditions:
+    - label != stale
+    - or:
+      - files~=^vllm_omni/worker/
+      - title~=(?i)ModelRunner
+      - title~=(?i)worker
+  actions:
+    label:
+      add:
+        - ModelRunner
+    assign:
+      users:
+        - "gcanlin" # To check whether need to sync to NPU
+
+- name: auto-rebase if approved, ready, and 30 commits behind main
+  conditions:
+    - base = main
+    - label=ready
+    - "#approved-reviews-by >= 1"
+    - "#commits-behind >= 30"
+    - -closed
+    - -draft
+    - -conflict
+  actions:
+    rebase: {}
+
+- name: ping author on conflicts and add 'needs-rebase' label
+  conditions:
+    - label != stale
+    - conflict
+    - -closed
+  actions:
+    label:
+      add:
+        - needs-rebase
+    comment:
+      message: |
+       This pull request has merge conflicts that must be resolved before it can be
+       merged. Please rebase the PR, @{{author}}.
+
+       https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/working-with-forks/syncing-a-fork
+
+- name: remove 'needs-rebase' label when conflict is resolved
+  conditions:
+    - -conflict
+    - -closed
+  actions:
+    label:
+      remove:
+        - needs-rebase
+
+- name: label-bug
+  description: Automatically apply bug label
+  conditions:
+    - label != stale
+    - or:
+      - title~=(?i)\bbug\b
+      - title~=(?i)\bbugfix\b
+  actions:
+    label:
+      add:
+        - bug

.github/scripts/pr_reviewer.py

@@ -1,629 +0,0 @@
-#!/usr/bin/env python3
-"""
-PR Reviewer using GLM API for vllm-omni project.
-"""
-
-import json
-import logging
-import os
-import sys
-import time
-from dataclasses import dataclass
-from typing import Any, TypedDict
-
-import requests
-
-
-# Type definitions for API responses
-class PRDetails(TypedDict):
-    """Type definition for GitHub PR details response."""
-
-    title: str
-    body: str
-    number: int
-    state: str
-    user: dict[str, Any]
-
-
-class GLMMessage(TypedDict):
-    """Type definition for GLM API message."""
-
-    role: str
-    content: str
-
-
-class GLMChoice(TypedDict):
-    """Type definition for GLM API choice."""
-
-    message: GLMMessage
-    finish_reason: str
-
-
-class GLMResponse(TypedDict):
-    """Type definition for GLM API response."""
-
-    choices: list[GLMChoice]
-    usage: dict[str, int] | None
-
-
-class GitHubComment(TypedDict):
-    """Type definition for GitHub comment."""
-
-    id: int
-    body: str
-    created_at: str
-    user: dict[str, Any]
-
-
-# Configuration
-TRIGGER_PHRASE: str = "@vllm-omni-reviewer"
-DEFAULT_GLM_API_URL: str = "https://open.bigmodel.cn/api/paas/v4/chat/completions"  # noqa: E501
-DEFAULT_GLM_MODEL: str = "glm-5"
-DEFAULT_COOLDOWN_MINUTES: int = 5
-DEFAULT_MAX_RETRIES: int = 3
-DEFAULT_RETRY_DELAY: float = 1.0
-MAX_DIFF_SIZE: int = 100_000  # Maximum diff size in characters
-
-
-@dataclass
-class Config:
-    """Configuration for the PR reviewer."""
-
-    glm_api_url: str
-    glm_model: str
-    cooldown_minutes: int
-    max_retries: int
-    retry_delay: float
-    max_diff_size: int
-
-
-# Setup logging
-logging.basicConfig(
-    level=logging.INFO,
-    format="[PR Reviewer] %(asctime)s - %(levelname)s - %(message)s",
-    datefmt="%Y-%m-%d %H:%M:%S",
-)
-logger: logging.Logger = logging.getLogger(__name__)
-
-
-def get_config() -> Config:
-    """Load configuration from environment variables with defaults."""
-    return Config(
-        glm_api_url=os.getenv("GLM_API_URL", DEFAULT_GLM_API_URL),
-        glm_model=os.getenv("GLM_MODEL", DEFAULT_GLM_MODEL),
-        cooldown_minutes=int(
-            os.getenv(
-                "PR_REVIEWER_COOLDOWN_MINUTES",
-                str(DEFAULT_COOLDOWN_MINUTES),
-            )
-        ),
-        max_retries=int(
-            os.getenv(
-                "PR_REVIEWER_MAX_RETRIES",
-                str(DEFAULT_MAX_RETRIES),
-            )
-        ),
-        retry_delay=float(os.getenv("PR_REVIEWER_RETRY_DELAY", str(DEFAULT_RETRY_DELAY))),
-        max_diff_size=int(os.getenv("PR_REVIEWER_MAX_DIFF_SIZE", str(MAX_DIFF_SIZE))),  # noqa: E501
-    )
-
-
-def get_env_var(name: str) -> str:
-    """
-    Get an environment variable or raise an error.
-
-    Args:
-        name: Name of the environment variable.
-
-    Returns:
-        The value of the environment variable.
-
-    Raises:
-        SystemExit: If the environment variable is not set.
-    """
-    value = os.environ.get(name)
-    if not value:
-        logger.error(f"Environment variable {name} is not set")
-        sys.exit(1)
-    return value
-
-
-def check_trigger(comment_body: str) -> bool:
-    """
-    Check if the comment contains the trigger phrase.
-
-    Args:
-        comment_body: The body of the comment to check.
-
-    Returns:
-        True if the trigger phrase is found, False otherwise.
-    """
-    return TRIGGER_PHRASE in comment_body
-
-
-def fetch_pr_diff(
-    repo_name: str,
-    pr_number: int,
-    token: str,
-    max_size: int = MAX_DIFF_SIZE,
-) -> str | None:
-    """
-    Fetch the diff for a pull request.
-
-    Args:
-        repo_name: The repository name in format "owner/repo".
-        pr_number: The pull request number.
-        token: GitHub authentication token.
-        max_size: Maximum diff size in characters.
-
-    Returns:
-        The diff content as a string, or None if fetching failed.
-        Returns empty string if diff is larger than max_size.
-    """
-    url: str = f"https://api.github.com/repos/{repo_name}/pulls/{pr_number}"
-    headers: dict[str, str] = {
-        "Authorization": f"Bearer {token}",
-        "Accept": "application/vnd.github.v3.diff",
-    }
-
-    logger.info(f"Fetching PR diff from {url}")
-    response = requests.get(url, headers=headers, timeout=30)
-
-    if response.status_code == 200:
-        diff: str = response.text
-        if len(diff) > max_size:
-            logger.warning(
-                f"Diff size ({len(diff)} bytes) exceeds maximum "
-                f"({max_size} bytes), truncating to first "
-                f"{max_size} bytes"
-            )
-            return diff[:max_size] + "\n\n... [Diff truncated due to size] ..."
-        logger.info(f"Successfully fetched diff ({len(diff)} bytes)")
-        return diff
-    else:
-        logger.error(f"Failed to fetch PR diff: {response.status_code}")
-        logger.error(f"Response: {response.text}")
-        return None
-
-
-def fetch_pr_details(
-    repo_name: str,
-    pr_number: int,
-    token: str,
-) -> PRDetails | None:
-    """
-    Fetch PR details including title and description.
-
-    Args:
-        repo_name: The repository name in format "owner/repo".
-        pr_number: The pull request number.
-        token: GitHub authentication token.
-
-    Returns:
-        A dictionary containing PR details, or None if fetching failed.
-    """
-    url: str = f"https://api.github.com/repos/{repo_name}/pulls/{pr_number}"
-    headers: dict[str, str] = {
-        "Authorization": f"Bearer {token}",
-        "Accept": "application/vnd.github.v3+json",
-    }
-
-    logger.info(f"Fetching PR details from {url}")
-    response = requests.get(url, headers=headers, timeout=30)
-
-    if response.status_code == 200:
-        return response.json()
-    else:
-        logger.error(f"Failed to fetch PR details: {response.status_code}")
-        return None
-
-
-def build_review_prompt(pr_title: str, pr_description: str, diff: str) -> str:
-    """
-    Build the prompt for the GLM-4.7 API.
-
-    Args:
-        pr_title: The title of the pull request.
-        pr_description: The description/body of the pull request.
-        diff: The diff content of the pull request.
-
-    Returns:
-        The formatted prompt string for the API.
-    """
-    return f"""You are an expert code reviewer for the VLLM-Omni project. \
-Please review the following pull request:
-
-## Pull Request Details
-**Title:** {pr_title}
-
-**Description:**
-{pr_description if pr_description else "No description provided."}
-
-## Code Changes (Diff)
-{diff}
-
-## Review Guidelines
-
-Please provide a comprehensive code review with the following sections:
-
-### 1. Overview
-- Brief summary of the changes
-- Overall assessment (positive, neutral, or concerns)
-
-### 2. Code Quality
-- Code style and consistency
-- Potential bugs or edge cases
-- Performance considerations
-- Error handling
-
-### 3. Architecture & Design
-- Integration with existing codebase
-- Design patterns and best practices
-- Potential improvements
-
-### 4. Security & Safety
-- Security concerns (if any)
-- Resource management
-- Input validation
-
-### 5. Testing & Documentation
-- Test coverage considerations
-- Documentation completeness
-- Examples and usage clarity
-
-### 6. Specific Suggestions
-- Line-by-line specific feedback (use `file:line` format)
-- Concrete actionable suggestions
-- Code examples for improvements (if applicable)
-
-### 7. Approval Status
-- **LGTM** (Looks Good To Me) if the PR is ready to merge
-- **LGTM with suggestions** if the PR is good but has minor suggestions
-- **Changes requested** if significant changes are needed
-
-## Important Notes
-- Be constructive and helpful
-- Focus on objective technical feedback
-- Acknowledge good practices when you see them
-- Prioritize critical issues over nitpicks
-- If the diff is empty or minimal, acknowledge this and provide
-  any relevant context-specific guidance
-
-Please format your response in Markdown with clear section headers.
-"""
-
-
-def validate_glm_response(data: dict[str, Any]) -> str | None:
-    """
-    Validate and extract content from GLM API response.
-
-    Args:
-        data: The response data from GLM API.
-
-    Returns:
-        The review content string if valid, None otherwise.
-    """
-    # Check if choices exists and is a non-empty list
-    if "choices" not in data:
-        logger.error("GLM API response missing 'choices' field")
-        logger.error(f"Response structure: {json.dumps(data, indent=2)}")
-        return None
-
-    choices = data["choices"]
-    if not isinstance(choices, list):
-        logger.error(f"GLM API 'choices' is not a list: {type(choices)}")
-        return None
-
-    if len(choices) == 0:
-        logger.error("GLM API 'choices' is an empty list")
-        return None
-
-    # Check if first choice has message
-    try:
-        first_choice = choices[0]
-        if not isinstance(first_choice, dict):
-            logger.error(f"GLM API choice is not a dict: {type(first_choice)}")
-            return None
-
-        if "message" not in first_choice:
-            logger.error("GLM API choice missing 'message' field")
-            logger.error(f"Choice structure: {json.dumps(first_choice, indent=2)}")  # noqa: E501
-            return None
-
-        message = first_choice["message"]
-        if not isinstance(message, dict):
-            logger.error(f"GLM API message is not a dict: {type(message)}")
-            return None
-
-        if "content" not in message:
-            logger.error("GLM API message missing 'content' field")
-            logger.error(f"Message structure: {json.dumps(message, indent=2)}")
-            return None
-
-        content = message["content"]
-        if not isinstance(content, str):
-            logger.error(f"GLM API content is not a string: {type(content)}")
-            return None
-
-        return content
-
-    except (KeyError, IndexError, TypeError) as e:
-        logger.error(f"Failed to parse GLM API response: {e}")
-        logger.error(f"Response: {json.dumps(data, indent=2)}")
-        return None
-
-
-def call_glm_api(prompt: str, api_key: str, config: Config) -> str | None:
-    """
-    Call the GLM-4.7 API to get code review with retry logic.
-
-    Args:
-        prompt: The prompt to send to the API.
-        api_key: The GLM API key.
-        config: Configuration object.
-
-    Returns:
-        The review content as a string, or None if all retries failed.
-    """
-    headers: dict[str, str] = {
-        "Authorization": f"Bearer {api_key}",
-        "Content-Type": "application/json",
-    }
-
-    payload: dict[str, Any] = {
-        "model": config.glm_model,
-        "messages": [{"role": "user", "content": prompt}],
-        "temperature": 0.3,
-        "max_tokens": 32000,
-        "top_p": 0.9,
-    }
-
-    last_error: str | None = None
-
-    for attempt in range(config.max_retries):
-        try:
-            logger.info(f"Calling GLM API ({config.glm_model}) - Attempt {attempt + 1}/{config.max_retries}")
-            response = requests.post(
-                config.glm_api_url,
-                headers=headers,
-                json=payload,
-                timeout=120,
-            )
-
-            if response.status_code == 200:
-                data = response.json()
-                review = validate_glm_response(data)
-                if review:
-                    logger.info(f"Successfully received review ({len(review)} chars)")  # noqa: E501
-                    return review
-                else:
-                    last_error = "Failed to validate API response structure"
-                    logger.error(last_error)
-            else:
-                last_error = f"GLM API request failed: {response.status_code} - {response.text}"
-                logger.error(last_error)
-
-        except requests.exceptions.Timeout:
-            last_error = f"GLM API request timed out (attempt {attempt + 1})"
-            logger.error(last_error)
-        except requests.exceptions.RequestException as e:
-            last_error = f"GLM API request exception: {e}"
-            logger.error(last_error)
-        except json.JSONDecodeError as e:
-            last_error = f"Failed to decode GLM API response as JSON: {e}"
-            logger.error(last_error)
-
-        # Exponential backoff before retry
-        if attempt < config.max_retries - 1:
-            wait_time: float = config.retry_delay * (2**attempt)
-            logger.info(f"Waiting {wait_time}s before retry...")  # noqa: E501
-            time.sleep(wait_time)
-
-    logger.error(
-        f"All {config.max_retries} attempts failed. Last error: {last_error}"  # noqa: E501
-    )
-    return None
-
-
-def check_cooldown(  # noqa: E501
-    repo_name: str,
-    pr_number: int,
-    token: str,
-    cooldown_minutes: int,
-) -> bool:
-    """
-    Check if the PR is within the cooldown period.
-
-    Args:
-        repo_name: The repository name in format "owner/repo".
-        pr_number: The pull request number.
-        token: GitHub authentication token.
-        cooldown_minutes: Cooldown period in minutes.
-
-    Returns:
-        True if within cooldown period (should skip), False otherwise.
-    """
-    from datetime import datetime, timedelta
-
-    url: str = (
-        f"https://api.github.com/repos/{repo_name}/issues/"
-        f"{pr_number}/comments"  # noqa: E501
-    )
-    headers: dict[str, str] = {
-        "Authorization": f"Bearer {token}",
-        "Accept": "application/vnd.github.v3+json",
-    }
-
-    logger.info(f"Checking cooldown period ({cooldown_minutes} minutes)")
-    response = requests.get(url, headers=headers, timeout=30)
-
-    if response.status_code != 200:
-        logger.warning(f"Failed to check cooldown: {response.status_code}, proceeding with review")
-        return False
-
-    comments: list[dict[str, Any]] = response.json()
-    cutoff_time: datetime = datetime.utcnow() - timedelta(minutes=cooldown_minutes)  # noqa: E501
-
-    for comment in reversed(comments):
-        # Check if this is a bot comment
-        body: str = comment.get("body", "")
-        if "VLLM-Omni PR Review" in body or "PR Reviewer Bot" in body:
-            created_at_str: str = comment.get("created_at", "")
-            try:
-                # Parse GitHub timestamp format
-                created_at = datetime.fromisoformat(created_at_str.replace("Z", "+00:00"))
-                created_at = created_at.replace(tzinfo=None)
-                if created_at > cutoff_time:
-                    logger.info(f"PR is within cooldown period (last review: {created_at_str})")
-                    return True
-            except ValueError:
-                logger.warning(f"Failed to parse comment timestamp: {created_at_str}")  # noqa: E501
-                continue
-
-    logger.info("PR is outside cooldown period, proceeding with review")
-    return False
-
-
-def post_review_comment(  # noqa: E501
-    repo_name: str,
-    pr_number: int,
-    token: str,
-    review: str,
-) -> bool:
-    """
-    Post the review as a comment on the PR.
-
-    Args:
-        repo_name: The repository name in format "owner/repo".
-        pr_number: The pull request number.
-        token: GitHub authentication token.
-        review: The review content to post.
-
-    Returns:
-        True if posting succeeded, False otherwise.
-    """
-    url: str = (
-        f"https://api.github.com/repos/{repo_name}/issues/"
-        f"{pr_number}/comments"  # noqa: E501
-    )
-    headers: dict[str, str] = {
-        "Authorization": f"Bearer {token}",
-        "Accept": "application/vnd.github.v3+json",
-    }
-
-    # Format the review comment
-    comment_body: str = f"""## 🤖 VLLM-Omni PR Review
-
-{review}
-
----
-*This review was generated automatically by the VLLM-Omni PR Reviewer Bot
-using {os.getenv("GLM_MODEL", DEFAULT_GLM_MODEL)}.*
-"""
-
-    payload: dict[str, str] = {"body": comment_body}
-
-    logger.info(f"Posting review comment to PR #{pr_number}")
-    response = requests.post(url, headers=headers, json=payload, timeout=30)
-
-    if response.status_code == 201:
-        logger.info("Successfully posted review comment")
-        return True
-    else:
-        logger.error(f"Failed to post comment: {response.status_code}")
-        logger.error(f"Response: {response.text}")
-        return False
-
-
-def main() -> int:
-    """
-    Main entry point for the PR reviewer bot.
-
-    Returns:
-        0 on success, 1 on error.
-    """
-    logger.info("VLLM-Omni PR Reviewer Bot starting...")
-
-    # Load configuration
-    config: Config = get_config()
-    logger.info(
-        f"Configuration: model={config.glm_model}, "
-        f"cooldown={config.cooldown_minutes}min, "
-        f"max_retries={config.max_retries}"
-    )
-
-    # Get environment variables
-    token: str = get_env_var("GITHUB_TOKEN")
-    api_key: str = get_env_var("GLM_API_KEY")
-    repo_name: str = get_env_var("REPO_NAME")
-    pr_number_str: str = get_env_var("PR_NUMBER")
-    comment_body: str = get_env_var("COMMENT_BODY")
-
-    try:
-        pr_number: int = int(pr_number_str)
-    except ValueError:
-        logger.error(f"Invalid PR number: {pr_number_str}")
-        return 1
-
-    logger.info(f"Repository: {repo_name}")
-    logger.info(f"PR Number: {pr_number}")
-
-    # Check if the comment contains the trigger phrase
-    if not check_trigger(comment_body):
-        logger.info(
-            f"Comment does not contain trigger phrase '{TRIGGER_PHRASE}', exiting"  # noqa: E501
-        )
-        return 0
-
-    logger.info("Trigger phrase detected! Starting review process...")
-
-    # Check cooldown period
-    if check_cooldown(repo_name, pr_number, token, config.cooldown_minutes):
-        logger.info("Skipping review due to cooldown period")
-        return 0
-
-    # Fetch PR details
-    logger.info("Step 1/4: Fetching PR details...")
-    pr_details: PRDetails | None = fetch_pr_details(repo_name, pr_number, token)  # noqa: E501
-    if not pr_details:
-        logger.error("Failed to fetch PR details")
-        return 1
-
-    pr_title: str = pr_details.get("title", "Unknown")
-    pr_description: str = pr_details.get("body", "")
-
-    logger.info(f"PR Title: {pr_title}")
-
-    # Fetch PR diff
-    logger.info("Step 2/4: Fetching PR diff...")
-    diff: str | None = fetch_pr_diff(repo_name, pr_number, token, config.max_diff_size)
-    if diff is None:
-        logger.error("Failed to fetch PR diff")
-        return 1
-
-    if not diff:
-        logger.warning("Warning: Empty diff - this might be a draft PR or no code changes")
-
-    # Build prompt
-    logger.info("Step 3/4: Building review prompt...")
-    prompt: str = build_review_prompt(pr_title, pr_description, diff)
-
-    # Call GLM API
-    logger.info("Step 4/4: Calling GLM API...")
-    review: str | None = call_glm_api(prompt, api_key, config)
-    if not review:
-        logger.error("Failed to get review from GLM API")
-        return 1
-
-    # Post review comment
-    logger.info("Posting review comment...")
-    if not post_review_comment(repo_name, pr_number, token, review):
-        logger.error("Failed to post review comment")
-        return 1
-
-    logger.info("PR review completed successfully!")
-    return 0
-
-
-if __name__ == "__main__":
-    sys.exit(main())

.github/workflows/pr-reviewer.yml

@@ -1,62 +0,0 @@
-name: VLLM-Omni PR Reviewer
-
-on:
-  issue_comment:
-    types: [created]
-
-permissions:
-  contents: read
-  pull-requests: write
-  issues: write
-
-jobs:
-  pr-reviewer:
-    name: Review Pull Request
-    runs-on: ubuntu-latest
-    timeout-minutes: 10
-    # Only run when the comment is from a collaborator/owner/member
-    if: |
-      github.event_name == 'issue_comment' &&
-      github.event.issue.pull_request != null &&
-      contains(github.event.comment.body, '@vllm-omni-reviewer') &&
-      (github.event.comment.author_association == 'MEMBER' ||
-       github.event.comment.author_association == 'COLLABORATOR' ||
-       github.event.comment.author_association == 'OWNER')
-    steps:
-      - name: Checkout repository
-        uses: actions/checkout@v6.0.2
-        with:
-          fetch-depth: 0
-
-      - name: Set up Python
-        uses: actions/setup-python@v6.2.0
-        with:
-          python-version: '3.11'
-          cache: 'pip'
-
-      - name: Install dependencies
-        run: |
-          python -m pip install --upgrade pip==24.0
-          pip install requests==2.31.0 pyyaml==6.0.1
-
-      - name: Run PR Reviewer
-        id: reviewer
-        env:
-          GLM_API_KEY: ${{ secrets.GLM_API_KEY }}
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-          PR_NUMBER: ${{ github.event.issue.number || github.event.pull_request.number }}
-          COMMENT_BODY: ${{ github.event.comment.body }}
-          REPO_NAME: ${{ github.repository }}
-          PR_REVIEWER_COOLDOWN_MINUTES: 5
-          PR_REVIEWER_MAX_RETRIES: 3
-        run: |
-          python .github/scripts/pr_reviewer.py 2>&1 | tee "/tmp/pr_review_${PR_NUMBER}.log"
-
-      - name: Upload review logs
-        if: always()
-        uses: actions/upload-artifact@v7.0.0
-        with:
-          name: pr-review-logs-${{ github.event.issue.number || github.event.pull_request.number }}
-          path: /tmp/pr_review_*.log
-          retention-days: 7
-          if-no-files-found: ignore

.github/workflows/pre-commit.yml

@@ -24,3 +24,9 @@ jobs:
           python-version: "3.12"
           cache: 'pip'
       - uses: pre-commit/action@v3.0.1
+        # Don't run the pytest mark check against everything in the CI
+        # for now since a lot of files are still missing them. This check
+        # is used to help contributors remember to add the marks when they
+        # commit.
+        env:
+          SKIP: check-test-ci-coverage

.gitignore

@@ -208,6 +208,7 @@ cache/
 .cache/
 diffusion_cache/
 kv_cache/
+!vllm_omni/experimental/ar_diffusion/kv_cache/
 
 # Logs
 logs/
@@ -239,12 +240,15 @@ data/
 datasets/
 *.csv
 *.json
+!tests/dfx/perf/tests/*.json
 !apps/ComfyUI-vLLM-Omni/example_workflows/*.json
+!tests/assets/soulxsinger/*.json
 *.jsonl
 *.parquet
 
 # Output files
 outputs/
+!vllm_omni/outputs/
 results/
 generated/
 output_*/
@@ -255,7 +259,6 @@ configs/production.yaml
 configs/development.yaml
 
 # Docker
-.dockerignore
 Dockerfile.dev
 discussion
 tmp_test
@@ -264,5 +267,11 @@ tmp_test
 vllm_omni/_version.py
 # output files
 *.wav
+# Vendored test reference audio (small, intentionally checked in)
+!tests/assets/**/*.wav
+.worktrees/
 # CI overlay yamls materialized from tests/utils.py:_CI_OVERLAYS at test time
 tests/.ci_generated/
+
+# Local Lance precision-matching dump scripts (not part of the repo)
+lance_compare/

.pre-commit-config.yaml

@@ -54,6 +54,13 @@ repos:
       verbose: true
       stages: [commit-msg]
 
+    - id: check-test-ci-coverage
+      name: Ensure test files have pytestmarks
+      entry: python tools/pre_commit/check_test_marks.py
+      language: python
+      types: [python]
+      files: ^tests/
+
     # Keep `suggestion` last
     - id: suggestion
       name: Suggestion

.readthedocs.yml

@@ -10,6 +10,11 @@ build:
   jobs:
     post_checkout:
       - git fetch --unshallow || true
+    pre_install:
+      # Pre-install CPU torch so the subsequent `pip install .[docs]` does not
+      # pull ~2 GB of CUDA 13 libs (cuBLAS/cuDNN/NCCL/...) that exceed RTD's
+      # 1 GB memory cap and time out the mkdocs build.
+      - python -m pip install --index-url https://download.pytorch.org/whl/cpu torch
 
 mkdocs:
   configuration: mkdocs.yml

CONTRIBUTING.md

@@ -1,3 +1,5 @@
 # Contributing to vLLM-Omni
 
 You may find information about contributing to vLLM-Omni on [Contributing](https://vllm-omni.readthedocs.io/en/latest/contributing/)
+
+Before submitting a PR, run the [precheck-pr skill](https://github.com/vllm-project/vllm-omni/blob/main/.claude/skills/precheck-pr/SKILL.md) with the code agent for a self-check against project conventions.

README.md

@@ -9,19 +9,21 @@ Easy, fast, and cheap omni-modality model serving for everyone
 </h3>
 
 <p align="center">
-| <a href="https://vllm-omni.readthedocs.io/en/latest/"><b>Documentation</b></a> | <a href="https://discuss.vllm.ai"><b>User Forum</b></a> | <a href="https://slack.vllm.ai"><b>Developer Slack</b></a> | <a href="docs/assets/WeChat.jpg"><b>WeChat</b></a> | <a href="https://arxiv.org/abs/2602.02204"><b>Paper</b></a> | <a href="https://docs.google.com/presentation/d/1XJWgv79lORl8rbaVvp2d5Sqs6ZEBgAgj/edit?slide=id.p1#slide=id.p1"><b>Slides</b></a> |
+| <a href="https://vllm-omni.readthedocs.io/en/latest/"><b>Documentation</b></a> | <a href="https://deepwiki.com/vllm-project/vllm-omni"><b>DeepWiki</b></a> | <a href="https://discuss.vllm.ai"><b>User Forum</b></a> | <a href="https://slack.vllm.ai"><b>Developer Slack</b></a> | <a href="docs/assets/WeChat.jpg"><b>WeChat</b></a> | <a href="https://arxiv.org/abs/2602.02204"><b>Paper</b></a> | <a href="https://docs.google.com/presentation/d/111-L8zF7A1j_YI_cR8JsblofdScdRr2f/edit?usp=sharing&ouid=110473603432222024453&rtpof=true&sd=true"><b>Slides</b></a> |
 </p>
 
 
 ---
 
 *Latest News* 🔥
+- [2026/07] We released [0.24.0](https://github.com/vllm-project/vllm-omni/releases/tag/v0.24.0) - aligned with the vLLM 0.24 release line, expanding production-ready coverage across TTS, speech, diffusion, image/video generation, and robot-policy serving, with major Omni stage runtime refactoring, diffusion request-level batching, async output materialization, quantization/cache/memory improvements, and broad CUDA/ROCm/XPU/NPU support.
+- [2026/06] We released [0.22.0](https://github.com/vllm-project/vllm-omni/releases/tag/v0.22.0) - an **omnimodal world-model** release aligned with vLLM 0.22, featuring [Nvidia Cosmos3](recipes/cosmos3/Cosmos3-Nano.md)/DreamZero world model support, expanded quantization coverage across Blackwell/NPU/XPU, TTS production improvements, new models including MiniCPM-o 4.5, MOSS-TTS, and Lance, plus RL integration with [VeRL-Omni](https://github.com/verl-project/verl-omni).
+- [2026/05] We released [0.20.0](https://github.com/vllm-project/vllm-omni/releases/tag/v0.20.0) - refreshes the serving/runtime stack for large-scale omni workloads, and improves diffusion model performance, quantization, and hardware readiness across CUDA, ROCm, MUSA, NPU, and XPU backends.
 - [2026/03] We released [0.18.0](https://github.com/vllm-project/vllm-omni/releases/tag/v0.18.0) - strengthens the core runtime through a large entrypoint refactor and scheduler/runtime cleanups, expands unified quantization and diffusion execution, broadens multimodal model coverage, and improves production readiness across audio, omni, image, video, RL, and multi-platform deployments.
 - [2026/03] Check out our first public [project deepdive](https://youtu.be/sgwNfsNnR9I) at the vLLM Hong Kong Meetup!
 - [2026/03] **[vllm-omni-skills](https://github.com/hsliuustc0106/vllm-omni-skills)** is a community-driven collection of AI assistant skills that help developers work with vLLM-Omni more effectively. These skills can be used with popular agentic AI coding assistants like **Cursor IDE**, **Claude**, **Codex**, and more.
 - [2026/02] We released [0.16.0](https://github.com/vllm-project/vllm-omni/releases/tag/v0.16.0) - A major alignment + capability release that rebases onto **upstream vLLM v0.16.0** and significantly expands performance, distributed execution, and production readiness across **Qwen3-Omni / Qwen3-TTS**, **Bagel**, **MiMo-Audio**, **GLM-Image** and the **Diffusion (DiT) image/video stack**—while also improving platform coverage (CUDA / ROCm / NPU / XPU), CI quality, and documentation.
 - [2026/02] We released [0.14.0](https://github.com/vllm-project/vllm-omni/releases/tag/v0.14.0) - This is the first **stable release** of vLLM-Omni that expands Omni’s diffusion / image-video generation and audio / TTS stack, improves distributed execution and memory efficiency, and broadens platform/backend coverage (GPU/ROCm/NPU/XPU). It also brings meaningful upgrades to serving APIs, profiling & benchmarking, and overall stability. Please check our latest [paper](https://arxiv.org/abs/2602.02204) for architecture design and performance results.
-- [2026/01] We released [0.12.0rc1](https://github.com/vllm-project/vllm-omni/releases/tag/v0.12.0rc1) - a major RC milestone focused on maturing the diffusion stack, strengthening OpenAI-compatible serving, expanding omni-model coverage, and improving stability across platforms (GPU/NPU/ROCm).
 - [2025/11] vLLM community officially released [vllm-project/vllm-omni](https://github.com/vllm-project/vllm-omni) in order to support omni-modality models serving.
 
 ---
@@ -30,9 +32,9 @@ Easy, fast, and cheap omni-modality model serving for everyone
 
 [vLLM](https://github.com/vllm-project/vllm) was originally designed to support large language models for text-based autoregressive generation tasks. vLLM-Omni is a framework that extends its support for omni-modality model inference and serving:
 
-- **Omni-modality**: Text, image, video, and audio data processing
+- **Omni-modality**: Text, image, audio, video, and action data processing
 - **Non-autoregressive Architectures**: extend the AR support of vLLM to Diffusion Transformers (DiT) and other parallel generation models
-- **Heterogeneous outputs**: from traditional text generation to multimodal outputs
+- **Heterogeneous outputs**: from traditional text generation to multimodal and action outputs
 
 <p align="center">
   <picture>
@@ -56,8 +58,10 @@ vLLM-Omni is flexible and easy to use with:
 
 vLLM-Omni seamlessly supports most popular open-source models on HuggingFace, including:
 
-- Omni-modality models (e.g. Qwen-Omni)
-- Multi-modality generation models (e.g. Qwen-Image)
+- **Omni-modality models** (e.g. Qwen3-Omni, Cosmos3, HunyuanImage, BAGEL)
+- **TTS models** (e.g. Qwen3-TTS, VoxCPM2, Ming-Omni-TTS, CosyVoice3)
+- **Diffusion models** — image, video, and audio generation (e.g. Qwen-Image, Wan2.2, FLUX)
+- **Robot-policy and action models** (e.g. GR00T-N1.7, DreamZero-DROID, InternVLA-A1, Cosmos3 action policy)
 
 ## Getting Started
 
@@ -66,6 +70,7 @@ Visit our [documentation](https://vllm-omni.readthedocs.io/en/latest/) to learn
 - [Installation](https://vllm-omni.readthedocs.io/en/latest/getting_started/installation/)
 - [Quickstart](https://vllm-omni.readthedocs.io/en/latest/getting_started/quickstart/)
 - [List of Supported Models](https://vllm-omni.readthedocs.io/en/latest/models/supported_models/)
+- [Deployment Recipes](https://recipes.vllm.ai) for vLLM-Omni model serving
 
 ## Contributing
 

apps/lvsa-omni-showcase/README.md

@@ -0,0 +1,87 @@
+# LVSA Long-Video Showcase
+
+**Training-free block-sparse attention for long-video diffusion**, running on
+vLLM-Omni via the public plugin entry points (no core changes). LVSA speeds up
+extended-horizon generation and prevents the freeze/loop failure mode dense
+attention exhibits beyond a model's training length — with no fine-tuning and no
+weight changes.
+
+Upstream: **[LongVideoSparseAttention](https://github.com/JiusiServe/LongVideoSparseAttention)**
+(algorithm, all model adapters, full docs).
+
+## Results
+
+Wan 2.1 1.3B, single A100 80 GB, mean over 5 prompts (Dense vs LVSA-FlashInfer):
+
+| Horizon | 1× | 2× | 3× | 4× | 5× | 6× |
+|---|---|---|---|---|---|---|
+| **Speedup** | 0.9× | 1.4× | 1.9× | 2.5× | 3.0× | **3.5×** |
+| **VQeval composite** (Dense→LVSA) | 63→63 | 59→63 | 61→63 | 61→64 | 59→64 | 59→63 |
+
+The speedup **grows with length** (LVSA is the dense regime at 1× by design, so
+the win starts at 2×), and quality *improves* at extension — LVSA's rotating
+keyframes prevent the looping/static output dense produces beyond its horizon
+(VQeval loop-quality +27 to +34 at ≥3×). At long horizons dense **OOMs on 80 GB**
+(HunyuanVideo ≥2×, Wan2.1-14B and Cosmos at 6×) where LVSA still generates.
+
+→ Full sweep (5 models × 6 horizons, VQeval + VBench-Long) and the
+quality-metric methodology:
+**[benchmarks/](https://github.com/JiusiServe/LongVideoSparseAttention/tree/main/benchmarks)**.
+
+## Quickstart
+
+```bash
+# 1. vLLM-Omni 0.22.0 (stable) + matching vLLM
+pip install "vllm==0.22.0"
+pip install --no-build-isolation \
+  "vllm-omni @ git+https://github.com/vllm-project/vllm-omni.git@v0.22.0"
+
+# 2. LVSA core + plugin
+git clone https://github.com/JiusiServe/LongVideoSparseAttention.git
+( cd LongVideoSparseAttention && pip install -e . && pip install -e lvsa-vllm-omni/ )
+
+# 3. this showcase's client dep
+pip install requests
+```
+
+FlashInfer (used by the default `flashinfer` backend) **ships with vLLM** — step 1
+already pulls the pinned `flashinfer-python` + `flashinfer-cubin`, so the fused
+backend works with no extra install. Two optional extras:
+
+```bash
+# (optional) faster startup — prebuilt JIT cache, matched to your CUDA + flashinfer
+CU=cu128   # match your CUDA: cu121 / cu124 / cu128 / …
+FIVER=$(python -c "import importlib.metadata as m; print(m.version('flashinfer-python'))")
+pip install --extra-index-url "https://flashinfer.ai/whl/${CU}" \
+  "flashinfer-jit-cache==${FIVER}+${CU}"
+
+# (fallback) no FlashInfer in your environment? run the SDPA backend instead:
+export LVSA_BACKEND=sdpa
+```
+
+```bash
+# Serve an LVSA-enabled endpoint (Wan 2.1 1.3B at 2× horizon):
+MODEL=/path/to/Wan2.1-T2V-1.3B-Diffusers MODEL_FAMILY=wan FRAMES=161 \
+  bash serve_lvsa.sh
+
+# In another shell — generate:
+python generate.py --frames 161 --height 480 --width 832 \
+  --prompt "A dog running through a sunlit forest." --out wan_lvsa.mp4
+```
+
+`serve_lvsa.sh` sets the LVSA env vars for the chosen `MODEL_FAMILY`
+(`wan` / `hunyuan`). Wan and HunyuanVideo run through the LVSA **attention
+backend** by default — it stays sparse even under sequence-parallel, unlike the
+monkey-patch hooks.
+
+## Configuration
+
+The one knob you must get right is the model's training horizon
+(`LVSA_REFERENCE_LATENT_FRAMES`: Wan = 21, HunyuanVideo = 33); `serve_lvsa.sh`
+sets it per family. For the full environment-variable reference, the multi-GPU
+support matrix, and tuning (`sparsity_scale`, rotating keyframes, non-standard
+resolutions), see the upstream docs:
+
+- **[lvsa-vllm-omni plugin reference](https://github.com/JiusiServe/LongVideoSparseAttention/blob/main/lvsa-vllm-omni/README.md)** — every `LVSA_*` env var
+- **[docs/parallelism.md](https://github.com/JiusiServe/LongVideoSparseAttention/blob/main/docs/parallelism.md)** — TP / Ulysses / HSDP support per model
+- **[docs/tuning.md](https://github.com/JiusiServe/LongVideoSparseAttention/blob/main/docs/tuning.md)** — picking knobs for your model and horizon

apps/lvsa-omni-showcase/generate.py

@@ -0,0 +1,93 @@
+#!/usr/bin/env python
+"""Generate a long video through an LVSA-enabled vLLM-Omni server.
+
+A thin client over the Omni video API — start the server first with
+``serve_lvsa.sh``. The sparse attention runs server-side; this just submits a
+job and saves the result.
+
+Uses the **asynchronous** endpoint (``POST /v1/videos`` → poll
+``GET /v1/videos/{id}`` → download ``GET /v1/videos/{id}/content``) rather than
+``POST /v1/videos/sync``. The sync endpoint is capped by a fixed server-side
+timeout (``VIDEO_SYNC_TIMEOUT_S``, 600 s) and is documented as
+"for benchmark and testing scenarios" — long videos (LVSA's use case) routinely
+exceed it. The async job has no such cap.
+
+  python generate.py --frames 481 --height 480 --width 832 \
+      --prompt "A golden retriever running through a sunlit forest." \
+      --out dog_forest_lvsa.mp4
+"""
+
+import argparse
+import os
+import time
+
+import requests
+
+
+def main():
+    ap = argparse.ArgumentParser()
+    ap.add_argument("--server", default=os.environ.get("OMNI_SERVER", "http://localhost:8000"))
+    ap.add_argument("--prompt", default="A dog running through a sunlit forest.")
+    ap.add_argument("--frames", type=int, default=161)  # Wan 2x horizon
+    ap.add_argument("--height", type=int, default=480)
+    ap.add_argument("--width", type=int, default=832)
+    ap.add_argument("--fps", type=int, default=16)
+    ap.add_argument("--steps", type=int, default=50)
+    ap.add_argument("--guidance", type=float, default=5.0)
+    ap.add_argument("--out", default="lvsa_long_video.mp4")
+    ap.add_argument("--poll-interval", type=float, default=5.0, help="Seconds between status polls.")
+    ap.add_argument(
+        "--max-wait", type=float, default=7200.0, help="Give up if the job hasn't finished after this many seconds."
+    )
+    args = ap.parse_args()
+
+    base = args.server.rstrip("/")
+    form = {
+        "prompt": args.prompt,
+        "size": f"{args.width}x{args.height}",
+        "num_frames": str(args.frames),
+        "fps": str(args.fps),
+        "num_inference_steps": str(args.steps),
+        "guidance_scale": str(args.guidance),
+    }
+
+    # ── 1. submit the async job ──────────────────────────────────────────────
+    print(f"[lvsa-app] POST {base}/v1/videos  {args.width}x{args.height} {args.frames}f steps={args.steps}")
+    resp = requests.post(f"{base}/v1/videos", data=form, timeout=120)
+    resp.raise_for_status()
+    job = resp.json()
+    vid = job["id"]
+    print(f"[lvsa-app] job {vid} ({job.get('status', '?')})")
+
+    # ── 2. poll until completed / failed ─────────────────────────────────────
+    deadline = time.time() + args.max_wait
+    last = None
+    while True:
+        if time.time() > deadline:
+            raise SystemExit(f"[lvsa-app] timed out after {args.max_wait:.0f}s (job {vid} still running)")
+        time.sleep(args.poll_interval)
+        r = requests.get(f"{base}/v1/videos/{vid}", timeout=60)
+        if r.status_code != 200:
+            # retrieve_video returns a non-200 JSON when the job has FAILED
+            raise SystemExit(f"[lvsa-app] job failed: {r.text}")
+        st = r.json()
+        status = st.get("status")
+        progress = st.get("progress", 0)
+        if (status, progress) != last:
+            print(f"[lvsa-app] {status}  {progress}%")
+            last = (status, progress)
+        if status == "completed":
+            break
+        if status == "failed":
+            raise SystemExit(f"[lvsa-app] job failed: {st.get('error')}")
+
+    # ── 3. download the rendered mp4 ─────────────────────────────────────────
+    c = requests.get(f"{base}/v1/videos/{vid}/content", timeout=600)
+    c.raise_for_status()
+    with open(args.out, "wb") as f:
+        f.write(c.content)
+    print(f"[lvsa-app] saved {args.out}")
+
+
+if __name__ == "__main__":
+    main()

apps/lvsa-omni-showcase/serve_lvsa.sh

@@ -0,0 +1,45 @@
+#!/usr/bin/env bash
+# Start an LVSA-enabled vLLM-Omni server for Wan or HunyuanVideo.
+#
+# LVSA is selected as the attention backend via --diffusion-attention-config (see
+# below); the `LVSA_*` env vars only tune it. The `lvsa-vllm-omni` plugin registers
+# itself through vLLM-Omni's `general_plugins` entry points — nothing in vLLM-Omni
+# core is modified. Watch the log for "[LVSA] Geometry detected" to confirm sparse
+# attention is active for your request geometry.
+#
+#   MODEL=/path/to/Wan2.1-T2V-1.3B-Diffusers MODEL_FAMILY=wan     FRAMES=161 bash serve_lvsa.sh
+#   MODEL=/path/to/HunyuanVideo-1.5-...       MODEL_FAMILY=hunyuan FRAMES=193 bash serve_lvsa.sh
+set -euo pipefail
+
+MODEL=${MODEL:?set MODEL=/path/to/Wan-or-HunyuanVideo-Diffusers}
+MODEL_FAMILY=${MODEL_FAMILY:-wan}     # wan | hunyuan
+PORT=${PORT:-8000}
+FRAMES=${FRAMES:-161}                 # target horizon in frames
+VAE_T=${VAE_T:-4}                     # VAE temporal compression (Wan/Hunyuan = 4)
+BACKEND=${LVSA_BACKEND:-flashinfer}   # flashinfer (fused, fastest) | sdpa
+
+T_LAT=$(( (FRAMES - 1) / VAE_T + 1 ))
+
+# Per-family training reference horizon (Wan train ref = 21, Hunyuan = 33).
+case "$MODEL_FAMILY" in
+  wan)     REFERENCE=${REFERENCE:-21} ;;
+  hunyuan) REFERENCE=${REFERENCE:-33} ;;
+  *) echo "MODEL_FAMILY must be 'wan' or 'hunyuan'"; exit 1 ;;
+esac
+export LVSA_BACKEND="$BACKEND"
+export LVSA_TOTAL_LATENT_FRAMES="$T_LAT"
+export LVSA_REFERENCE_LATENT_FRAMES="$REFERENCE"
+export LVSA_ROTATE_KEYFRAMES=${LVSA_ROTATE_KEYFRAMES:-1}
+
+echo "[lvsa-app] serving $MODEL ($MODEL_FAMILY) on :$PORT"
+echo "[lvsa-app]   LVSA backend=$BACKEND  T_lat=$T_LAT (frames=$FRAMES)  reference=$REFERENCE"
+echo "[lvsa-app]   confirm activation by grep '\\[LVSA\\] Geometry detected' in this log"
+
+# vLLM-Omni 0.22 selects the LVSA attention backend per-role via
+# --diffusion-attention-config. The backend is what drives sparse attention
+# here; it engages under tensor-parallel too, unlike the LVSA_*_HOOK monkey-patch
+# path (which falls back to dense under sequence-parallel), so we deliberately do
+# NOT set a hook.
+exec vllm serve "$MODEL" --omni --host 0.0.0.0 --port "$PORT" \
+     --diffusion-attention-config '{"per_role": {"self": {"backend": "LVSA"}}}' \
+     --init-timeout 1800 ${EXTRA_ARGS:-}

benchmarks/README.md

@@ -1,44 +1,68 @@
-# Benchmarks Overview and Architecture
+# Benchmarks
 
-This document explains the benchmark architecture across all benchmark assets in this repo. It describes what we measure, and where to find or plug in new scenarios. Per-task details remain in subfolder READMEs (e.g., `benchmarks/<model>/README.md`).
+This directory contains benchmark suites for evaluating different model families and infrastructure components in vLLM-Omni. Each subfolder targets a different benchmark family with its own scripts, configs, and metrics. See the per-subfolder READMEs for detailed usage.
 
-## Scope and goals
-- Establish repeatable latency/throughput measurements for multimodal LLM pipelines.
-- Provide both HF Transformers (offline) and vLLM-Omni (multi-stage/pipeline) baselines.
-- Make it easy to plug in new datasets and models with minimal changes to the runner scripts.
+## Benchmark families
 
-## Dataset and inputs
-- Default example: SeedTTS top-100 prompts (`benchmarks/build_dataset/top100.txt`) via `benchmarks/build_dataset/`.
-- Extensible: drop in new prompt files or modality-aligned payloads; keep the expected format for the consuming scripts (e.g., one prompt per line).
-- If you add a new dataset, document it under `benchmarks/<model>/README.md` and point scripts to your data path.
+### [TTS](tts/README.md) — Text-to-Speech
 
-## Directory layout
-- `benchmarks/build_dataset/` — dataset prep utilities (e.g., SeedTTS top100).
-- `benchmarks/<model>/vllm_omni/` — vLLM-Omni pipeline benchmarks, logs, outputs.
-- `benchmarks/accuracy/` — accuracy benchmark integrations that adapt external
-  benchmark suites to vLLM-Omni serving and evaluation flows.
-- Add new tasks under `benchmarks/<model>/...` with the same pattern: `transformers/`, `vllm_omni/`, task-specific README, and (optionally) dataset prep notes.
+Model-agnostic serving benchmarks for TTS models, including Qwen3-TTS and VoxCPM2.
 
-## Reference workflows
-- **HF Transformers (offline, single process)**  
-  Script (example): `benchmarks/<model>/transformers/eval_qwen3_moe_omni_transformers.sh`  
-  Outputs: `benchmark_results/perf_stats.json`, `benchmark_results/results.json`, `benchmark_results/audio/` (if audio is produced).
+- **Layout**: `tts/bench_tts.py` (serving benchmark driver), `tts/model_configs.yaml` (model registry), `tts/plot_results.py` (result plotting)
+- **Dataset**: Seed-TTS full or text-only datasets, plus bundled smoke/design prompts under `build_dataset/`
+- **Key metrics**: TTFP (time to first audio packet), E2E latency, RTF (real-time factor), throughput (audio seconds / wall-clock second)
 
-- **vLLM-Omni end-to-end pipeline**  
-  Script (example): `benchmarks/<model>/vllm_omni/eval_qwen3_moe_omni.sh`  
-  Outputs: `vllm_omni/logs/*.stats.jsonl` (per-stage/overall latency & TPS), `vllm_omni/logs/stage*.log`, `vllm_omni/outputs/` (text/audio artifacts).
+### [Diffusion](diffusion/README.md) — Image and Video Generation
 
-- **Adding a new task/model**  
-  1) Create `benchmarks/<model>/transformers/` and/or `benchmarks/<model>/vllm_omni/` with scripts referencing your model and dataset.  
-  2) Add a task README describing dataset, configs, and expected outputs.  
-  3) Keep the output/log structure similar for easy comparison (perf_stats/results/audio or text outputs; stats.jsonl/logs for pipeline).
+Online-serving benchmark for diffusion image/video models, sending requests to the configured vLLM serving endpoint (`/v1/chat/completions`, `/v1/images/generations`, or `/v1/videos`, depending on backend/task).
 
-## Metrics to watch
-- **Throughput**: `overall_tps`, `*_tps_avg` per stage.
-- **Latency distribution**: look for long tails in `*.stats.jsonl`.
-- **Quality/completeness**: missing outputs or errors in stage logs indicate pipeline failures or misconfigurations.
+- **Tasks**: text-to-image, text-to-video, image-to-image, image-to-video, text+image-to-image, text+image-to-video
+- **Datasets**: `vbench`, `trace`, `random`
+- **Key metrics**: request throughput, latency percentiles, optional SLO attainment
 
-## Troubleshooting
-- Verify GPU/driver/FlashAttention2 requirements for your chosen model/config.
-- Ensure network access for dataset/model downloads (Google Drive, Hugging Face, etc.).
-- If outputs are missing or slow, inspect per-stage logs and `*.stats.jsonl` for errors, stragglers, or contention.
+### [GLM-Image](glm_image/README.md) — Text-to-Image and Image-to-Image
+
+Benchmarks for GLM-Image performance across HuggingFace baseline, vLLM-Omni offline inference, and vLLM-Omni online serving.
+
+- **Layout**: `glm_image/huggingface/` (HF baseline), `glm_image/vllm-omni/` (offline inference), `glm_image/benchmark_glm_image.py` (online serving)
+- **Tasks**: text-to-image and image-to-image
+- **Key metrics**: request/image throughput, latency percentiles, optional per-stage pipeline timings
+
+### [LingBot-Video](lingbot_video/README.md) — Dense and MoE Parity
+
+Manual cross-runtime validation for the dense LingBot-Video pipeline and the
+LingBot-Video MoE transformer.
+
+- **Dense pipeline**: decoded-video MAE, MSE, PSNR, latency, and optional steady-state timings
+- **MoE transformer**: bitwise router, sparse-block, shared-expert, and full-transformer parity
+
+### [Distributed](distributed/omni_connectors/README.md) — RDMA Connector Testing
+
+RDMA environment setup and transfer tests for `MooncakeTransferEngineConnector`, including pytest-based single-node checks and manual cross-node benchmarks.
+
+- **Transfer modes**: `copy`, `zerocopy`, `gpu` (GPUDirect)
+- **Supports**: single-node pytest suites and manual multi-node/cross-node transfer testing
+
+### [Accuracy](accuracy/README.md) — Image Generation and Editing Quality
+
+Accuracy benchmarks for image generation/editing models, adapting external suites to vLLM-Omni serving and local judge-evaluation flows.
+
+- **Layout**: `accuracy/text_to_image/` (GEBench), `accuracy/image_to_image/` (GEdit-Bench)
+- **Method**: generation and judge scoring both run through local `vllm-omni serve` endpoints
+
+### Common serving metrics framework
+
+`vllm_omni/benchmarks/` extends `vllm bench serve --omni` with Omni-specific datasets, backends, and multimodal metrics. Key metrics include:
+
+- **Text output**: TTFT (time to first token), TPOT (time per output token), ITL (inter-token latency)
+- **Audio output**: TTFP (time to first audio packet), E2E latency, RTF (real-time factor)
+- **Throughput**: request throughput, output token throughput, total token throughput, audio throughput
+
+See `vllm_omni/benchmarks/serve.py` for the `vllm bench serve --omni` runner wrapper and `vllm_omni/benchmarks/metrics/` for Omni metric definitions.
+
+## Adding a new benchmark
+
+1. Create a subfolder under `benchmarks/<name>/` with scripts, configs if needed, and a `README.md`.
+2. If comparing against another runtime, use clear backend subfolders where applicable, such as `huggingface/` and `vllm-omni/`, or follow the shared TTS serving benchmark pattern in `tts/`.
+3. Add reusable dataset or prompt-building utilities to `build_dataset/` if applicable.
+4. Update this README with a link to the new benchmark family.

benchmarks/accuracy/README.md

@@ -23,5 +23,5 @@ Test guidance:
 - Local static/self-checks live in `tests/benchmarks/test_accuracy_bench_utils.py`.
 - End-to-end generation/evaluation should be validated in a remote GPU
   environment. In the current repo marker system there is `L4` but no `L5`
-  marker, so benchmark smoke tests should be wired as `advanced_model +
-  benchmark + L4` when GPU capacity is available.
+  marker, so benchmark smoke tests should be wired as `full_model +
+  benchmark + L4` for nightly when GPU capacity is available.

benchmarks/accuracy/common.py

@@ -151,6 +151,9 @@ class VllmOmniImageClient:
         seed: int | None = None,
         negative_prompt: str | None = None,
         output_compression: int | None = None,
+        bot_task: str | None = None,
+        sys_type: str | None = None,
+        system_prompt: str | None = None,
     ) -> Image.Image:
         if not isinstance(images, list):
             images = [images]
@@ -170,6 +173,12 @@ class VllmOmniImageClient:
             data["negative_prompt"] = negative_prompt
         if output_compression is not None:
             data["output_compression"] = str(output_compression)
+        if bot_task is not None:
+            data["bot_task"] = bot_task
+        if sys_type is not None:
+            data["sys_type"] = sys_type
+        if system_prompt is not None:
+            data["system_prompt"] = system_prompt
 
         files = [
             (

benchmarks/accuracy/image_to_image/README.md

@@ -99,5 +99,5 @@ Notes:
 - This flow requires the optional Hugging Face `datasets` package.
 - `generate` writes `generation_manifest.json` with local output coverage.
 - The current repo marker set exposes `L4` but not `L5`, so if you promote an
-  end-to-end smoke test into CI, use the existing `advanced_model`, `benchmark`,
-  and `L4` markers or introduce a new repo-wide marker explicitly first.
+  end-to-end smoke test into CI, use the `full_model`, `benchmark`,
+  and `L4` markers for nightly (or `advanced_model` for merge) or introduce a new repo-wide marker explicitly first.

benchmarks/accuracy/image_to_image/gedit_bench.py

@@ -64,6 +64,47 @@ def resolve_model_name(*, model_name: str | None, model: str | None = None, outp
     raise ValueError("model-name is required when it cannot be inferred from model or output-root")
 
 
+def _coerce_score_int(value: Any) -> int:
+    if isinstance(value, bool):
+        raise TypeError(f"Expected numeric score, got bool: {value!r}")
+    if isinstance(value, (int, float)):
+        return int(value)
+    if isinstance(value, str):
+        stripped = value.strip()
+        if stripped.lstrip("-").isdigit():
+            return int(stripped)
+    raise TypeError(f"Expected numeric score, got {type(value).__name__}: {value!r}")
+
+
+def _expand_score_entry(entry: Any) -> list[int]:
+    if isinstance(entry, dict):
+        if "score" in entry:
+            return [_coerce_score_int(entry["score"])]
+        if "value" in entry:
+            return [_coerce_score_int(entry["value"])]
+        numeric_values = [_coerce_score_int(value) for value in entry.values()]
+        if numeric_values:
+            return numeric_values
+        raise TypeError(f"Score dict has no numeric values: {entry!r}")
+    return [_coerce_score_int(entry)]
+
+
+def _normalize_score_list(score: Any) -> list[int]:
+    if isinstance(score, dict):
+        entries: list[Any] = [score]
+    elif isinstance(score, list):
+        entries = score
+    else:
+        entries = [score]
+
+    normalized: list[int] = []
+    for entry in entries:
+        normalized.extend(_expand_score_entry(entry))
+    if not normalized:
+        raise ValueError(f"Empty score list after normalization: {score!r}")
+    return normalized
+
+
 def parse_score_payload(raw_text: str) -> dict[str, Any]:
     try:
         parsed = extract_json_object(raw_text)
@@ -73,10 +114,7 @@ def parse_score_payload(raw_text: str) -> dict[str, Any]:
             parsed = {"score": json.loads(stripped), "reasoning": ""}
         else:
             raise
-    score = parsed.get("score", [])
-    if not isinstance(score, list):
-        score = [score]
-    parsed["score"] = [int(value) for value in score]
+    parsed["score"] = _normalize_score_list(parsed.get("score", []))
     return parsed
 
 
@@ -391,6 +429,9 @@ class GEditBenchRunner:
         num_inference_steps: int = 20,
         guidance_scale: float | None = None,
         seed: int | None = 42,
+        bot_task: str | None = None,
+        sys_type: str | None = None,
+        system_prompt: str | None = None,
     ):
         self.dataset_ref = dataset_ref
         self.output_root = output_root
@@ -400,6 +441,9 @@ class GEditBenchRunner:
         self.num_inference_steps = num_inference_steps
         self.guidance_scale = guidance_scale
         self.seed = seed
+        self.bot_task = bot_task
+        self.sys_type = sys_type
+        self.system_prompt = system_prompt
         self.client = VllmOmniImageClient(base_url=base_url, api_key=api_key)
 
     def generate(
@@ -478,6 +522,9 @@ class GEditBenchRunner:
             num_inference_steps=self.num_inference_steps,
             guidance_scale=self.guidance_scale,
             seed=self.seed,
+            bot_task=self.bot_task,
+            sys_type=self.sys_type,
+            system_prompt=self.system_prompt,
         )
         edited_image.save(output_path)
         return {
@@ -633,6 +680,25 @@ def build_parser() -> argparse.ArgumentParser:
     generate.add_argument("--num-inference-steps", type=int, default=20)
     generate.add_argument("--guidance-scale", type=float, default=None)
     generate.add_argument("--seed", type=int, default=42)
+    generate.add_argument(
+        "--bot-task",
+        type=str,
+        default=None,
+        choices=["think", "recaption", "think_recaption", "vanilla"],
+        help="HunyuanImage-3 prompting mode for /v1/images/edits (optional).",
+    )
+    generate.add_argument(
+        "--sys-type",
+        type=str,
+        default=None,
+        help="Hunyuan system-prompt type (e.g. en_unified). Defaults from --bot-task when omitted.",
+    )
+    generate.add_argument(
+        "--system-prompt",
+        type=str,
+        default=None,
+        help="Custom Hunyuan system prompt body (sys_type=custom).",
+    )
     generate.add_argument("--workers", type=int, default=1)
     generate.add_argument("--max-samples", type=int, default=None)
     generate.add_argument("--samples-per-group", type=int, default=None)
@@ -675,6 +741,9 @@ def main(argv: list[str] | None = None) -> int:
             num_inference_steps=args.num_inference_steps,
             guidance_scale=args.guidance_scale,
             seed=args.seed,
+            bot_task=args.bot_task,
+            sys_type=args.sys_type,
+            system_prompt=args.system_prompt,
         )
         records = runner.generate(
             model_name=model_name,

benchmarks/build_dataset/seed_tts_design/en/meta.lst

@@ -0,0 +1,30 @@
+vd001|||The quick brown fox jumps over the lazy dog.|A warm, friendly female voice with a slight American Midwest accent, speaking at a moderate pace with natural inflection.
+vd002|||Welcome to the future of text-to-speech synthesis.|A deep, authoritative male news anchor voice, clear and professional with a measured cadence.
+vd003|||The sunset painted the sky in brilliant shades of orange and pink.|A gentle elderly female voice, soft and wise, with a slight Southern American accent.
+vd004|||Scientists have discovered a new species of deep-sea creature.|A young male voice with an Australian accent, curious and enthusiastic.
+vd005|||Breaking news: a major climate summit opens today in Geneva.|A crisp female newsreader voice, neutral accent, confident and precise.
+vd006|||In the beginning, there was darkness and silence across the universe.|A rich, dramatic bass male narrator voice, slow and deeply resonant.
+vd007|||Come closer, I have something important to tell you.|A soft, intimate female voice, slightly whispery, warm and gentle.
+vd008|||And they're off! The horses race toward the first turn at incredible speed.|An energetic male sports commentator, fast-paced and excited.
+vd009|||Once upon a time, in a land far away, lived a very clever fox.|A light, playful voice with childlike enthusiasm, bright and clear.
+vd010|||The ancient manuscript reveals secrets hidden for a thousand years.|A wise, measured elderly male voice, slow and deliberate, British English accent.
+vd011|||Good evening, ladies and gentlemen, and welcome to our show.|A sophisticated female voice with a slight French accent speaking English, elegant and refined.
+vd012|||System initialized. Running diagnostics. All systems nominal.|A clear, precise robotic-sounding voice, neutral and monotone with slight synthetic quality.
+vd013|||I hear what you are saying, and it is completely understandable to feel that way.|A warm, empathetic female therapist voice, calm and reassuring, unhurried pace.
+vd014|||Attention all units: proceed to grid reference seven-seven-alpha.|A firm, authoritative military male voice, clipped and commanding.
+vd015|||Oh my goodness, you have to try this amazing new recipe I just found!|An enthusiastic, bubbly female voice, high energy and friendly.
+vd016|||Dude, the waves were totally amazing out there today. Super happy about it!|A relaxed male voice with a California accent, casual and laid-back.
+vd017|||The quarterly results exceed expectations across all major metrics.|A sharp, businesslike female voice, confident and efficient, fast-paced delivery.
+vd018|||Chapter one. The morning sun filtered gently through the forest canopy.|A smooth, rich male audiobook narrator voice, expressive and engaging.
+vd019|||To be or not to be, that is the question.|A theatrical female voice, dramatic and expressive, stage projection quality.
+vd020|||And that is all for tonight. Stay well out there, everyone.|A warm, velvety male late-night radio DJ voice, smooth and intimate.
+vd021|||Please fasten your seatbelts; we will be experiencing some turbulence shortly.|A calm, professional female flight attendant voice, reassuring and clear.
+vd022|||The fossil record suggests these creatures lived more than sixty million years ago.|A measured male documentary narrator voice, scientific and curious.
+vd023|||Hey kiddo, did you finish your homework before dinner today?|A gentle, caring father voice, warm and slightly tired after a long day.
+vd024|||Live from courtside, the home team has just taken a commanding lead.|A high-energy male sports broadcaster, urgent and engaged.
+vd025|||Please remain on the line while I transfer you to the next available agent.|A polite, neutral female call-center voice, professional and patient.
+vd026|||I have a dream that one day this nation will rise up and live out the true meaning of its creed.|A powerful, resonant male orator voice, slow and impassioned.
+vd027|||Mama always said life was like a box of chocolates; you never know what you're gonna get.|A folksy male voice with a Southern American accent, warm and reflective.
+vd028|||Ahoy there, matey! Hoist the colors and prepare to set sail.|A theatrical pirate male voice, gruff and boisterous.
+vd029|||Ladies and gentlemen, please welcome to the stage our keynote speaker.|A polished female event MC voice, gracious and articulate.
+vd030|||Once you've completed the form, click submit and we'll review it within two business days.|A friendly female product-onboarding voice, encouraging and clear.

benchmarks/build_dataset/seed_tts_smoke/en/meta.lst

@@ -0,0 +1,30 @@
+smoke001|||The quick brown fox jumps over the lazy dog near the riverbank at sunset.
+smoke002|||Welcome to the future of text-to-speech synthesis in production systems.
+smoke003|||Yesterday the team finished rolling out the new authentication flow.
+smoke004|||She walked carefully across the wet cobblestones, careful not to slip.
+smoke005|||The conference call is scheduled for nine in the morning, Pacific time.
+smoke006|||Please remember to save your work before closing the editor.
+smoke007|||Two plus two equals four, but five hundred and forty three digits is long.
+smoke008|||I would like a coffee with oat milk and a chocolate croissant please.
+smoke009|||The library closes at eight on weekdays and six on Saturdays.
+smoke010|||During the Renaissance, art and science flourished in European cities.
+smoke011|||He whispered the secret word so quietly that no one else could hear.
+smoke012|||Our flight departs from gate twenty three at eleven fifteen.
+smoke013|||The storm knocked out power for six hours, but the backup generator kicked in.
+smoke014|||Reading a good book on a rainy afternoon is one of life's great pleasures.
+smoke015|||When the kettle whistled, she poured the hot water over the fresh tea leaves.
+smoke016|||The algorithm runs in linear time, which is a big improvement over the previous approach.
+smoke017|||In the distance, the mountains were shrouded in thick morning fog.
+smoke018|||Our company reported record revenue for the fourth quarter of the fiscal year.
+smoke019|||She explained the new policy in detail during the staff meeting this morning.
+smoke020|||The children laughed and played in the garden until the sun began to set.
+smoke021|||A long line of cars stretched along the coastal highway after the lighthouse opened.
+smoke022|||The chef recommended the seared scallops with a lemon butter sauce as the special.
+smoke023|||Our quarterly engineering review begins promptly at two in the afternoon on Thursday.
+smoke024|||Despite the heavy rain, the marathon runners kept a steady pace through downtown.
+smoke025|||She finished the crossword puzzle in under fifteen minutes while sipping her tea.
+smoke026|||The new release adds streaming output, faster startup, and better error reporting.
+smoke027|||His grandfather taught him how to whittle small wooden birds from cedar offcuts.
+smoke028|||Travelers were advised to check the timetable since trains were running on a holiday schedule.
+smoke029|||The startup launched a public beta after eighteen months of closed development.
+smoke030|||At night the desert sky filled with stars so bright the Milky Way was clearly visible.

benchmarks/build_dataset/sound_effect_smoke/en/meta.lst

@@ -0,0 +1,10 @@
+se001|ocean waves crashing on a rocky beach|10.0
+se002|heavy rain on a metal roof at night|8.0
+se003|a dog barking in the distance|6.0
+se004|a crackling campfire in a quiet forest|10.0
+se005|footsteps on dry leaves|6.0
+se006|thunder rolling across a wide valley|8.0
+se007|a busy coffee shop with chatter and espresso machines|10.0
+se008|wind howling through tall pine trees|8.0
+se009|a small stream flowing over pebbles|10.0
+se010|a creaking wooden door slowly opening|6.0

benchmarks/build_dataset/ttsd_smoke/en/meta.lst

@@ -0,0 +1,10 @@
+ttsd001|prompt-wavs/spk1.wav|prompt-wavs/spk2.wav|[S1] Hello and welcome to the show today. [S2] Thanks for having me, it's a pleasure to be here. [S1] Let's dive right in.
+ttsd002|prompt-wavs/spk1.wav|prompt-wavs/spk2.wav|[S1] Did you finish the report? [S2] Almost — I'm just polishing the conclusion. [S1] Great, send it over when you're done.
+ttsd003|prompt-wavs/spk1.wav|prompt-wavs/spk2.wav|[S1] Have you tried that new restaurant downtown? [S2] Yes, the pasta is amazing. [S1] I'll book a table for Friday.
+ttsd004|prompt-wavs/spk1.wav|prompt-wavs/spk2.wav|[S1] What did you think of yesterday's meeting? [S2] Productive, but a bit long. [S1] Agreed, we should set an hour cap next time.
+ttsd005|prompt-wavs/spk1.wav|prompt-wavs/spk2.wav|[S1] The weather is beautiful today. [S2] Perfect for a hike. [S1] Want to drive out to the trails this afternoon?
+ttsd006|prompt-wavs/spk1.wav|prompt-wavs/spk2.wav|[S1] How did the demo go? [S2] Better than expected — the client signed off on the prototype. [S1] Fantastic news!
+ttsd007|prompt-wavs/spk1.wav|prompt-wavs/spk2.wav|[S1] Coffee or tea this morning? [S2] Coffee please, with a splash of milk. [S1] Coming right up.
+ttsd008|prompt-wavs/spk1.wav|prompt-wavs/spk2.wav|[S1] Did you read the email about the deadline change? [S2] I did, we now have until next Friday. [S1] That should give us enough buffer.
+ttsd009|prompt-wavs/spk1.wav|prompt-wavs/spk2.wav|[S1] Are you joining the team dinner tonight? [S2] I'll be there around seven. [S1] See you then.
+ttsd010|prompt-wavs/spk1.wav|prompt-wavs/spk2.wav|[S1] Have we backed up the database? [S2] Last backup ran two hours ago. [S1] Good, I'll trigger a fresh snapshot before the migration.

benchmarks/diffusion/README.md

@@ -30,7 +30,7 @@ python3 benchmarks/diffusion/diffusion_benchmark_serving.py \
 
 **Notes**
 
-- The benchmark talks to `http://<host>:<port>/v1/chat/completions`.
+- By default, image tasks talk to `http://<host>:<port>/v1/chat/completions`; video tasks talk to `/v1/videos`.
 - If you run the server on another host or port, pass `--base-url` accordingly.
 
 ## 2. Supported Datasets
@@ -95,8 +95,9 @@ You can point to your own trace using `--dataset-path`.
 
 ### Basic flags
 
-- `--base-url`: Server address (the script calls `.../v1/chat/completions`).
+- `--base-url`: Server address; `--endpoint` selects the path appended to this base URL.
 - `--model`: The OpenAI-compatible `model` field.
+- `--endpoint`: API endpoint path. Leading `/` is optional, e.g. `/v1/videos` or `v1/videos`.
 - `--task`: Task type (e.g., `t2i`, `t2v`, `i2i`, `i2v`).
 - `--dataset`: Dataset mode (`vbench` / `trace` / `random`).
 - `--num-prompts`: Number of requests to send.
@@ -130,9 +131,21 @@ Warmup flags:
 
 - `--warmup-requests`: Number of warmup requests.
 - `--warmup-num-inference-steps`: Steps used during warmup.
+- `--warmup-concurrency`: Maximum concurrent warmup requests. Use this to warm
+  the same batch shape as the measured run instead of warming only batch=`1`.
 - For `--task t2v`: warmup requests are forced to use `num_frames=1` to make warmup faster and less noisy.
 
 Traffic / concurrency flags:
 
 - `--request-rate`: Target request rate (requests/second). If set to `inf`, the script sends all requests immediately.
 - `--max-concurrency`: Max number of in-flight requests (default: `1`). This can hard-cap the achieved QPS: if it is too small, requests will queue behind the semaphore, and both achieved throughput and observed SLO attainment can be skewed.
+
+### Batched warmup note
+
+For batched serving runs, warm the same in-flight shape you plan to measure.
+For example, a run with `--max-concurrency 8` should usually also use
+`--warmup-requests 8 --warmup-concurrency 8`; otherwise the first measured
+batch may still pay compile or CUDA-graph capture cost.
+
+For a Qwen-Image continuous-batching replay example, see
+[`performance_dashboard/qwen_image_serving_performance.md`](./performance_dashboard/qwen_image_serving_performance.md).

benchmarks/diffusion/backends.py

@@ -10,6 +10,8 @@ from typing import Any
 import aiohttp
 from tqdm import tqdm
 
+DEFAULT_EDITS_BOT_TASK = "think"
+
 
 @dataclass
 class RequestFuncInput:
@@ -26,7 +28,9 @@ class RequestFuncInput:
     slo_ms: float | None = None
     extra_body: dict[str, Any] = field(default_factory=dict)
     image_paths: list[str] | None = None
+    video_paths: list[str] | None = None
     request_id: str = field(default_factory=lambda: str(uuid.uuid4()))
+    default_bot_task: str | None = DEFAULT_EDITS_BOT_TASK
 
 
 @dataclass
@@ -53,6 +57,95 @@ def _encode_image_as_data_url(path: str) -> str:
     return f"data:{mime};base64,{encoded}"
 
 
+async def async_request_image_edits(
+    input: RequestFuncInput,
+    session: aiohttp.ClientSession,
+    pbar: tqdm | None = None,
+    enable_diffusion_pipeline_profiler: bool = False,
+) -> RequestFuncOutput:
+    """POST /v1/images/edits (multipart)."""
+    del enable_diffusion_pipeline_profiler
+    output = RequestFuncOutput()
+    output.start_time = time.perf_counter()
+
+    extra_body = dict(input.extra_body)
+    width = input.width or extra_body.get("width") or 1024
+    height = input.height or extra_body.get("height") or 1024
+    edits_url = input.api_url
+
+    form = aiohttp.FormData()
+    form.add_field("model", input.model)
+    form.add_field("prompt", input.prompt)
+    form.add_field("size", f"{width}x{height}")
+    form.add_field("response_format", "b64_json")
+
+    if input.num_inference_steps is not None:
+        form.add_field("num_inference_steps", str(input.num_inference_steps))
+    elif extra_body.get("num_inference_steps") is not None:
+        form.add_field("num_inference_steps", str(extra_body["num_inference_steps"]))
+
+    if input.seed is not None:
+        form.add_field("seed", str(input.seed))
+    elif extra_body.get("seed") is not None:
+        form.add_field("seed", str(extra_body["seed"]))
+
+    if extra_body.get("guidance_scale") is not None:
+        form.add_field("guidance_scale", str(extra_body["guidance_scale"]))
+    if extra_body.get("negative_prompt") is not None:
+        form.add_field("negative_prompt", str(extra_body["negative_prompt"]))
+    if extra_body.get("true_cfg_scale") is not None:
+        form.add_field("true_cfg_scale", str(extra_body["true_cfg_scale"]))
+    if extra_body.get("sys_type") is not None:
+        form.add_field("sys_type", str(extra_body["sys_type"]))
+    if extra_body.get("system_prompt") is not None:
+        form.add_field("system_prompt", str(extra_body["system_prompt"]))
+
+    bot_task = extra_body.get("bot_task")
+    if bot_task is None and input.default_bot_task is not None:
+        bot_task = input.default_bot_task
+    if bot_task is not None:
+        form.add_field("bot_task", str(bot_task))
+
+    assert input.image_paths is not None
+    for img_path in input.image_paths:
+        if not os.path.exists(img_path):
+            output.error = f"Image file not found: {img_path}"
+            output.success = False
+            if pbar:
+                pbar.update(1)
+            return output
+        with open(img_path, "rb") as img_f:
+            image_bytes = img_f.read()
+        form.add_field(
+            "image",
+            image_bytes,
+            filename=os.path.basename(img_path),
+            content_type=_guess_mime_type(img_path),
+        )
+
+    try:
+        async with session.post(edits_url, data=form) as response:
+            if response.status == 200:
+                resp_json = await response.json()
+                output.response_body = resp_json
+                output.success = True
+            else:
+                output.error = f"HTTP {response.status}: {await response.text()}"
+                output.success = False
+    except Exception as e:
+        output.error = str(e)
+        output.success = False
+
+    output.latency = time.perf_counter() - output.start_time
+
+    if output.success and input.slo_ms is not None:
+        output.slo_achieved = (output.latency * 1000.0) <= float(input.slo_ms)
+
+    if pbar:
+        pbar.update(1)
+    return output
+
+
 async def async_request_chat_completions(
     input: RequestFuncInput,
     session: aiohttp.ClientSession,
@@ -122,6 +215,18 @@ async def async_request_chat_completions(
                                     output.peak_memory_mb = first_item.get("peak_memory_mb", 0.0)
                 except (IndexError, TypeError, AttributeError):
                     pass
+
+                if (not output.stage_durations or output.peak_memory_mb == 0.0) and isinstance(
+                    resp_json.get("metrics"), dict
+                ):
+                    m = resp_json["metrics"]
+                    if not output.stage_durations and isinstance(m.get("stage_durations"), dict):
+                        output.stage_durations = m.get("stage_durations") or {}
+                    if output.peak_memory_mb == 0.0 and m.get("peak_memory_mb") is not None:
+                        try:
+                            output.peak_memory_mb = float(m.get("peak_memory_mb") or 0.0)
+                        except (TypeError, ValueError):
+                            pass
             else:
                 output.error = f"HTTP {response.status}: {await response.text()}"
                 output.success = False
@@ -139,7 +244,7 @@ async def async_request_chat_completions(
     return output
 
 
-async def async_request_openai_images(
+async def async_request_openai_image_generations(
     input: RequestFuncInput,
     session: aiohttp.ClientSession,
     pbar: tqdm | None = None,
@@ -234,6 +339,11 @@ async def async_request_v1_videos(
         form.add_field(k, str(v))
 
     image_file = None
+    if input.image_paths and input.video_paths:
+        output.error = "Only one of image_paths or video_paths can be provided"
+        output.success = False
+        return output
+
     if input.image_paths and len(input.image_paths) > 0:
         image_path = input.image_paths[0]
         image_file = open(image_path, "rb")
@@ -241,7 +351,16 @@ async def async_request_v1_videos(
             "input_reference",
             image_file,
             filename=os.path.basename(image_path),
-            content_type="application/octet-stream",
+            content_type=_guess_mime_type(image_path),
+        )
+    elif input.video_paths and len(input.video_paths) > 0:
+        video_path = input.video_paths[0]
+        image_file = open(video_path, "rb")
+        form.add_field(
+            "input_reference",
+            image_file,
+            filename=os.path.basename(video_path),
+            content_type=_guess_mime_type(video_path),
         )
 
     job_id = None
@@ -336,12 +455,39 @@ async def async_request_v1_videos(
     return output
 
 
+LEGACY_BACKEND_ENDPOINT_ALIASES = {
+    "vllm-omni": "/v1/chat/completions",
+    "openai": "/v1/images/generations",
+}
+
+
+def normalize_endpoint(value: str) -> str:
+    endpoint = str(value).strip()
+    if not endpoint:
+        raise ValueError("endpoint must not be empty.")
+    endpoint = LEGACY_BACKEND_ENDPOINT_ALIASES.get(
+        endpoint,
+        LEGACY_BACKEND_ENDPOINT_ALIASES.get(endpoint.lstrip("/"), endpoint),
+    )
+    if not endpoint.startswith("/"):
+        endpoint = f"/{endpoint}"
+    return endpoint
+
+
+def endpoint_filename_token(value: str) -> str:
+    token = normalize_endpoint(value).lstrip("/")
+    for bad in ("/", "\\", ":", "*", "?", '"', "<", ">", "|"):
+        token = token.replace(bad, "_")
+    return token or "endpoint"
+
+
 backends_function_mapping = {
     "2i": {
-        "vllm-omni": (async_request_chat_completions, "/v1/chat/completions"),
-        "openai": (async_request_openai_images, "/v1/images/generations"),
+        "/v1/chat/completions": (async_request_chat_completions, "/v1/chat/completions"),
+        "/v1/images/generations": (async_request_openai_image_generations, "/v1/images/generations"),
+        "/v1/images/edits": (async_request_image_edits, "/v1/images/edits"),
     },
     "2v": {
-        "v1/videos": (async_request_v1_videos, "/v1/videos"),
+        "/v1/videos": (async_request_v1_videos, "/v1/videos"),
     },
 }

benchmarks/diffusion/bench_attention_backends.py

@@ -0,0 +1,328 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
+"""Standalone diagnostic for the diffusion attention backends added in #3079.
+
+Exercises the same synthetic attention shape through each backend and each
+torch SDPA sub-kernel, with and without an attn_mask, so we can pinpoint why
+CUDNN_ATTN underperforms on some SKUs. Two typical causes:
+
+  (a) cuDNN lacks a tuned kernel for this (SM, head_dim, seq) combination
+      and silently falls back to MATH.
+  (b) attn_mask is non-None and the cuDNN SDPA dispatch rejects it,
+      walking CUDNN -> FLASH -> MATH.
+
+Run:
+    python benchmarks/diffusion/bench_attention_backends.py --preset hv15
+    python benchmarks/diffusion/bench_attention_backends.py --preset wan22
+    python benchmarks/diffusion/bench_attention_backends.py \
+        --batch 1 --heads 24 --seq 14336 --head-dim 128
+
+Optional extras:
+    pip install --pre flash-attn-4        # FA4 is currently pre-release only
+    pip install -U flashinfer              # latest FlashInfer (0.6.9)
+
+The table at the end is the data we want on the PR — surface the row where a
+backend is >1.5x the SDPA baseline, that's the one to gate off the auto-route.
+"""
+
+from __future__ import annotations
+
+import argparse
+import time
+
+import torch
+from torch.nn.attention import SDPBackend, sdpa_kernel
+
+# Presets mirror the DiT hot-path attention call (single-stream block) at the
+# shapes we run in the PR's validation configs.
+_PRESETS = {
+    # HunyuanVideo-1.5 480p/33f: (30 * 52 * 9) latent + ~256 text tokens.
+    "hv15": {"batch": 1, "heads": 24, "seq": 14336, "head_dim": 128},
+    # Wan 2.2 480p/33f rough estimate; adjust when we measure real shapes.
+    "wan22": {"batch": 1, "heads": 40, "seq": 16384, "head_dim": 128},
+    # Smaller image-gen shape for quick smoke tests.
+    "flux": {"batch": 1, "heads": 24, "seq": 4096, "head_dim": 128},
+}
+
+_SDPA_BACKENDS = [
+    ("CUDNN_ATTENTION", [SDPBackend.CUDNN_ATTENTION]),
+    ("FLASH_ATTENTION", [SDPBackend.FLASH_ATTENTION]),
+    ("EFFICIENT_ATTENTION", [SDPBackend.EFFICIENT_ATTENTION]),
+    ("MATH", [SDPBackend.MATH]),
+    # The PR's CUDNN_ATTN impl uses this priority chain; keep it here to show
+    # what torch actually picks when multiple are allowed.
+    ("CUDNN_ATTN_CHAIN", [SDPBackend.CUDNN_ATTENTION, SDPBackend.FLASH_ATTENTION, SDPBackend.MATH]),
+]
+
+
+def _env_report() -> None:
+    if not torch.cuda.is_available():
+        raise SystemExit("CUDA not available; this diagnostic needs a GPU.")
+    cc = torch.cuda.get_device_capability()
+    print("=" * 72)
+    print(f"GPU           : {torch.cuda.get_device_name()}")
+    print(f"SM capability : sm_{cc[0]}{cc[1]}")
+    print(f"torch         : {torch.__version__}")
+    print(f"cuDNN version : {torch.backends.cudnn.version()}")
+    print("flashinfer    : ", end="")
+    try:
+        import flashinfer
+
+        print(getattr(flashinfer, "__version__", "present"))
+    except Exception as e:
+        print(f"not installed ({type(e).__name__})")
+    print("=" * 72)
+
+
+def _make_qkv(batch: int, heads: int, seq: int, head_dim: int, device: str, dtype: torch.dtype):
+    # Layout: (B, S, H, D) — what the PR's backends accept as input.
+    q = torch.randn(batch, seq, heads, head_dim, device=device, dtype=dtype)
+    k = torch.randn(batch, seq, heads, head_dim, device=device, dtype=dtype)
+    v = torch.randn(batch, seq, heads, head_dim, device=device, dtype=dtype)
+    return q, k, v
+
+
+def _make_mask(batch: int, seq: int, device: str, dtype: torch.dtype, pad_tokens: int = 128) -> torch.Tensor:
+    # Mirrors a text-encoder padding mask: last `pad_tokens` positions masked.
+    mask = torch.zeros(batch, 1, seq, seq, device=device, dtype=dtype)
+    if pad_tokens > 0:
+        mask[..., -pad_tokens:] = float("-inf")
+    return mask
+
+
+def _time_call(fn, *args, warmup: int = 3, iters: int = 10) -> tuple[float, str]:
+    """Return (median ms, error string). Skips on any exception so that a
+    backend that rejects our args (wrong dtype, missing JIT module, unsupported
+    kwarg value) doesn't abort the whole sweep."""
+    try:
+        for _ in range(warmup):
+            fn(*args)
+        torch.accelerator.synchronize()
+        times = []
+        for _ in range(iters):
+            torch.accelerator.synchronize()
+            t0 = time.perf_counter()
+            fn(*args)
+            torch.accelerator.synchronize()
+            times.append((time.perf_counter() - t0) * 1000.0)
+        times.sort()
+        return times[len(times) // 2], ""
+    except Exception as e:  # noqa: BLE001 — probe script, keep the whole table going
+        msg = str(e).split("\n", 1)[0][:60]
+        return float("nan"), f"{type(e).__name__}: {msg}" if msg else type(e).__name__
+
+
+def _run_sdpa_variants(q, k, v, attn_mask, scale: float) -> list[tuple[str, float, str]]:
+    # q/k/v come in (B, S, H, D); F.sdpa wants (B, H, S, D).
+    q_t = q.permute(0, 2, 1, 3)
+    k_t = k.permute(0, 2, 1, 3)
+    v_t = v.permute(0, 2, 1, 3)
+    rows: list[tuple[str, float, str]] = []
+    for name, backends in _SDPA_BACKENDS:
+
+        def _call(mask=attn_mask):
+            with sdpa_kernel(backends):
+                return torch.nn.functional.scaled_dot_product_attention(
+                    q_t, k_t, v_t, attn_mask=mask, dropout_p=0.0, is_causal=False, scale=scale
+                )
+
+        ms, err = _time_call(_call)
+        rows.append((name, ms, err))
+    return rows
+
+
+def _run_flashinfer(q, k, v, scale: float, backend: str | None = None, attn_mask=None) -> tuple[float, str]:
+    """Call FlashInfer's dense single-prefill.
+
+    ``backend`` hints at cutlass/fa3/trtllm if the installed version exposes
+    that kwarg. ``attn_mask`` is the same additive 4D mask we pass to SDPA —
+    we convert it to the 2D boolean ``custom_mask`` form FlashInfer accepts.
+    Per FlashInfer docs, ``custom_mask`` only applies when ``causal=False``.
+    """
+    try:
+        import inspect
+
+        from flashinfer.prefill import single_prefill_with_kv_cache
+    except Exception as e:
+        return float("nan"), f"import-{type(e).__name__}"
+
+    kwargs: dict = {"sm_scale": scale, "causal": False, "return_lse": False}
+    if backend is not None:
+        sig = inspect.signature(single_prefill_with_kv_cache)
+        if "backend" not in sig.parameters:
+            return float("nan"), "no-backend-kwarg"
+        kwargs["backend"] = backend
+
+    if attn_mask is not None:
+        # Collapse (B, 1, S, S) additive float mask to (S, S) boolean.
+        # FlashInfer expects True = keep, False = masked out.
+        mask_2d = attn_mask[0, 0]
+        kwargs["custom_mask"] = mask_2d != float("-inf")
+
+    def _call():
+        out = single_prefill_with_kv_cache(q[0], k[0], v[0], **kwargs)
+        return out.unsqueeze(0)
+
+    return _time_call(_call)
+
+
+def _run_fa4(q, k, v, scale: float) -> tuple[float, str]:
+    """Call FlashAttention-4 directly (``pip install flash-attn-4``). FA4
+    ships a Blackwell-native kernel via CuTe-DSL; on sm_120 it should beat
+    cuDNN by ~20%. API lives under ``flash_attn.cute``; the older
+    ``flash_attn.flash_attn_func`` path is FA2/FA3 only."""
+    try:
+        from flash_attn.cute import flash_attn_func
+    except Exception as e:
+        return float("nan"), f"import-{type(e).__name__}"
+
+    def _call():
+        # FA4 accepts (B, S, H, D) directly — same layout the PR's backends use.
+        return flash_attn_func(q, k, v, softmax_scale=scale, causal=False)
+
+    return _time_call(_call)
+
+
+def _run_flashinfer_cudnn_batch(q, k, v, scale: float) -> tuple[float, str]:
+    """Call FlashInfer's direct cuDNN wrapper. Bypasses PyTorch SDPA dispatch
+    (which has a few hundred ns of overhead per call) and talks to cuDNN FMHA
+    straight. Useful as a ceiling for 'pure cuDNN, no SDPA' on Blackwell."""
+    try:
+        from flashinfer.prefill import cudnn_batch_prefill_with_kv_cache
+    except Exception as e:
+        return float("nan"), f"import-{type(e).__name__}"
+
+    b, s, h, d = q.shape
+    qo_indptr = torch.tensor([0, s], dtype=torch.int32, device=q.device)
+    kv_indptr = torch.tensor([0, s], dtype=torch.int32, device=q.device)
+
+    def _call():
+        return cudnn_batch_prefill_with_kv_cache(
+            q.reshape(b * s, h, d),
+            k.reshape(b * s, h, d),
+            v.reshape(b * s, h, d),
+            qo_indptr=qo_indptr,
+            kv_indptr=kv_indptr,
+            max_qo_len=s,
+            max_kv_len=s,
+            sm_scale=scale,
+            causal=False,
+        )
+
+    return _time_call(_call)
+
+
+def _print_table(title: str, rows: list[tuple[str, float, str]], baseline_name: str | None = None) -> None:
+    print(f"\n{title}")
+    print("-" * 88)
+    header = f"{'backend':<24} {'median (ms)':>14}    {'vs baseline':>12}    status"
+    print(header)
+    print("-" * 88)
+    baseline_ms = None
+    if baseline_name is not None:
+        for name, ms, _ in rows:
+            if name == baseline_name and ms == ms:  # not NaN
+                baseline_ms = ms
+                break
+    for name, ms, err in rows:
+        ms_str = f"{ms:>14.3f}" if ms == ms else f"{'n/a':>14}"
+        if baseline_ms is not None and ms == ms:
+            ratio = baseline_ms / ms
+            ratio_str = f"{ratio:>11.2f}x"
+        else:
+            ratio_str = f"{'—':>12}"
+        status = "ok" if not err else f"FAILED ({err})"
+        print(f"{name:<24} {ms_str}    {ratio_str}    {status}")
+
+
+def _pick_winner(rows: list[tuple[str, float, str]]) -> tuple[str, float] | None:
+    """Return (backend, ms) of the fastest non-failing row, or None."""
+    ok_rows = [(n, m) for n, m, err in rows if not err and m == m]
+    if not ok_rows:
+        return None
+    return min(ok_rows, key=lambda x: x[1])
+
+
+def _bench_one_shape(shape: dict, dtype: torch.dtype, device: str) -> tuple[list, list]:
+    torch.manual_seed(0)
+    q, k, v = _make_qkv(**shape, device=device, dtype=dtype)
+    scale = 1.0 / (shape["head_dim"] ** 0.5)
+
+    rows_nomask = _run_sdpa_variants(q, k, v, attn_mask=None, scale=scale)
+    rows_nomask.append(("FLASHINFER (default)", *_run_flashinfer(q, k, v, scale)))
+    # `trtllm-gen` has no sm_120 cubins (NVIDIA/TensorRT-LLM#11799) — skipped.
+    for fi_backend in ("fa2", "fa3", "cutlass", "auto"):
+        rows_nomask.append((f"FLASHINFER ({fi_backend})", *_run_flashinfer(q, k, v, scale, backend=fi_backend)))
+    rows_nomask.append(("FLASHINFER (cudnn-batch)", *_run_flashinfer_cudnn_batch(q, k, v, scale)))
+    rows_nomask.append(("FA4 (direct)", *_run_fa4(q, k, v, scale)))
+
+    mask = _make_mask(shape["batch"], shape["seq"], device=device, dtype=dtype, pad_tokens=256)
+    rows_mask = _run_sdpa_variants(q, k, v, attn_mask=mask, scale=scale)
+    rows_mask.append(("FLASHINFER (dense)", *_run_flashinfer(q, k, v, scale, attn_mask=mask)))
+    rows_mask.append(("FA4 (direct)", float("nan"), "mask-not-supported"))
+
+    return rows_nomask, rows_mask
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter)
+    parser.add_argument("--preset", choices=list(_PRESETS.keys()), default="hv15")
+    parser.add_argument("--sweep", action="store_true", help="Run all presets and print a ranking")
+    parser.add_argument("--batch", type=int, default=None)
+    parser.add_argument("--heads", type=int, default=None)
+    parser.add_argument("--seq", type=int, default=None)
+    parser.add_argument("--head-dim", type=int, default=None)
+    parser.add_argument("--dtype", choices=["bf16", "fp16"], default="bf16")
+    parser.add_argument("--device", default="cuda:0")
+    args = parser.parse_args()
+
+    dtype = torch.bfloat16 if args.dtype == "bf16" else torch.float16
+    _env_report()
+    print(f"dtype         : {args.dtype}")
+
+    presets = list(_PRESETS.keys()) if args.sweep else [args.preset]
+    summary: dict[str, tuple[str, float] | None] = {}
+
+    for preset_name in presets:
+        shape = dict(_PRESETS[preset_name])
+        if not args.sweep:
+            for k in ("batch", "heads", "seq", "head_dim"):
+                v = getattr(args, k if k != "head_dim" else "head_dim")
+                if v is not None:
+                    shape[k] = v
+
+        print("\n" + "=" * 88)
+        print(
+            f"Preset: {preset_name}  |  batch={shape['batch']} heads={shape['heads']} "
+            f"seq={shape['seq']} head_dim={shape['head_dim']}"
+        )
+        print("=" * 88)
+
+        rows_nomask, rows_mask = _bench_one_shape(shape, dtype, args.device)
+        _print_table(f"[{preset_name}] No attention mask", rows_nomask, baseline_name="CUDNN_ATTENTION")
+        _print_table(
+            f"[{preset_name}] With attention mask (pad 256 tokens)", rows_mask, baseline_name="CUDNN_ATTENTION"
+        )
+
+        summary[preset_name] = _pick_winner(rows_nomask)
+
+    if args.sweep or len(presets) > 1:
+        print("\n" + "=" * 88)
+        print("Winners per preset (no-mask path)")
+        print("=" * 88)
+        for preset_name, winner in summary.items():
+            if winner is None:
+                print(f"  {preset_name:<10} — no successful backend")
+            else:
+                name, ms = winner
+                print(f"  {preset_name:<10} {name:<24} {ms:>8.3f} ms")
+
+    print("\nNotes:")
+    print("  * Ratios are relative to CUDNN_ATTENTION. >1.0x means faster than cuDNN.")
+    print("  * Mask-path winner inherits CUDNN_ATTN's fallback in the PR's backends.")
+    print("  * `trtllm-gen` and FA4 4.0.0b10 are known-broken on sm_120 as of Apr 2026.")
+    print("  * For e2e timings, run text_to_video.py / text_to_image.py while varying DIFFUSION_ATTENTION_BACKEND.")
+
+
+if __name__ == "__main__":
+    main()

benchmarks/diffusion/bench_sage_attn3_matrix.sh

@@ -0,0 +1,108 @@
+#!/usr/bin/env bash
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
+#
+# 2x2 timing matrix (backend x compile mode) for SAGE_ATTN_3 vs TORCH_SDPA,
+# on HunyuanVideo-1.5 480p T2V and Wan 2.2 TI2V-5B. Run from the repo root:
+#
+#     bash benchmarks/diffusion/bench_sage_attn3_matrix.sh
+#
+# Requires: sageattn3 installed (for SAGE_ATTN_3 rows) and a Blackwell GPU.
+
+set -u
+
+STEPS=${STEPS:-50}
+SEED=${SEED:-42}
+PROMPT=${PROMPT:-"A dog running across a field of golden wheat."}
+OUT_DIR=${OUT_DIR:-/tmp/bench_sage3}
+mkdir -p "$OUT_DIR"
+
+declare -A TOTAL PER_STEP TEXT_ENC TRANSFORMER VAE
+
+run() {
+    local name="$1"; shift
+    local model="$1"; shift
+    local h="$1"; shift
+    local w="$1"; shift
+    local f="$1"; shift
+    local g="$1"; shift
+    local log="$OUT_DIR/${name}.log"
+    echo "[$(date +%T)] Running $name ..."
+    env "$@" python examples/offline_inference/text_to_video/text_to_video.py \
+        --model "$model" \
+        --prompt "$PROMPT" \
+        --height "$h" --width "$w" --num-frames "$f" \
+        --num-inference-steps "$STEPS" --seed "$SEED" --guidance-scale "$g" \
+        --enable-diffusion-pipeline-profiler \
+        --output "$OUT_DIR/${name}.mp4" > "$log" 2>&1 || {
+            echo "  FAILED — see $log"
+            TOTAL[$name]="FAIL"; PER_STEP[$name]="FAIL"
+            TEXT_ENC[$name]="-"; TRANSFORMER[$name]="-"; VAE[$name]="-"
+            return
+        }
+
+    local wait_ms
+    wait_ms=$(grep "add_req_and_wait=" "$log" | tail -1 | sed -nE 's/.*add_req_and_wait=([0-9.]+) ms.*/\1/p')
+    local total_s
+    total_s=$(grep -oE "Total generation time: [0-9.]+ seconds" "$log" | tail -1 | sed -nE 's/.*: ([0-9.]+) seconds.*/\1/p')
+    local text_enc_s
+    text_enc_s=$(grep -oE "text_encoder.forward took [0-9.]+s" "$log" | awk '{sub(/s$/,"",$NF); sum+=$NF} END {printf "%.3f", sum+0}')
+    local vae_s
+    vae_s=$(grep -oE "vae.decode took [0-9.]+s" "$log" | awk '{sub(/s$/,"",$NF); sum+=$NF} END {printf "%.3f", sum+0}')
+    local pipeline_s
+    pipeline_s=$(grep -oE "Pipeline\.forward took [0-9.]+s" "$log" | tail -1 | sed -nE 's/.*took ([0-9.]+)s.*/\1/p')
+    local transformer_s
+    transformer_s=$(awk "BEGIN {printf \"%.3f\", ${pipeline_s:-0} - ${vae_s:-0} - ${text_enc_s:-0}}")
+
+    TOTAL[$name]="${total_s:-0}"
+    PER_STEP[$name]=$(awk "BEGIN {printf \"%.3f\", ${wait_ms:-0} / 1000 / $STEPS}")
+    TEXT_ENC[$name]="$text_enc_s"
+    TRANSFORMER[$name]="$transformer_s"
+    VAE[$name]="$vae_s"
+
+    echo "  total=${TOTAL[$name]}s  per-step=${PER_STEP[$name]}s/it  transformer=${TRANSFORMER[$name]}s  vae=${VAE[$name]}s"
+}
+
+# ========== HunyuanVideo-1.5 480p T2V (480x832x33, CFG=6.0) ==========
+HV_MODEL=hunyuanvideo-community/HunyuanVideo-1.5-Diffusers-480p_t2v
+run hv15_sdpa_eager     "$HV_MODEL" 480 832 33 6.0 TORCH_COMPILE_DISABLE=1
+run hv15_sdpa_compiled  "$HV_MODEL" 480 832 33 6.0
+run hv15_sage3_eager    "$HV_MODEL" 480 832 33 6.0 TORCH_COMPILE_DISABLE=1 DIFFUSION_ATTENTION_BACKEND=SAGE_ATTN_3
+run hv15_sage3_compiled "$HV_MODEL" 480 832 33 6.0                         DIFFUSION_ATTENTION_BACKEND=SAGE_ATTN_3
+
+# ========== Wan 2.2 TI2V-5B (704x1280x33, CFG=5.0) ==========
+WAN_MODEL=Wan-AI/Wan2.2-TI2V-5B-Diffusers
+run wan22_sdpa_eager     "$WAN_MODEL" 704 1280 33 5.0 TORCH_COMPILE_DISABLE=1
+run wan22_sdpa_compiled  "$WAN_MODEL" 704 1280 33 5.0
+run wan22_sage3_eager    "$WAN_MODEL" 704 1280 33 5.0 TORCH_COMPILE_DISABLE=1 DIFFUSION_ATTENTION_BACKEND=SAGE_ATTN_3
+run wan22_sage3_compiled "$WAN_MODEL" 704 1280 33 5.0                         DIFFUSION_ATTENTION_BACKEND=SAGE_ATTN_3
+
+print_table() {
+    local title="$1"; shift
+    echo
+    echo "$title (${STEPS} steps, seed=${SEED})"
+    echo
+    printf "| %-23s | %-9s | %-15s | %-12s | %-15s | %-14s |\n" \
+        "Config" "Total (s)" "Per-step (s/it)" "Text Enc (s)" "Transformer (s)" "VAE Decode (s)"
+    printf "| %-23s | %-9s | %-15s | %-12s | %-15s | %-14s |\n" \
+        "-----------------------" "---------" "---------------" "------------" "---------------" "--------------"
+    for row in "$@"; do
+        case "$row" in
+            *_sdpa_eager)     label="SDPA + Eager" ;;
+            *_sdpa_compiled)  label="SDPA + Compiled" ;;
+            *_sage3_eager)    label="SAGE_ATTN_3 + Eager" ;;
+            *_sage3_compiled) label="SAGE_ATTN_3 + Compiled" ;;
+        esac
+        printf "| %-23s | %-9s | %-15s | %-12s | %-15s | %-14s |\n" \
+            "$label" "${TOTAL[$row]}" "${PER_STEP[$row]}" "${TEXT_ENC[$row]}" "${TRANSFORMER[$row]}" "${VAE[$row]}"
+    done
+}
+
+print_table "HunyuanVideo-1.5 480p T2V, 832x480x33" \
+    hv15_sdpa_eager hv15_sdpa_compiled hv15_sage3_eager hv15_sage3_compiled
+
+print_table "Wan 2.2 TI2V-5B, 1280x704x33" \
+    wan22_sdpa_eager wan22_sdpa_compiled wan22_sage3_eager wan22_sage3_compiled
+
+echo
+echo "Logs: $OUT_DIR/*.log"