[Fix] Restore MammothModa2 support

Signed-off-by: hsliuustc0106 <hsliuustc0106@users.noreply.github.com>
2026-08-27 23:40:02 +08:00
464 changed files with 3216 additions and 37338 deletions
@@ -1,7 +1,4 @@
 #!/bin/bash
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
-
 # vllm-omni customized version
 # Based on: https://github.com/vllm-project/ci-infra/blob/main/buildkite/bootstrap-amd.sh
 # Last synced: 2025-12-15
@@ -9,9 +6,6 @@

 set -euo pipefail

-# The bootstrap runs from the repository root; ShellCheck does not follow
-# sourced files unless invoked with -x.
-# shellcheck disable=SC1091
 source .buildkite/common/scripts/resolve_skip_ci.sh

 if [[ -z "${RUN_ALL:-}" ]]; then
@@ -69,8 +63,6 @@ upload_pipeline() {
    # Install minijinja
    ls .buildkite || buildkite-agent annotate --style error 'Please merge upstream main branch for buildkite CI'
    curl -sSfL https://github.com/mitsuhiko/minijinja/releases/download/2.3.1/minijinja-cli-installer.sh | sh
-    # Installed by the minijinja bootstrap above and only present on the CI agent.
-    # shellcheck disable=SC1091
    source /var/lib/buildkite-agent/.cargo/env

    if [[ $BUILDKITE_PIPELINE_SLUG == "fastcheck" ]]; then
@@ -98,82 +90,22 @@ upload_pipeline() {
    cd .buildkite/amd

    # Select test definition file: merge suite for main, ready suite for PRs.
-    # For debugging, DEBUG_TEST_YAML accepts a comma-separated list containing
-    # "merge" and/or "ready" (case-insensitive). Multiple suites are combined
-    # before rendering so they share one amd-build step.
+    # For debugging, DEBUG_TEST_YAML can override the selection — accepts
+    # "merge" or "ready" (case-insensitive).
    if [[ -n "${DEBUG_TEST_YAML:-}" ]]; then
-        declare -a DEBUG_TEST_SPECS=()
-        declare -A SEEN_DEBUG_TESTS=()
-        IFS=',' read -ra REQUESTED_DEBUG_TESTS <<< "${DEBUG_TEST_YAML,,}"
-
-        for requested_test in "${REQUESTED_DEBUG_TESTS[@]}"; do
-            # Permit readable values such as "ready, merge," and ignore the
-            # empty item produced by a trailing comma.
-            requested_test="${requested_test//[[:space:]]/}"
-            [[ -z "$requested_test" ]] && continue
-
-            if [[ -n "${SEEN_DEBUG_TESTS[$requested_test]:-}" ]]; then
-                echo "ERROR: duplicate DEBUG_TEST_YAML suite '$requested_test'" >&2
+        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
-            fi
-            SEEN_DEBUG_TESTS[$requested_test]=1
-
-            case "$requested_test" in
-                ready)
-                    DEBUG_TEST_SPECS+=("READY_TESTS:test-amd-ready.yml")
-                    ;;
-                merge)
-                    DEBUG_TEST_SPECS+=("MERGE_TESTS:test-amd-merge.yml")
-                    ;;
-                *)
-                    echo "ERROR: DEBUG_TEST_YAML entries must be 'merge' or 'ready', got '$requested_test'" >&2
-                    exit 1
-                    ;;
-            esac
-        done
-
-        if [[ ${#DEBUG_TEST_SPECS[@]} -eq 0 ]]; then
-            echo "ERROR: DEBUG_TEST_YAML did not contain a test suite" >&2
-            exit 1
-        elif [[ ${#DEBUG_TEST_SPECS[@]} -eq 1 ]]; then
-            TEST_YAML="${DEBUG_TEST_SPECS[0]#*:}"
-        else
-            TEST_YAML=$(mktemp "${TMPDIR:-/tmp}/amd-debug-tests.XXXXXX.yml")
-            python - "$TEST_YAML" "${DEBUG_TEST_SPECS[@]}" <<'PY'
-import sys
-
-import yaml
-
-
-output_path, *suite_specs = sys.argv[1:]
-combined = {"env": {}, "steps": []}
-
-for suite_spec in suite_specs:
-    group_name, input_path = suite_spec.split(":", 1)
-    with open(input_path, encoding="utf-8") as test_file:
-        suite = yaml.safe_load(test_file)
-
-    for name, value in (suite.get("env") or {}).items():
-        previous = combined["env"].get(name, value)
-        if previous != value:
-            raise ValueError(
-                f"Conflicting environment value for {name}: {previous!r} != {value!r}"
-            )
-        combined["env"][name] = value
-
-    suite_steps = []
-    for entry in suite.get("steps") or []:
-        if "group" in entry:
-            suite_steps.extend(entry.get("steps") or [])
-        else:
-            suite_steps.append(entry)
-    combined["steps"].append({"group": group_name, "steps": suite_steps})
-
-with open(output_path, "w", encoding="utf-8") as output_file:
-    yaml.safe_dump(combined, output_file, sort_keys=False)
-PY
-        fi
-        echo "DEBUG_TEST_YAML override: using ${DEBUG_TEST_SPECS[*]}"
+                ;;
+        esac
+        echo "DEBUG_TEST_YAML override: using $TEST_YAML"
    elif [[ $BUILDKITE_BRANCH == "main" ]]; then
        TEST_YAML="test-amd-merge.yml"
    else
@@ -198,20 +130,19 @@ PY
            > pipeline.yaml
    )
    cat pipeline.yaml
-    if [[ "$TEST_YAML" == "${TMPDIR:-/tmp}/amd-debug-tests."*.yml ]]; then
-        rm -f -- "$TEST_YAML"
-    fi
    buildkite-agent artifact upload pipeline.yaml
    buildkite-agent pipeline upload pipeline.yaml
    exit 0
 }

 get_diff() {
-    git diff --name-only --diff-filter=ACMDR "$(git merge-base origin/main HEAD)"
+    $(git add .)
+    echo $(git diff --name-only --diff-filter=ACMDR $(git merge-base origin/main HEAD))
 }

 get_diff_main() {
-    git diff --name-only --diff-filter=ACMDR HEAD~1
+    $(git add .)
+    echo $(git diff --name-only --diff-filter=ACMDR HEAD~1)
 }

 file_diff=$(get_diff)
@@ -250,7 +181,7 @@ for file in $file_diff; do
    # First check if file matches any pattern
    matches_pattern=0
    for pattern in "${patterns[@]}"; do
-        if [[ $file == "$pattern"* ]] || [[ $file == "$pattern" ]]; then
+        if [[ $file == $pattern* ]] || [[ $file == $pattern ]]; then
            matches_pattern=1
            break
        fi
@@ -260,7 +191,7 @@ for file in $file_diff; do
    if [[ $matches_pattern -eq 1 ]]; then
        matches_ignore=0
        for ignore in "${ignore_patterns[@]}"; do
-            if [[ $file == "$ignore"* ]] || [[ $file == "$ignore" ]]; then
+            if [[ $file == $ignore* ]] || [[ $file == $ignore ]]; then
                matches_ignore=1
                break
            fi
@@ -21,7 +21,7 @@ steps:

    - label: "Simple · Diffusion Test · Shard %N/%t"
      agent_pool: mi300_1
-      parallelism: 4
+      parallelism: 2
      depends_on: amd-build
      mirror_hardwares: [amdproduction]
      grade: Blocking
@@ -71,45 +71,26 @@ steps:
      commands:
        - export VLLM_LOGGING_LEVEL=DEBUG
        - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-        - pytest -s -v tests/diffusion/offloader/test_diffusion_cpu_offload.py
-        - >-
-          pytest -s -v tests/diffusion/offloader/test_diffusion_layerwise_offload.py
-          --deselect='tests/diffusion/offloader/test_diffusion_layerwise_offload.py::test_layerwise_offload_diffusion_model[stabilityai/stable-audio-open-1.0]'
+        - |
+          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))
+          '

-    # TODO: Diagnose unstable Stable Audio layerwise-offload memory savings on ROCm.
-    - label: "Diffusion Stable Audio Layerwise Offload Test (quarantined)"
-      agent_pool: mi300_1
-      depends_on: amd-build
-      mirror_hardwares: [amdproduction]
-      grade: NonBlocking
-      commands:
-        - export VLLM_LOGGING_LEVEL=DEBUG
-        - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-        - >-
-          pytest -s -v
-          'tests/diffusion/offloader/test_diffusion_layerwise_offload.py::test_layerwise_offload_diffusion_model[stabilityai/stable-audio-open-1.0]'
-
-    - label: "Diffusion Sequence Parallelism Test (Need 4 GPUs) · Shard 1/2"
+    - label: "Diffusion Sequence Parallelism Test (Need 4 GPUs)"
      agent_pool: mi300_4
      depends_on: amd-build
      mirror_hardwares: [amdproduction]
-      timeout_in_minutes: 40
      grade: Blocking
      commands:
        - export VLLM_LOGGING_LEVEL=DEBUG
        - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-        - pytest -s -v tests/diffusion/distributed/test_sequence_parallel.py
-
-    - label: "Diffusion Sequence Parallelism Test (Need 4 GPUs) · Shard 2/2"
-      agent_pool: mi300_4
-      depends_on: amd-build
-      mirror_hardwares: [amdproduction]
-      timeout_in_minutes: 40
-      grade: Blocking
-      commands:
-        - export VLLM_LOGGING_LEVEL=DEBUG
-        - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-        - pytest -s -v tests/diffusion/distributed/test_ulysses_uaa_perf.py
+        - 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"
@@ -190,45 +171,21 @@ steps:
      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 tests/e2e/offline_inference/test_qwen3_omni_colocate_async.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+        - 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 · Shard %N/%t"
      agent_pool: mi300_1
-      parallelism: 3
+      parallelism: 2
      depends_on: amd-build
      mirror_hardwares: [amdproduction]
      grade: Blocking
-      timeout_in_minutes: 80
      commands:
        - |
-          timeout --signal=TERM --kill-after=2m 80m bash -c '
+          bash -c '
            export VLLM_LOGGING_LEVEL=DEBUG
            export VLLM_WORKER_MULTIPROC_METHOD=spawn
            export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
-            TESTS=(
-              "tests/e2e/online_serving/test_qwen3_tts_customvoice.py::test_default_cuda_graph_startup[async_chunk]"
-              "tests/e2e/online_serving/test_qwen3_tts_customvoice.py::test_text_to_audio_002[async_chunk]"
-              "tests/e2e/offline_inference/test_qwen3_tts_customvoice.py::test_text_to_audio_001[no_cuda_graph]"
-            )
-            TEST="$${TESTS[$$BUILDKITE_PARALLEL_JOB]}"
-            pytest -s -v "$$TEST" -m "advanced_model" --run-level "advanced_model"
-          '
-
-    # TODO: Restore this case to the blocking sharded step once its long-running
-    # Whisper validation on AMD CI is resolved.
-    - label: "Qwen3-TTS CustomVoice async_chunk Test (quarantined)"
-      agent_pool: mi300_1
-      depends_on: amd-build
-      mirror_hardwares: [amdproduction]
-      grade: NonBlocking
-      timeout_in_minutes: 80
-      commands:
-        - |
-          timeout --signal=TERM --kill-after=2m 75m bash -c '
-            export VLLM_LOGGING_LEVEL=DEBUG
-            export VLLM_WORKER_MULTIPROC_METHOD=spawn
-            export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
-            pytest -s -v "tests/e2e/online_serving/test_qwen3_tts_customvoice.py::test_text_to_audio_001[async_chunk]" -m "advanced_model" --run-level "advanced_model"
+            pytest -s -v tests/e2e/online_serving/test_qwen3_tts_customvoice.py tests/e2e/offline_inference/test_qwen3_tts_customvoice.py -m "advanced_model" --run-level "advanced_model" --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB
          '

    - label: "Qwen3-TTS Base E2E Test"
@@ -238,10 +195,13 @@ steps:
      grade: Blocking
      timeout_in_minutes: 50
      commands:
-        - export VLLM_LOGGING_LEVEL=DEBUG
-        - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-        - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
-        - timeout 40m 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"
+        - |
+          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
@@ -191,7 +191,10 @@ steps:
      mirror_hardwares: [amdproduction]
      grade: Blocking
      commands:
-        - pytest -s -v tests/e2e/online_serving/test_cosyvoice3_tts_expansion.py -m "slow" --run-level "core_model"
+        - |
+          timeout 45m 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
@@ -85,9 +85,6 @@ steps:
          - vllm_omni/model_executor/models/common/alias_free_activation.py
          - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
          - vllm_omni/deploy/qwen3_tts.yaml
-          - vllm_omni/entrypoints/openai/serving_speech.py
-          - vllm_omni/entrypoints/openai/tts_adapters/
-          - vllm_omni/engine/stage_pool.py
        commands:
          - |
            timeout 40m bash -c "
@@ -107,9 +104,6 @@ steps:
          - vllm_omni/model_executor/models/common/alias_free_activation.py
          - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
          - vllm_omni/deploy/qwen3_tts.yaml
-          - vllm_omni/entrypoints/openai/serving_speech.py
-          - vllm_omni/entrypoints/openai/tts_adapters/
-          - vllm_omni/engine/stage_pool.py
          - .buildkite/common/scripts/run_cov_split.sh
          - pyproject.toml
        # 40 -> 60: the coverage split runs the model twice.
@@ -128,7 +122,6 @@ steps:
      - label: "Omni · Qwen3-Omni Test"
        source_file_dependencies:
          - tests/e2e/offline_inference/test_qwen3_omni.py
-          - tests/e2e/offline_inference/test_qwen3_omni_colocate_async.py
          - tests/e2e/online_serving/test_qwen3_omni.py
          - tests/helpers/runtime.py
          - vllm_omni/entrypoints/openai/serving_chat.py
@@ -139,10 +132,9 @@ steps:
          - vllm_omni/model_executor/models/qwen3_tts/cuda_graph_decoder_wrapper.py
          - vllm_omni/model_executor/stage_input_processors/qwen3_omni.py
          - vllm_omni/deploy/qwen3_omni_moe.yaml
-          - vllm_omni/deploy/qwen3_omni_moe_thinking.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 tests/e2e/offline_inference/test_qwen3_omni_colocate_async.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+          - 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"
@@ -174,7 +166,6 @@ steps:
          - tests/e2e/online_serving/test_minicpmo_realtime_duplex_drivers.py
          - tests/e2e/online_serving/helpers/minicpmo_realtime_duplex_scenarios.py
          - tests/e2e/online_serving/run_minicpmo_realtime_duplex_multi_session.py
-          - tests/e2e/online_serving/run_minicpmo_realtime_duplex_server_vad.py
          - tests/e2e/online_serving/run_minicpmo_realtime_duplex_soft_interrupt.py
          - vllm_omni/deploy/minicpmo_4_5.yaml
          - tests/e2e/online_serving/helpers/minicpmo_4_5_duplex.py
@@ -225,50 +216,3 @@ steps:
              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 · HunyuanVideo-1.5 Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_hunyuan_video_15_expansion.py
-          - vllm_omni/diffusion/models/hunyuan_video/
-        commands:
-          - |
-            timeout 40m bash -c "
-              export VLLM_IMAGE_FETCH_TIMEOUT=60
-              pytest -s -v tests/e2e/online_serving/test_hunyuan_video_15_expansion.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
-            "
-        mirror_hardwares: h100_1
-
-      - label: "Diffusion · MiniMax-H3 L3 DLO DP2 and FastH3"
-        source_file_dependencies:
-          - tests/e2e/online_serving/minimax_h3/
-          - tests/helpers/assertions.py
-          - tests/helpers/mark.py
-          - tests/helpers/media.py
-          - tests/helpers/runtime.py
-          - vllm_omni/diffusion/models/minimax_h3/
-          - vllm_omni/diffusion/lora/
-          - vllm_omni/diffusion/offloader/
-          - vllm_omni/diffusion/distributed/
-          - vllm_omni/entrypoints/openai/
-        # Keep four server lifecycles in one Buildkite job, but never in one
-        # pytest process. DLO uses multiprocessing shared-memory queues; the
-        # parent's resource_tracker can retain a stale /psm_* entry during
-        # teardown and strand the next Ref2VA collective.
-        # A per-file process plus a hard timeout also makes a regression fail
-        # instead of leaving the merge queue stuck indefinitely.
-        timeout_in_minutes: 150
-        commands:
-          - |
-            set -eu
-            export VLLM_IMAGE_FETCH_TIMEOUT=60
-            if [ -f vllm_omni/diffusion/models/minimax_h3/lora.py ]; then
-              export VLLM_TEST_MINIMAX_H3_TURBO_LORA=$(python3 -c 'from huggingface_hub import hf_hub_download; print(hf_hub_download(repo_id="lightx2v/Minimax-h3-Turbo", filename="minimax_h3_fl2v_turbo_4step_v1.0_768p_bf16.safetensors"))')
-            fi
-            if [ -f vllm_omni/diffusion/models/minimax_h3/fasth3.py ]; then
-              export VLLM_TEST_MINIMAX_H3_FASTH3_LORA=$(python3 -c 'from huggingface_hub import hf_hub_download; print(hf_hub_download(repo_id="FastVideo/FastVideo-FastH3-4-step-Preview-v1-LoRA", filename="dense-datafree/adapter_model.safetensors"))')
-            fi
-            timeout 35m pytest -s -v tests/e2e/online_serving/minimax_h3/test_minimax_h3_dlo_dp2_fl2va.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
-            timeout 35m pytest -s -v tests/e2e/online_serving/minimax_h3/test_minimax_h3_dlo_dp2_ref2va.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
-            timeout 35m pytest -s -v tests/e2e/online_serving/minimax_h3/test_minimax_h3_turbo_lora.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
-            timeout 35m pytest -s -v tests/e2e/online_serving/minimax_h3/test_minimax_h3_fasth3_lora.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
-        mirror_hardwares: h100_2
@@ -12,22 +12,16 @@ steps:
    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 · Single-GPU"
-        timeout_in_minutes: 90
+      - 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 and cards_1" --ignore=tests/e2e/accuracy --ignore=tests/e2e/online_serving/test_minicpmo_4_5_duplex_expansion.py
-        mirror_hardwares: h100_1
-
-      - label: ":full_moon: Omni · Function Test with H100 · 2-GPU"
-        timeout_in_minutes: 90
-        commands:
-          - pytest -sv tests/e2e --run-level "full_model" -m "full_model and H100 and omni and cards_2" --ignore=tests/e2e/accuracy --ignore=tests/e2e/online_serving/test_qwen3_omni_multi_replicas.py
+          - 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 --ignore=tests/e2e/online_serving/test_minicpmo_4_5_duplex_expansion.py
        mirror_hardwares: h100_2

      - label: ":full_moon: Omni · MiniCPM-o 4.5 Duplex Test"
        timeout_in_minutes: 50
        commands:
-          - pytest -s -v tests/e2e/online_serving/test_minicpmo_4_5_duplex_expansion.py -m "full_model and cuda and H100 and omni and cards_1" --run-level "full_model"
+          - pytest -s -v tests/e2e/online_serving/test_minicpmo_4_5_duplex_expansion.py -m "full_model and cuda and H100 and omni" --run-level "full_model"
        mirror_hardwares: h100_1

      - label: ":full_moon: Omni · Doc Test with L4"
@@ -37,10 +31,10 @@ steps:
          - 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 · 2-GPU"
+      - 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 and cards_2" --run-level "full_model"
+          - 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"
@@ -142,29 +136,15 @@ steps:
          - pytest -s -v tests/e2e/ -m "full_model and L4 and tts" --run-level "full_model" --ignore=tests/e2e/accuracy
        mirror_hardwares: l4_1

-      - label: ":full_moon: TTS · Perf Test · Single-GPU"
-        key: nightly-tts-performance-single
+      - 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 -m "H100 and cards_1"
-            EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
-            exit $$EXIT
-        mirror_hardwares: h100_1
-
-      - label: ":full_moon: TTS · Perf Test · 2-GPU"
-        key: nightly-tts-performance-2gpu
-        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 -m "H100 and cards_2"
+            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
@@ -199,6 +179,7 @@ steps:
      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.

      - label: ":full_moon: Diffusion X2I(&A&T) · Function Test with H100 · Single-GPU"
        timeout_in_minutes: 120
@@ -206,24 +187,24 @@ steps:
          - >-
            pytest -sv
            tests/e2e/online_serving/test_qwen_image_expansion.py
-            -m "full_model and diffusion and H100 and cards_1"
+            -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 · 2-GPU"
+      - 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
-            -m "full_model and diffusion and H100 and cards_2"
+            -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 L4 · Multi-GPU"
+      - 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 hunyuan" -m "full_model and diffusion and L4 and not cards_1" --run-level "full_model" --ignore=tests/e2e/accuracy
+          - pytest -sv tests/e2e/ -k "not test_wan 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"
@@ -254,8 +235,8 @@ steps:
          - pytest -s -v tests/e2e/accuracy/test_hidream_o1_image.py -m full_model --run-level full_model
        mirror_hardwares: h100_2

-      - label: ":full_moon: Diffusion X2I(&A&T) · Perf Test · Qwen-Image · Single-GPU"
-        key: nightly-diffusion-x2iat-performance-qwen-image-single
+      - 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
@@ -263,23 +244,7 @@ steps:
          - export CACHE_DIT_VERSION=1.5.0
          - |
            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 -m "H100 and cards_1"
-            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 · Qwen-Image · 4-GPU"
-        key: nightly-diffusion-x2iat-performance-qwen-image-4gpu
-        timeout_in_minutes: 180
-        commands:
-          - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
-          - export DIFFUSION_ATTENTION_BACKEND=FLASH_ATTN
-          - export CACHE_DIT_VERSION=1.5.0
-          - |
-            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 -m "H100 and cards_4"
+            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"
@@ -295,34 +260,16 @@ steps:
      build.pull_request.labels includes "diffusion-x2v-test"
    steps:

-      - label: ":full_moon: Diffusion X2V · Wan2.2 T2V Function Test · Single-GPU"
+      - 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 and H100 and cards_1" --run-level "full_model"
-        mirror_hardwares: h100_1
-
-      - label: ":full_moon: Diffusion X2V · Wan2.2 T2V Function Test · 2-GPU"
-        timeout_in_minutes: 90
-        commands:
-          - pytest -s -v tests/e2e/online_serving/test_wan22_expansion.py -k "t2v" -m "full_model and cuda and H100 and cards_2" --run-level "full_model"
+          - 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 · Single-GPU"
+      - 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 and H100 and cards_1" --run-level "full_model"
-        mirror_hardwares: h100_1
-
-      - label: ":full_moon: Diffusion X2V · Wan2.2 I2V/TI2V Function Test · 2-GPU"
-        timeout_in_minutes: 90
-        commands:
-          - pytest -s -v tests/e2e/online_serving/test_wan22_expansion.py -k "i2v" -m "full_model and cuda and H100 and cards_2" --run-level "full_model"
-        mirror_hardwares: h100_2
-
-      - label: ":full_moon: Diffusion X2V · HunyuanVideo-1.5 Function Test"
-        timeout_in_minutes: 90
-        commands:
-          - pytest -s -v tests/e2e/online_serving/test_hunyuan_video_15_expansion.py -m "full_model and cuda" --run-level "full_model"
+          - 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 · Doc Test"
@@ -357,49 +304,17 @@ steps:
          - pytest -s -v tests/e2e/accuracy/hunyuanvideo15_i2v/test_hunyuanvideo15_i2v_video_similarity.py -m full_model --run-level full_model
        mirror_hardwares: h100_2

-      - label: ":full_moon: Diffusion X2V · Perf Test · Single-GPU"
-        key: nightly-diffusion-x2v-performance-single
+      - 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 -m "H100 and cards_1"
-            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 X2V · Perf Test · 2-GPU"
-        key: nightly-diffusion-x2v-performance-2gpu
-        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 -m "H100 and cards_2"
+            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=$$?
-            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_cosmos3_vllm_omni.json -m "H100 and cards_2"
-            EXIT2=$$?
-            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 $$EXIT2
-        mirror_hardwares: h100_2
-
-      - label: ":full_moon: Diffusion X2V · Perf Test · HunyuanVideo-1.5"
-        key: nightly-diffusion-x2v-performance-hunyuanvideo15
-        timeout_in_minutes: 180
-        commands:
-          - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
-          - |
-            set +e
-            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_hunyuanvideo15_t2v_vllm_omni.json
-            EXIT1=$$?
-            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_hunyuanvideo15_i2v_vllm_omni.json
+            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"
@@ -433,16 +348,10 @@ steps:
          - pytest -sv tests/diffusion/distributed/ -m 'full_model and cuda and L4' --run-level "full_model"
        mirror_hardwares: l4_4

-      - label: ":full_moon: Diffusion · Distributed Test with H100 · 2-GPU"
+      - label: ":full_moon: Diffusion · Distributed Test with H100"
        timeout_in_minutes: 60
        commands:
-          - pytest -sv tests/diffusion/distributed/ -m 'full_model and cuda and H100 and cards_2' --run-level "full_model"
-        mirror_hardwares: h100_2
-
-      - label: ":full_moon: Diffusion · Distributed Test with H100 · 4-GPU"
-        timeout_in_minutes: 60
-        commands:
-          - pytest -sv tests/diffusion/distributed/ -m 'full_model and cuda and H100 and cards_4' --run-level "full_model"
+          - pytest -sv tests/diffusion/distributed/ -m 'full_model and cuda and H100' --run-level "full_model"
        mirror_hardwares: h100_4

      - label: ":full_moon: Diffusion · Offloader Test"
@@ -477,25 +386,19 @@ steps:
    depends_on:
      - nightly-omni-performance-no-async-chunk
      - nightly-omni-performance-async-chunk
-      - nightly-tts-performance-single
-      - nightly-tts-performance-2gpu
-      - nightly-diffusion-x2iat-performance-qwen-image-single
-      - nightly-diffusion-x2iat-performance-qwen-image-4gpu
-      - nightly-diffusion-x2v-performance-single
-      - nightly-diffusion-x2v-performance-2gpu
+      - nightly-tts-performance
+      - nightly-diffusion-x2iat-performance-qwen-image
+      - nightly-diffusion-x2v-performance
    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-single
-      - buildkite-agent artifact download "tests/dfx/perf/results/*.json" . --step nightly-tts-performance-2gpu
+      - 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-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-diffusion-x2iat-performance-qwen-image-single
-      - buildkite-agent artifact download "tests/dfx/perf/results/*.json" . --step nightly-diffusion-x2iat-performance-qwen-image-4gpu
-      - buildkite-agent artifact download "tests/dfx/perf/results/*.json" . --step nightly-diffusion-x2v-performance-single
-      - buildkite-agent artifact download "tests/dfx/perf/results/*.json" . --step nightly-diffusion-x2v-performance-2gpu
+      - 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-x2v-performance
      - 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"
@@ -205,7 +205,6 @@ steps:
          - tests/e2e/online_serving/test_minicpmo_realtime_duplex_drivers.py
          - tests/e2e/online_serving/helpers/minicpmo_realtime_duplex_scenarios.py
          - tests/e2e/online_serving/run_minicpmo_realtime_duplex_multi_session.py
-          - tests/e2e/online_serving/run_minicpmo_realtime_duplex_server_vad.py
          - tests/e2e/online_serving/run_minicpmo_realtime_duplex_soft_interrupt.py
          - vllm_omni/deploy/minicpmo_4_5.yaml
          - tests/e2e/online_serving/helpers/minicpmo_4_5_duplex.py
@@ -252,9 +251,6 @@ steps:
          - vllm_omni/model_executor/models/common/alias_free_activation.py
          - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
          - vllm_omni/deploy/qwen3_tts.yaml
-          - vllm_omni/entrypoints/openai/serving_speech.py
-          - vllm_omni/entrypoints/openai/tts_adapters/
-          - vllm_omni/engine/stage_pool.py
        commands:
          - |
            timeout 20m bash -c "
@@ -263,6 +259,25 @@ steps:
            "
        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/models/common/qwen3_code_predictor.py
+          - vllm_omni/model_executor/models/common/snake_activation.py
+          - vllm_omni/model_executor/models/common/alias_free_activation.py
+          - 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: "Diffusion · Wan22 Test"
        source_file_dependencies:
          - tests/e2e/online_serving/test_wan22_t2v.py
@@ -322,15 +337,3 @@ steps:
              pytest -s -v tests/e2e/online_serving/test_qwen_image.py -m 'core_model' --run-level 'core_model'
            "
        mirror_hardwares: h100_1
-
-      - label: "Diffusion · HunyuanVideo-1.5 Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_hunyuan_video_15_expansion.py
-          - vllm_omni/diffusion/models/hunyuan_video/
-        commands:
-          - |
-            timeout 40m bash -c "
-              export VLLM_IMAGE_FETCH_TIMEOUT=60
-              pytest -s -v tests/e2e/online_serving/test_hunyuan_video_15_expansion.py -m 'core_model and diffusion' --run-level 'core_model'
-            "
-        mirror_hardwares: h100_1
@@ -44,16 +44,10 @@ steps:
          - pytest -s -v tests/dfx/reliability/test_reliability_hunyuan_image.py -m "slow"
        mirror_hardwares: h100_4

-      - label: "Reliability Test · Invalid parameters · H100 · Single-GPU"
-        timeout_in_minutes: 90
+      - label: "Reliability Test · Invalid parameters · H100"
+        timeout_in_minutes: 120
        commands:
-          - pytest -s -v tests/dfx/reliability/invalid_param_test/ -m "slow and H100 and cards_1"
-        mirror_hardwares: h100_1
-
-      - label: "Reliability Test · Invalid parameters · H100 · 2-GPU"
-        timeout_in_minutes: 90
-        commands:
-          - pytest -s -v tests/dfx/reliability/invalid_param_test/ -m "slow and H100 and cards_2"
+          - pytest -s -v tests/dfx/reliability/invalid_param_test/ -m "slow and H100"
        mirror_hardwares: h100_2

      - label: "Reliability Test · Invalid parameters · L4"
@@ -111,37 +105,27 @@ steps:
    if: build.env("NON_CRITICAL") == "1"
    steps:

-      # Qwen2.5 expansion (CUDA 4-card / online 2-card) stays on this l4_4 job.
-      - label: "Omni · L4 · Multi-GPU"
+      - 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 -m "slow and L4 and omni and not cards_1" --run-level "core_model"
+          - 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 and not cards_1"
+            -m "slow and L4 and omni"
            --run-level "full_model"
        mirror_hardwares: l4_4

-      - label: "Omni · H100 · Single-GPU"
+      - label: "Omni · H100"
        timeout_in_minutes: 90
        commands:
          - >-
            pytest -sv
-            tests/e2e/ -m "slow and H100 and omni and cards_1"
+            tests/e2e/ -m "slow and H100 and omni"
            --run-level "full_model"
-        mirror_hardwares: h100_1
-
-      - label: "Omni · H100 · 4-GPU"
-        timeout_in_minutes: 90
-        commands:
-          - >-
-            pytest -sv
-            tests/e2e/ -m "slow and H100 and omni and cards_4"
-            --run-level "full_model"
-        mirror_hardwares: h100_4
+        mirror_hardwares: h100_2

      - label: "TTS · L4"
        timeout_in_minutes: 120
@@ -185,21 +169,12 @@ steps:
            --run-level "full_model"
        mirror_hardwares: l4_1

-      - label: "Diffusion · L4 · Single-GPU"
+      - label: "Diffusion · L4"
        timeout_in_minutes: 60
        commands:
          - >-
            pytest -sv
-            tests/e2e/ -m "slow and diffusion and L4 and cards_1"
-            --run-level "full_model"
-        mirror_hardwares: l4_1
-
-      - label: "Diffusion · L4 · Multi-GPU"
-        timeout_in_minutes: 60
-        commands:
-          - >-
-            pytest -sv
-            tests/e2e/ -m "slow and diffusion and L4 and not cards_1"
+            tests/e2e/ -m "slow and diffusion and L4"
            --run-level "full_model"
        mirror_hardwares: l4_4

@@ -208,24 +183,15 @@ steps:
        commands:
          - >-
            pytest -sv
-            tests/e2e/ -m "slow and diffusion and H100 and cards_1"
+            tests/e2e/ -m "slow and diffusion and H100 and not distributed_cuda"
            --run-level "full_model"
        mirror_hardwares: h100_1

-      - label: "Diffusion · H100 · 2-GPU"
+      - label: "Diffusion · H100 · Multi-GPU"
        timeout_in_minutes: 90
        commands:
          - >-
            pytest -sv
-            tests/e2e/ -m "slow and diffusion and H100 and cards_2"
-            --run-level "full_model"
-        mirror_hardwares: h100_2
-
-      - label: "Diffusion · H100 · 4-GPU"
-        timeout_in_minutes: 90
-        commands:
-          - >-
-            pytest -sv
-            tests/e2e/ -m "slow and diffusion and H100 and cards_4"
+            tests/e2e/ -m "slow and diffusion and H100 and distributed_cuda"
            --run-level "full_model"
        mirror_hardwares: h100_4
@@ -10,7 +10,7 @@ steps:
          DOCKER_BUILDKIT: "1"
          # Buildkite will automatically replace this with the actual commit hash
          VLLM_IMAGE_TAG: "${BUILDKITE_COMMIT}"
-          VLLM_VERSION: "v0.28.0"
+          VLLM_VERSION: "v0.27.0"
        priority: 100
        timeout_in_minutes: 60
        soft_fail: false
@@ -5,12 +5,7 @@ set -ex

 omni_source_dir=$(git rev-parse --show-toplevel)

-: "${VLLM_VERSION:?VLLM_VERSION must be set}"
-# Official upstream XPU image. It is published from vllm's own
-# docker/Dockerfile.xpu, which is exactly what build_vllm_base() clones and
-# builds below, so pulling it is equivalent to building the fallback base.
-upstream_base_image="vllm/vllm-openai-xpu:${VLLM_VERSION}"
-local_base_image="xpu/vllm-omni-ci-base:${VLLM_VERSION}"
+base_image_name="xpu/vllm-omni-ci-base:${VLLM_VERSION:?VLLM_VERSION must be set}"
 image_name="xpu/vllm-omni-ci:${BUILDKITE_COMMIT:?BUILDKITE_COMMIT must be set}"
 container_name="xpu_${BUILDKITE_COMMIT}_$(
    tr -dc A-Za-z0-9 </dev/urandom | head -c 10
@@ -54,36 +49,9 @@ build_vllm_base() (
        "${vllm_source_dir}"
 )

-# Resolve the vLLM base image, in priority order:
-#   1. the published upstream image for VLLM_VERSION, pulled fresh (so moving
-#      tags like nightly are refreshed rather than served stale from disk);
-#   2. a local copy of that same image, when the registry is unreachable;
-#   3. VLLM_BASE, if that image is already on disk;
-#   4. otherwise build VLLM_BASE from the matching vLLM source tag (~30min).
-# VLLM_BASE names the fallback base and defaults to ${local_base_image}; set it
-# to reuse or produce a base under a different tag.
-resolve_vllm_base() {
-    if docker pull "${upstream_base_image}"; then
-        base_image_name="${upstream_base_image}"
-        return
-    fi
-
-    # Registry unreachable, but a previous run may have left the image behind.
-    if [ -n "$(docker images -q "${upstream_base_image}")" ]; then
-        echo "WARNING: could not pull ${upstream_base_image}; using local copy" >&2
-        base_image_name="${upstream_base_image}"
-        return
-    fi
-
-    echo "WARNING: ${upstream_base_image} unavailable remotely and locally" >&2
-    base_image_name="${VLLM_BASE:-${local_base_image}}"
-    if [ -z "$(docker images -q "${base_image_name}")" ]; then
-        echo "WARNING: building ${base_image_name} from vLLM ${VLLM_VERSION} source" >&2
-        build_vllm_base
-    fi
-}
-
-resolve_vllm_base
+if [ -z "$(docker images -q "${base_image_name}")" ]; then
+    build_vllm_base
+fi

 # Try building the docker image
 docker_build "${image_name}" --build-arg "VLLM_BASE=${base_image_name}" --build-arg "VLLM_VERSION=${VLLM_VERSION}"
@@ -10,18 +10,6 @@ steps:
        -m 'core_model and npu and A3 and omni'
        --run-level 'core_model'

-  - label: "Diffusion NPU layer tests"
-    key: ready-diffusion-npu-layer-tests
-    depends_on: upload-ready-pipeline
-    mirror_hardwares: a3_npu_2
-    commands:
-      - >-
-        pytest -s -v
-        tests/diffusion/layers/test_activation.py
-        tests/diffusion/layers/test_fused_qk_norm_rope_npu.py
-        -m 'core_model and npu and A3'
-        --run-level 'core_model'
-
  - group: ":card_index_dividers: Omni Model Test"
    key: ready-omni-group
    depends_on: upload-ready-pipeline
@@ -338,7 +338,7 @@ Classify the model's **CI priority** first:

 **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 (`cards_1` vs `not cards_1`).
+**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.

@@ -34,7 +34,7 @@ but the changes are correct — stage the modified files and re-commit.
 | `check-forbidden-imports` | Stdlib `re`/`base64`, pickle, Hugging Face Hub API, or direct Triton/TileLang | `import regex as re` and `pybase64`; use `vllm.transformers_utils.repo_utils`. Do **not** add the file to `allowed_files` without review |
 | `check-torch-cuda-call` | New `torch.cuda.*` call site | Use `current_omni_platform` / `OmniPlatform`; do not grow `ALLOWED_FILES` without review |
 | `check-tts-adapter-migration` | New `self._tts_model_type` branch in `serving_speech.py` | Put per-model logic in `tts_adapters/`. Lower `MAX_MODEL_TYPE_BRANCHES` when removing branches; do not raise it without review |
-| `check-mark` | New `tests/**/test_*.py` missing level/hardware mark, or direct `pytest.mark.H100` | Add `core_model`/`advanced_model`/… plus `cpu`/`cuda`/`hardware_test(`; SKU marks only via helpers |
+| `check-test-ci-coverage` | New `tests/**/test_*.py` missing level or hardware mark | Add `core_model`/`advanced_model`/… plus `cpu`/`cuda`/`hardware_test(` |
 | `mypy-3.10` | Type error on changed `vllm_omni/` files | Fix the types; model trees are excluded. Extra versions: `pre-commit run --hook-stage manual mypy-3.12` |
 | `markdownlint-cli2` | Docs/README markdown lint | Accept auto-fix under `docs/` / `recipes/` / root README |
 | `check-buildkite` | Invalid `.buildkite/*.yml` | Fix the YAML; do not grow `SKIP_FILES` without review |
@@ -52,8 +52,8 @@ Always attach markers deliberately:
  - `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`, …; SKU (`H100`, `L4`, …) and `cards_{n}` via `@hardware_test(...)` / `hardware_marks(...)` only — never `pytest.mark.H100` (`check-mark` rejects it)
- Optional: `slow`. Multi-card: `num_cards=N` on the helper (filter with `cards_N` / `not cards_1`)
+- **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`**.

@@ -274,7 +274,7 @@ def test_text_to_video_001(omni_server, online_client) -> None:

 **3. Cross-cutting rules**

- New `.py` files need the Omni SPDX header (`Copyright contributors to the vLLM-Omni project`, not `vLLM project`). `tests/` may keep stdlib `re`/`base64`; pickle is still banned unless the file is already on the pickle allowlist. Every collected `tests/**/test_*.py` needs a CI **level** mark and a **hardware** platform mark/helper (`check-mark` runs locally; GHA skips it). Do not write `pytest.mark.H100` / `pytest.mark.L4`; use the helpers so `cards_{n}` is attached. Do not grow `allowed_files` to land a test. Policy: [docs/contributing/README.md](../../../docs/contributing/README.md#linting).
+- New `.py` files need the Omni SPDX header (`Copyright contributors to the vLLM-Omni project`, not `vLLM project`). `tests/` may keep stdlib `re`/`base64`; pickle is still banned unless the file is already on the pickle allowlist. Every collected `tests/**/test_*.py` needs a CI **level** mark and a **hardware** mark/helper (`check-test-ci-coverage` runs locally; GHA skips it). Do not grow `allowed_files` to land a test. Policy: [docs/contributing/README.md](../../../docs/contributing/README.md#linting).
 - 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).
@@ -79,7 +79,7 @@ E2E (L2+) should not rely on mocks unless documenting a rare exception; prefer r
   - `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`, `cards_1`, …
+4. **Hardware**: `cpu`, `cuda`, `rocm`, `npu`, `L4`, `H100`, `distributed_cuda`, …
 5. Multi-card: `@hardware_test(...)` in `tests/helpers/mark.py`

 ## Command Templates
@@ -244,7 +244,7 @@ pytest -s -v dfx/reliability/invalid_param_test/test_invalid_image_generation.py

 ```bash
 cd tests
-pytest -s -v -m "core_model and L4 and not cards_1" --run-level=core_model
+pytest -s -v -m "core_model and distributed_cuda and L4" --run-level=core_model
 ```

 ### Concrete e2e paths (common in-tree)
@@ -27,4 +27,4 @@ jobs:
        # Don't run historical-debt hooks against the whole tree in CI yet.
        # They still run locally on changed files.
        env:
-          SKIP: check-mark,markdownlint-cli2,shellcheck,check-spdx-header,mypy-3.10
+          SKIP: check-test-ci-coverage,markdownlint-cli2,shellcheck,check-spdx-header,mypy-3.10
@@ -105,8 +105,8 @@ repos:
      <<: *mypy_common
      stages: [manual]

-    - id: check-mark
-      name: Ensure test files have CI marks (no direct SKU)
+    - id: check-test-ci-coverage
+      name: Ensure test files have pytestmarks
      entry: python tools/pre_commit/check_test_marks.py
      language: python
      types: [python]
@@ -16,7 +16,6 @@ Easy, fast, and cheap omni-modality model serving for everyone

 *Latest News* 🔥

- [2026/08] We released [0.28.0](https://github.com/vllm-project/vllm-omni/releases/tag/v0.28.0), featuring production-ready [MiniMax H3](recipes/MiniMaxAI/MiniMax-H3.md) serving on GPU and NPU, a unified AR/DiT paged KV cache runtime, and enhanced realtime full-duplex serving for the [MiniCPM-o series](recipes/OpenBMB/MiniCPM-o-4_5.md).
 - [2026/08] [VeRL-Omni](https://github.com/verl-project/verl-omni) `v0.2.0` is released: faster diffusion RL powered by vLLM-Omni (request-level/step-wise batching with FA3), rebuilt Qwen3-Omni multimodal training (DPO & GSPO), plus LTX-2.3, Qwen-Image-Edit support and more. See the [release notes](https://github.com/verl-project/verl-omni/releases/tag/v0.2.0).
 - [2026/08] We released [0.26.0](https://github.com/vllm-project/vllm-omni/releases/tag/v0.26.0) - aligned with the vLLM 0.26 release line, featuring [MiniMax H3](recipes/MiniMaxAI/MiniMax-H3.md) joint video/audio generation, an experimental full-duplex realtime runtime for [MiniCPM-o 4.5](recipes/OpenBMB/MiniCPM-o-4_5.md), distributed layerwise diffusion offload, and broader model, hardware, streaming, TTS, and quantization support.
 - [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.
@@ -58,9 +57,9 @@ 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. Qwen3-Omni, MiniCPM-o 4.5, Cosmos3, HunyuanImage, BAGEL)
- **TTS models** (e.g. Qwen3-TTS, IndexTTS 2.5, dots.tts, CosyVoice3)
- **Diffusion models** — image, video, and audio generation (e.g. MiniMax H3, LTX-2.5, SANA-Video, Wan2.2)
- **Robot-policy and action models** (e.g. π0, GR00T-N1.7, DreamZero-DROID, InternVLA-A1)
+- **TTS models** (e.g. Qwen3-TTS, VoxCPM2, Ming-Omni-TTS, CosyVoice3)
+- **Diffusion models** — image, video, and audio generation (e.g. MiniMax H3, Qwen-Image, Wan2.2, FLUX)
+- **Robot-policy and action models** (e.g. GR00T-N1.7, DreamZero-DROID, InternVLA-A1, Cosmos3 action policy)

 ## Getting Started

@@ -68,7 +67,6 @@ 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/)
- [Nightly CUDA Docker images](https://hub.docker.com/r/vllm/vllm-omni/tags?name=nightly) built from the latest `main`
 - [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

@@ -1,5 +1,5 @@
 ARG VLLM_ASCEND_IMAGE=quay.io/atlas-ci/vllm-ascend
-ARG VLLM_ASCEND_TAG=v0.28.0
+ARG VLLM_ASCEND_TAG=v0.27.0
 FROM ${VLLM_ASCEND_IMAGE}:${VLLM_ASCEND_TAG}

 # WORKDIR /vllm-workspace/vllm
@@ -30,7 +30,6 @@ RUN python3 -m pip install -v --no-deps \
      "stepaudio2-minicpmo>=0.1.1" \
      "step-audio2>=1.0.0"

-ENV VLLM_TARGET_DEVICE=npu
 ENV VLLM_WORKER_MULTIPROC_METHOD=spawn

 ENTRYPOINT []
@@ -1,5 +1,5 @@
 ARG VLLM_ASCEND_IMAGE=quay.io/atlas-ci/vllm-ascend
-ARG VLLM_ASCEND_TAG=v0.28.0-a3
+ARG VLLM_ASCEND_TAG=v0.27.0-a3
 FROM ${VLLM_ASCEND_IMAGE}:${VLLM_ASCEND_TAG}

 # WORKDIR /vllm-workspace/vllm
@@ -30,7 +30,6 @@ RUN python3 -m pip install -v --no-deps \
      "stepaudio2-minicpmo>=0.1.1" \
      "step-audio2>=1.0.0"

-ENV VLLM_TARGET_DEVICE=npu
 ENV VLLM_WORKER_MULTIPROC_METHOD=spawn

 ENTRYPOINT []
@@ -1,5 +1,5 @@
 ARG VLLM_ASCEND_IMAGE=quay.io/atlas-ci/vllm-ascend
-ARG VLLM_ASCEND_TAG=v0.28.0
+ARG VLLM_ASCEND_TAG=v0.27.0
 FROM ${VLLM_ASCEND_IMAGE}:${VLLM_ASCEND_TAG}

 ARG APP_DIR=/vllm-workspace/vllm-omni
@@ -65,7 +65,6 @@ RUN python3 -m pip install -v --no-deps \
      --trusted-host cache-service.nginx-pypi-cache.svc.cluster.local \
      --default-timeout=60

-ENV VLLM_TARGET_DEVICE=npu
 ENV VLLM_WORKER_MULTIPROC_METHOD=spawn

 ENTRYPOINT []
@@ -1,5 +1,5 @@
 ARG VLLM_ASCEND_IMAGE=quay.io/atlas-ci/vllm-ascend
-ARG VLLM_ASCEND_TAG=v0.28.0-a3
+ARG VLLM_ASCEND_TAG=v0.27.0-a3
 FROM ${VLLM_ASCEND_IMAGE}:${VLLM_ASCEND_TAG}

 ARG APP_DIR=/vllm-workspace/vllm-omni
@@ -66,7 +66,6 @@ RUN python3 -m pip install -v --no-deps \
      --trusted-host cache-service.nginx-pypi-cache.svc.cluster.local \
      --default-timeout=60

-ENV VLLM_TARGET_DEVICE=npu
 ENV VLLM_WORKER_MULTIPROC_METHOD=spawn

 ENTRYPOINT []
@@ -1,33 +1,140 @@
-# Target vLLM version: v0.28.0. Keep VLLM_VERSION aligned with the target vLLM
-# release on every rebase.
-#
-# This file layers vLLM-Omni on top of the published upstream XPU image, which
-# already ships a self-consistent XPU runtime (torch-xpu, oneCCL/UCX+NIXL, the
-# pinned UMD/compute-runtime stack, and vLLM itself) built from
-# https://github.com/vllm-project/vllm/blob/main/docker/Dockerfile.xpu. Docker
-# pulls that base automatically, so there is nothing to build here beyond the
-# Omni layers. Mirrors how docker/Dockerfile.ci consumes vllm/vllm-openai.
-#
-# If the tag does not exist yet (targeting an unreleased vLLM commit), build the
-# base from upstream's own Dockerfile and point VLLM_BASE at the result:
-#
-#   git clone --depth 1 --branch $VLLM_VERSION https://github.com/vllm-project/vllm /tmp/vllm
-#   docker build -t vllm-openai-xpu:local --target vllm-openai \
-#       -f /tmp/vllm/docker/Dockerfile.xpu /tmp/vllm
-#   docker build -f docker/Dockerfile.xpu --build-arg VLLM_BASE=vllm-openai-xpu:local .
-#
-# .buildkite/intel/scripts/run-xpu-test.sh automates that pull-then-build fallback.
-ARG VLLM_BASE_IMAGE=vllm/vllm-openai-xpu
-ARG VLLM_VERSION=v0.28.0
-ARG VLLM_BASE=${VLLM_BASE_IMAGE}:${VLLM_VERSION}
+# Argument to configure vllm base image if pre-built
+ARG VLLM_BASE=vllm-base
+
+FROM intel/deep-learning-essentials:2025.3.2-0-devel-ubuntu24.04 AS vllm-base
+
+WORKDIR /workspace/
+
+ARG PYTHON_VERSION=3.12
+ARG PIP_EXTRA_INDEX_URL="https://download.pytorch.org/whl/xpu"
+
+RUN wget -O- https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB | gpg --dearmor | tee /usr/share/keyrings/oneapi-archive-keyring.gpg > /dev/null && \
+    echo "deb [signed-by=/usr/share/keyrings/oneapi-archive-keyring.gpg] https://apt.repos.intel.com/oneapi all main" | tee /etc/apt/sources.list.d/oneAPI.list
+
+RUN apt clean && apt-get update -y && \
+    apt-get install -y --no-install-recommends --fix-missing \
+    curl \
+    ffmpeg \
+    git \
+    libsndfile1 \
+    libsm6 \
+    libxext6 \
+    libgl1 \
+    lsb-release \
+    libaio-dev \
+    numactl \
+    wget \
+    vim \
+    python3.12 \
+    python3.12-dev \
+    python3-pip
+
+# Add oneAPI repo, pin oneAPI to 2025.3, then install pinned packages in one layer.
+RUN wget -O- https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB | gpg --dearmor | tee /usr/share/keyrings/oneapi-archive-keyring.gpg > /dev/null && \
+    echo "deb [signed-by=/usr/share/keyrings/oneapi-archive-keyring.gpg] https://apt.repos.intel.com/oneapi all main" | tee /etc/apt/sources.list.d/oneAPI.list && \
+    printf '%s\n' \
+    'Package: intel-oneapi-* intel-deep-learning-essentials* intel-pti*' \
+    'Pin: version 2025.3*' \
+    'Pin-Priority: 1001' \
+    > /etc/apt/preferences.d/oneapi-2025.3.pref && \
+    apt-get update -y && \
+    apt-get install -y --no-install-recommends \
+    intel-oneapi-compiler-dpcpp-cpp-2025.3 \
+    intel-oneapi-mkl-devel-2025.3 \
+    intel-oneapi-dnnl-devel-2025.3 && \
+    rm -rf /var/lib/apt/lists/*
+
+# Install UMD with fixed version
+RUN mkdir neo && \
+    cd neo && \
+    wget https://github.com/intel/intel-graphics-compiler/releases/download/v2.24.8/intel-igc-core-2_2.24.8+20344_amd64.deb && \
+    wget https://github.com/intel/intel-graphics-compiler/releases/download/v2.24.8/intel-igc-opencl-2_2.24.8+20344_amd64.deb && \
+    wget https://github.com/intel/compute-runtime/releases/download/25.48.36300.8/intel-ocloc_25.48.36300.8-0_amd64.deb && \
+    wget https://github.com/intel/compute-runtime/releases/download/25.48.36300.8/intel-opencl-icd_25.48.36300.8-0_amd64.deb && \
+    wget https://github.com/intel/compute-runtime/releases/download/25.48.36300.8/libigdgmm12_22.8.2_amd64.deb && \
+    wget https://github.com/intel/compute-runtime/releases/download/25.48.36300.8/libze-intel-gpu1_25.48.36300.8-0_amd64.deb && \
+    wget https://github.com/oneapi-src/level-zero/releases/download/v1.26.0/level-zero_1.26.0+u24.04_amd64.deb && \
+    dpkg -i *.deb && \
+    cd .. && \
+    rm -rf neo
+
+ENV PATH="/root/.local/bin:$PATH"
+ENV VIRTUAL_ENV="/opt/venv"
+ENV UV_PYTHON_INSTALL_DIR=/opt/uv/python
+RUN curl -LsSf https://astral.sh/uv/install.sh | sh
+RUN uv venv --python ${PYTHON_VERSION} --seed ${VIRTUAL_ENV}
+ENV PATH="$VIRTUAL_ENV/bin:$PATH"
+
+# This oneccl contains the BMG support which is not the case for default version of oneapi 2025.3.
+ARG ONECCL_INSTALLER="intel-oneccl-2021.15.9.14_offline.sh"
+RUN wget "https://github.com/uxlfoundation/oneCCL/releases/download/2021.15.9/${ONECCL_INSTALLER}" && \
+    bash "${ONECCL_INSTALLER}" -a --silent --eula accept && \
+    rm "${ONECCL_INSTALLER}" && \
+    echo "source /opt/intel/oneapi/setvars.sh --force" >> /root/.bashrc && \
+    echo "source /opt/intel/oneapi/ccl/2021.15/env/vars.sh --force" >> /root/.bashrc
+RUN rm -f /opt/intel/oneapi/ccl/latest && \
+    ln -s /opt/intel/oneapi/ccl/2021.15 /opt/intel/oneapi/ccl/latest
+
+SHELL ["bash", "-c"]
+CMD ["bash", "-c", "source /root/.bashrc && exec bash"]
+
+WORKDIR /workspace/
+
+ENV UV_HTTP_TIMEOUT=500
+
+# Configure package index for XPU
+ENV PIP_EXTRA_INDEX_URL=${PIP_EXTRA_INDEX_URL}
+ENV UV_EXTRA_INDEX_URL=${PIP_EXTRA_INDEX_URL}
+ENV UV_INDEX_STRATEGY="unsafe-best-match"
+ENV UV_LINK_MODE="copy"
+
+ARG VLLM_VERSION=v0.27.0
+RUN git clone -b ${VLLM_VERSION} https://github.com/vllm-project/vllm
+WORKDIR /workspace/vllm
+
+RUN --mount=type=cache,target=/root/.cache/uv \
+    uv pip install --upgrade pip && \
+    uv pip install -r requirements/xpu.txt && \
+    uv pip install setuptools_scm &&\
+    uv pip install grpcio-tools protobuf nanobind && \
+    source /opt/intel/oneapi/setvars.sh --force && \
+    source /opt/intel/oneapi/ccl/2021.15/env/vars.sh --force && \
+    export CMAKE_PREFIX_PATH="$(python3 -c 'import site; print(site.getsitepackages()[0])'):${CMAKE_PREFIX_PATH}"
+
+ENV LD_LIBRARY_PATH="/opt/intel/oneapi/ccl/2021.15/lib:$LD_LIBRARY_PATH:/usr/local/lib/"
+
+ENV VLLM_TARGET_DEVICE=xpu
+ENV VLLM_WORKER_MULTIPROC_METHOD=spawn
+
+RUN --mount=type=cache,target=/root/.cache/uv \
+    uv pip uninstall triton triton-xpu && \
+    uv pip install triton-xpu==3.7.2 && \
+    uv pip install --no-build-isolation --no-deps .
+
+CMD ["/bin/bash"]
+
+FROM vllm-base AS vllm-openai
+
+# install nixl from source code
+RUN uv pip install nixl
+
+# install additional dependencies for openai api server
+RUN --mount=type=cache,target=/root/.cache/uv \
+    uv pip install accelerate hf_transfer pytest pytest_asyncio lm_eval[api] modelscope
+
+# install development dependencies (for testing)
+RUN uv pip install -e tests/vllm_test_utils
+
+# ensure vllm is properly installed
+RUN python -c "import vllm, inspect; print(vllm.__file__)"
+RUN uv pip show vllm
+
+CMD ["/bin/bash"]
+
+ENTRYPOINT []

 FROM ${VLLM_BASE} AS vllm-omni

-# The base image ships ENTRYPOINT ["vllm", "serve"]; clear it so this stage stays
-# usable as a plain shell/test container regardless of which VLLM_BASE was used.
-ENTRYPOINT []
-CMD ["/bin/bash"]
-
 RUN apt-get update && \
    apt-get install -y espeak-ng ffmpeg git jq && \
    apt-get clean && \
@@ -5,22 +5,14 @@ nav:
    - getting_started/quickstart.md
    - getting_started/installation/*
  - Serving:
-    - API Server:
-      - Overview: serving/README.md
-      - Core APIs:
-        - Chat Completions: serving/chat_completions_api.md
-        - Text to Speech: serving/speech_api.md
-        - Audio Generation: serving/audio_generate_api.md
-        - Image Generation: serving/image_generation_api.md
-        - Image Edit: serving/image_edit_api.md
-        - Videos: serving/videos_api.md
-      - Streaming and Realtime:
-        - Realtime Audio: serving/realtime_api.md
-        - Full Duplex: serving/full_duplex_api.md
-        - Streaming Video Input: serving/video_stream_api.md
-        - Streaming Video Output: serving/streaming_video_output_api.md
-        - OpenPI Robot Policy: serving/openpi_api.md
-      - Standalone Experimental Servers: serving/standalone_servers.md
+    - OpenAI-Compatible API:
+      - Chat Completions API: serving/chat_completions_api.md
+      - Image Generation: serving/image_generation_api.md
+      - Image Edit: serving/image_edit_api.md
+      - Text to Speech: serving/speech_api.md
+      - Videos: serving/videos_api.md
+      - Audio Generation: serving/audio_generate_api.md
+      - Streaming Video Input: serving/video_stream_api.md
  - Examples:
    - examples/README.md
    - Offline Inference:
@@ -37,7 +29,7 @@ nav:
      - Image-To-Video: user_guide/examples/online_serving/image_to_video.md
      - Speech-To-Video: user_guide/examples/online_serving/speech_to_video.md
      - Text-To-Image: user_guide/examples/online_serving/text_to_image.md
-      - Text-To-Speech (Online Serving): user_guide/examples/online_serving/text_to_speech.md
+      - Text-To-Speech: user_guide/examples/online_serving/text_to_speech.md
      - Text-To-Video: user_guide/examples/online_serving/text_to_video.md
  - General:
    - usage/*
@@ -127,8 +119,6 @@ nav:
        - Async Omni Output Materialization: design/feature/omni_async_output_materialization.md
        - Automatic Prefix Caching: design/feature/prefix_caching.md
        - Realtime AR-Diffusion: design/feature/realtime_ar_diffusion.md
-      - Model-Specific Capabilities:
-        - MiniMax H3 Temporal VAE Chunks: design/feature/minimax_h3_temporal_vae_chunks.md
      - Communication:
        - OmniConnector Implementations:
          - Mooncake Store Connector: design/feature/omni_connectors/mooncake_store_connector.md
@@ -57,9 +57,9 @@ 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. Qwen3-Omni, MiniCPM-o 4.5, Cosmos3, HunyuanImage, BAGEL)
- TTS models (e.g. Qwen3-TTS, IndexTTS 2.5, dots.tts, CosyVoice3)
- Diffusion models — image, video, and audio generation (e.g. MiniMax H3, LTX-2.5, SANA-Video, Wan2.2)
- Robot-policy and action models (e.g. π0, GR00T-N1.7, DreamZero-DROID, InternVLA-A1)
+- TTS models (e.g. Qwen3-TTS, VoxCPM2, Ming-Omni-TTS, CosyVoice3)
+- Diffusion models — image, video, and audio generation (e.g. MiniMax H3, Qwen-Image, Wan2.2, FLUX)
+- Robot-policy and action models (e.g. GR00T-N1.7, DreamZero-DROID, InternVLA-A1, Cosmos3 action policy)

 For more information, checkout the following:

@@ -2,7 +2,7 @@

 This section lists the most common options for running vLLM-Omni.

-For options within a vLLM Engine, please refer to the [vLLM 0.28 configuration guide](https://docs.vllm.ai/en/v0.28.0/configuration/index.html).
+For options within a vLLM Engine, please refer to the [vLLM 0.26 configuration guide](https://docs.vllm.ai/en/v0.26.0/configuration/index.html).

 Each model defines fixed topology in a registered `PipelineConfig` and runtime
 overrides in a deploy YAML.
@@ -151,7 +151,7 @@ their keys only.
 ## Inherited vLLM variables

 vLLM-Omni also reads variables through its aligned vLLM dependency. Refer to
-the [vLLM 0.28 environment-variable reference](https://docs.vllm.ai/en/v0.28.0/configuration/env_vars.html)
+the [vLLM 0.26 environment-variable reference](https://docs.vllm.ai/en/v0.26.0/configuration/env_vars.html)
 for their definitions. This includes vLLM launch, cache, logging, plugin, ROCm,
 XPU, ModelScope, and FlashInfer workspace settings.

@@ -49,7 +49,7 @@ vLLM-Omni's `pre-commit` hooks will now run automatically every time you commit.
    `--all-files` run stays green while historical debt is cleaned up. The
    current `SKIP` list in
    [`.github/workflows/pre-commit.yml`](https://github.com/vllm-project/vllm-omni/blob/main/.github/workflows/pre-commit.yml)
-    is `check-mark`, `markdownlint-cli2`, `shellcheck`,
+    is `check-test-ci-coverage`, `markdownlint-cli2`, `shellcheck`,
    `check-spdx-header`, and `mypy-3.10`. Those hooks still run on **your
    commit** for changed files. A passing GitHub pre-commit check does not mean
    they passed locally. Hooks **not** on that list (forbidden imports,
@@ -66,7 +66,7 @@ already there). `check-pickle-imports` is gone: pickle is now one rule inside
 | `markdownlint-cli2` | Markdown in `docs/`, `recipes/`, `README.md`, `CONTRIBUTING.md` (not `.claude/` / `.cursor/` / `CLAUDE.md`) | skipped |
 | `mypy-3.10` | Type-check changed `vllm_omni/` files (model trees excluded). `tests/` uses `--follow-imports skip` | skipped |
 | `mypy-3.11` / `3.12` / `3.13` | Same checker, extra Python versions | not installed; `pre-commit run --hook-stage manual mypy-3.12` |
-| `check-mark` | Every `tests/**/test_*.py` has a CI level mark and a hardware platform mark or helper; no direct `pytest.mark.H100` / SKU | skipped |
+| `check-test-ci-coverage` | Every `tests/**/test_*.py` has a CI level mark and a hardware mark/helper | skipped |
 | `check-tts-adapter-migration` | `self._tts_model_type` branches in `serving_speech.py` must not increase | runs |
 | `shellcheck` | `*.sh` quoting / undefined vars | skipped |
 | `check-spdx-header` | Omni SPDX header on `.py` / `.pyi` / `.sh` / `.rs` / `.proto` | skipped |
@@ -144,13 +144,11 @@ pre-commit run --hook-stage manual mypy-3.12

 #### Test CI marks

-`check-mark` requires each collected test module under `tests/` to
+`check-test-ci-coverage` requires each collected test module under `tests/` to
 have at least one CI **level** mark (`core_model`, `advanced_model`,
-`full_model`, `local_model`, or `slow`) and a **hardware** platform mark
-(`cpu`, `cuda`, …) or helper (`hardware_test(` / `hardware_marks(`). SKU marks
-(`H100`, `L4`, …) must come from those helpers so `cards_{n}` is attached;
-direct `pytest.mark.H100` is rejected. GitHub Actions skips this hook; local
-commit does not. See the
+`full_model`, `local_model`, or `slow`) and a **hardware** mark (`cpu`, `cuda`,
+`H100`, …) or helper (`hardware_test(` / `hardware_marks(`). GitHub Actions
+skips this hook; local commit does not. See the
 [test writing guide](./ci/test_writing_guide.md).

 #### TTS adapter ratchet
@@ -402,7 +402,7 @@ loads the model once per mode.
 Before splitting, confirm each half actually collects at least one test under the job's forwarded filters — pytest exits with code 5 ("no tests collected") if a half is empty, which the script reports as a failure, red-ing a job that used to pass as one combined invocation.

 Also check the tests' `num_cards` against the job's `mirror_hardwares` preset.
-`_cuda_marks` turns `num_cards > 1` into `skipif(device_count() < num_cards)`, so a
+`cuda_marks` turns `num_cards > 1` into `skipif(device_count() < num_cards)`, so a
 test asking for more GPUs than the preset provides is silently skipped and its
 coverage artifact comes back empty. Splitting such a job produces a file that
 looks fine and measures nothing — "Diffusion · Z Image Test" is the current
@@ -81,8 +81,8 @@ Supported form:

 | Form | Example | Effect |
 | ---- | ------- | ------ |
-| Array (single platform) | `"mark": [{"hardware_marks": {"res": {"cuda": "H100"}, "num_cards": 2}}, "full_model", "diffusion"]` | Hardware marks from `hardware_marks(...)`; `num_cards` adds `cards_{n}` (including `cards_1` when omitted). String entries become extra pytest marks. |
-| Array (multi-platform) | `"mark": [{"hardware_marks": {"res": {"cuda": "H100", "rocm": "MI325"}, "num_cards": 2}}, "full_model", "omni"]` | Same as above, but declares **multiple platforms** on one item. All platforms must share the same `num_cards` (int or identical dict values). Different counts need `@hardware_test` or one `hardware_marks()` per `pytest.param`. Filter examples: `-m "full_model and H100 and cards_2 and cuda"`, `-m "full_model and MI325 and cards_2 and rocm"`. |
+| Array (single platform) | `"mark": [{"hardware_marks": {"res": {"cuda": "H100"}, "num_cards": 2}}, "full_model", "diffusion"]` | Hardware marks from `hardware_marks(...)`; `num_cards` > 1 adds `distributed_*` (+ CUDA `skipif_cuda` when applicable). String entries become extra pytest marks. |
+| Array (multi-platform) | `"mark": [{"hardware_marks": {"res": {"cuda": "H100", "rocm": "MI325", "npu": "A2"}, "num_cards": {"cuda": 2, "rocm": 2, "npu": 1}}}, "full_model", "omni"]` | Same as above, but declares **multiple platforms** in one case. Filter examples: `-m "full_model and H100 and cuda"`, `-m "full_model and MI325 and rocm"`. |

 Recommended for L4 perf cases:

@@ -157,7 +157,7 @@ pytest -s -v tests/dfx/perf/scripts/run_benchmark.py \
 ```

 <!-- markdownlint-disable-next-line MD051 -->
-See also [Markers for Tests](#markers-for-tests) for registered hardware markers (`H100`, `L4`, `cuda`, `cards_1`, …).
+See also [Markers for Tests](#markers-for-tests) for registered hardware markers (`H100`, `L4`, `cuda`, `distributed_cuda`, …).

 #### `server_params` Configuration

@@ -88,7 +88,7 @@ Failed jobs: 1/2
    ```bash
    cd tests
    pytest -s -v test_xxxx.py --run-level=core_model
-    pytest -s -v -m "core_model and L4 and not cards_1" --run-level=core_model
+    pytest -s -v -m "core_model and distributed_cuda and L4" --run-level=core_model
    ```

 === "L3 level"
@@ -119,7 +119,7 @@ Failed jobs: 1/2

    ```bash
    pytest -s -v test_xxxx.py --run-level=advanced_model
-    pytest -s -v -m "advanced_model and L4 and not cards_1" --run-level=advanced_model
+    pytest -s -v -m "advanced_model and distributed_cuda and L4" --run-level=advanced_model
    ```

 === "L4 level"
@@ -153,12 +153,12 @@ Failed jobs: 1/2
    ```bash
    cd tests
    pytest -s -v test_xxxx.py --run-level=full_model
-    pytest -s -v -m "full_model and L4 and not cards_1" --run-level=full_model
+    pytest -s -v -m "full_model and distributed_cuda and L4" --run-level=full_model
    pytest -s -v -m "full_model and (omni or tts) and H100" --run-level=full_model
    ```
    If you only want to run specific test cases on a particular platform, you can use:
    ```bash
-    pytest -s -v -m "full_model and L4 and not cards_1"  --run-level=full_model
+    pytest -s -v -m "full_model and distributed_cuda and L4"  --run-level=full_model
    ```
    Note: ``run_benchmark.py`` and ``run_diffusion_benchmark.py`` accept an optional ``--test-config-file``. If omitted, each loads every ``*.json`` under ``tests/dfx/perf/tests/`` (omni/tts vs diffusion split by ``is_diffusion_perf_config``) and pytest ``-m`` filters by each case's JSON ``mark``:
    ```bash
@@ -48,12 +48,17 @@ Defined in `pyproject.toml`:
 | `MI325`            | Tests that require MI325 GPU (AMD/ROCm) *                                                               |
 | `A2`               | Tests that require A2 NPU *                                                                             |
 | `A3`               | Tests that require A3 NPU *                                                                             |
-| `cards_{n}`        | Tests that require *n* accelerator cards (`cards_1` … `cards_8`); auto-added from `num_cards` *         |
+| `distributed_cuda` | Tests that require multi cards on CUDA platform *                                                       |
+| `distributed_rocm` | Tests that require multi cards on ROCm platform  *                                                      |
+| `distributed_npu`  | Tests that require multi cards on NPU platform  *                                                       |
+| `skipif_cuda`      | Skip if the num of CUDA cards is less than the required *                                               |
+| `skipif_rocm`      | Skip if the num of ROCm cards is less than the required *                                               |
+| `skipif_npu`       | Skip if the num of NPU cards is less than the required *                                                |
 | `slow`             | Slow tests (may skip in quick CI)                                                                       |
 | `benchmark`        | Benchmark tests (decorator on runner test functions; perf JSON uses `full_model` + type marker instead) |
 | `local_model`      | Tests requiring local / non-HF-hub model weights                                                        |

-\* Means those markers are auto-added by `@hardware_test` (parametrization decorator) or `hardware_marks` (only returning the list of marks for flexibility). Do **not** write `pytest.mark.H100` / `pytest.mark.L4` (or other SKU names) by hand: `check-mark` rejects that so `cards_{n}` cannot be skipped.
+\* Means those markers are auto-added by `@hardware_test` (parametrization decorator) or `hardware_marks` (only returning the list of marks for flexibility).

 #### Example usage for markers

@@ -76,23 +81,25 @@ def test_video_to_audio()
 This decorator is intended to make hardware-aware, cross-platform test authoring easier and more robust for CI/CD environments. The `hardware_test` decorator in `vllm-omni/tests/helpers/mark.py` performs the following actions:

 1. **Applies platform and resource markers**  
-   Adds the appropriate pytest markers for each specified hardware platform (e.g., `cuda`, `rocm`, `xpu`, `npu`) and resource type (e.g., `L4`, `H100`, `H200`, `B200`, `MI325`, `B60`, `A2`, `A3`).
+   Adds the appropriate pytest markers for each specified hardware platform (e.g., `cuda`, `rocm`, `xpu`, `npu`) and resource type (e.g., `L4`, `H100`, `MI325`, `B60`, `A2`, `A3`).
   ```python
   @pytest.mark.cuda
   @pytest.mark.L4
   ```
-2. **Handles card-count filtering**  
-   Adds a filterable marker `cards_{n}` (including `cards_1` when `num_cards` is omitted / 1). Use this with the SKU and platform, e.g. `-m "H100 and cards_4 and cuda"` vs `-m "H100 and cards_2 and cuda"`. Any multi-card CUDA test is `-m "cuda and not cards_1"`.
-
-   When `res` lists more than one platform **and** `num_cards` differs by platform, `@hardware_test` collects **one variant per platform** so those `cards_{n}` marks do not stack. `-m "H100 and cards_4 and cuda"` then selects only the CUDA/H100 4-card variant, not a ROCm 2-card sibling. The same count on every platform stays a single item.
+2. **Handles multi-card (distributed) scenarios**  
+   For tests requiring multiple cards, it automatically adds distributed markers such as `distributed_cuda`, `distributed_rocm`, or `distributed_npu`.
+   ```python
+   @pytest.mark.distributed_cuda(num_cards=num_cards)
+   ```
 3. **Supports flexible card requirements**  
   Accepts `num_cards` as either a single integer for all platforms or as a dictionary with per-platform values. If not specified, defaults to 1 card per platform.

 4. **Integrates resource validation**  
-   On CUDA and XPU, when `num_cards > 1`, attaches pytest's built-in `skipif` if the machine has fewer devices than requested. This is not a named `skipif_*` marker and is not used for `-m` filtering.
+   On CUDA, adds a skip marker (`skipif_cuda`) if the system does not have the required number of devices.
+   Support for `skipif_rocm` and `skipif_npu` will be implemented later.

 5. **Works with pytest filtering**  
-   Allows tests to be filtered and selected at runtime using standard pytest marker expressions (e.g., `-m "H100 and cards_2"`, `-m "L4 and not cards_1"`).
+   Allows tests to be filtered and selected at runtime using standard pytest marker expressions (e.g., `-m "distributed_cuda and L4"`).

 ##### Example usage for decorator

@@ -110,14 +117,13 @@ This decorator is intended to make hardware-aware, cross-platform test authoring
        num_cards=2,
    )
    ```
- `res` must be a dict. SKU values come from `pyproject.toml` markers tagged `[hardware-resource]` plus the platform key (`[cuda]`, `[rocm]`, `[xpu]`, `[npu]`, `[musa]`).
+- `res` must be a dict; supported resources: CUDA (L4/H100), ROCm (MI325), XPU (B60), MUSA (S5000), NPU (A2/A3)
 - `num_cards` can be int (all platforms) or dict (per platform); defaults to 1 when missing
+- Distributed markers (`distributed_cuda`, `distributed_rocm`, `distributed_npu`) are auto-added for multi-card cases
 - Filtering examples:
-    - CUDA H100 4-card: `pytest -m "H100 and cards_4 and cuda"`
-    - CUDA H100 2-card: `pytest -m "H100 and cards_2 and cuda"`
-    - CUDA multi-card (any count): `pytest -m "L4 and not cards_1"`
-    - ROCm multi-card: `pytest -m "MI325 and not cards_1"`
-    - NPU multi-card: `pytest -m "npu and not cards_1"`
+    - CUDA only: `pytest -m "distributed_cuda and L4"`
+    - ROCm only: `pytest -m "distributed_rocm and MI325"`
+    - NPU only: `pytest -m "distributed_npu"`

 #### Function: `hardware_marks`

@@ -173,30 +179,37 @@ If you want to add support for a new platform (e.g., "tpu" for a new accelerator
   [tool.pytest.ini_options]
   markers = [
       # ... existing markers ...
-       "tpu: [hardware-platform] Tests that require TPU device",
-       # Add [gpu] next to [hardware-platform] if this platform should also get pytest.mark.gpu.
-       "TPU_V3: [hardware-resource] [tpu] Tests that require TPU v3 hardware",
-       "cards_1: [hardware-cards] Tests that require 1 accelerator card",
-       "cards_2: [hardware-cards] Tests that require 2 accelerator cards",
+       "tpu: Tests that require TPU device",
+       "TPU_V3: Tests that require TPU v3 hardware",
+       "distributed_tpu: Tests that require multiple TPU devices",
   ]
   ```
-   `check-mark` treats `[hardware-platform]` names (including `cpu`/`gpu`) as allowed `pytest.mark.<name>` values. `hardware_test(res=...)` still rejects `cpu`/`gpu` as `res` keys.
 2. **Implement a marker construction function for your platform** in `vllm-omni/tests/helpers/mark.py`:
   ```python
   # In vllm-omni/tests/helpers/mark.py

-   def _tpu_marks(*, res: str, num_cards: int):
-       _require_sku("tpu", res)
-       return [pytest.mark.tpu, getattr(pytest.mark, res), _cards_mark(num_cards)]
+   def tpu_marks(*, res: str, num_cards: int):
+       test_platform = pytest.mark.tpu
+       if res == "TPU_V3":
+           test_resource = pytest.mark.TPU_V3
+       else:
+           raise ValueError(
+               f"Invalid TPU resource type: {res}. Supported: TPU_V3")
+
+       if num_cards == 1:
+           return [test_platform, test_resource]
+       else:
+           test_distributed = pytest.mark.distributed_tpu(num_cards=num_cards)
+           # Optionally: add skipif_tpu when implemented
+           return [test_platform, test_resource, test_distributed]
   ```
-3. **Register the builder** in `_marks_for_platform`:
+3. **Update `hardware_marks` to recognize your new platform**:
+    In the relevant place (see the `hardware_marks` implementation), add:
    ```python
-    builders = {
-        # ... existing platforms ...
-        "tpu": _tpu_marks,
-    }
+    if platform == "tpu":
+        marks = tpu_marks(res=resource, num_cards=cards)
    ```
-    (`hardware_test` / `hardware_marks` both go through `_marks_for_platform`.)
+    (`hardware_test` calls `hardware_marks` internally, so both will pick up the change.)
 4. **(Recommended) Add a test using your new markers**:
   ```python
   @hardware_test(
@@ -210,11 +223,11 @@ If you want to add support for a new platform (e.g., "tpu" for a new accelerator
 **Summary**:  

 - Add pytest markers for your new platform/resources  
- Implement a private marker function (`_xxx_marks`)
- Register it in `_marks_for_platform`
- You're done: tests using `@hardware_test` or `hardware_marks` with your platform now automatically get the correct markers and `cards_{n}` filtering.
+- Implement a marker function (`xxx_marks`)  
+- Plug into `hardware_marks`  
+- You're done: tests using `@hardware_test` or `hardware_marks` with your platform now automatically get the correct markers, distribution, and isolation!

-See code in `vllm-omni/tests/helpers/mark.py` for existing examples (`_cuda_marks`, `_rocm_marks`, `_npu_marks`).
+See code in `vllm-omni/tests/helpers/mark.py` for existing examples (`cuda_marks`, `rocm_marks`, `npu_marks`).

 ## Test case style

@@ -438,7 +438,7 @@ selects the scheduler and worker family:
 Key topology fields belong to `StagePipelineConfig`:

 | Field | Description |
-| ------- | ------------- |
+|-------|-------------|
 | `model_stage` | Logical stage name (`ar_stage`, `decoder`, etc.). |
 | `execution_type` | Runtime family for this stage. |
 | `input_sources` | Upstream stage IDs that provide input. |
@@ -960,7 +960,7 @@ Common failures and fixes:
 | `check-forbidden-imports` | Stdlib `re`/`base64`, pickle, Hugging Face Hub API, or direct Triton/TileLang | `import regex as re` and `pybase64`; do not grow the allowlist without review |
 | `check-torch-cuda-call` | New `torch.cuda.*` call site | Use `current_omni_platform`; do not grow `ALLOWED_FILES` without review |
 | `check-tts-adapter-migration` | New `_tts_model_type` branch in `serving_speech.py` | Put logic in `tts_adapters/`; lower `MAX_MODEL_TYPE_BRANCHES` when removing branches |
-| `check-mark` | Test file missing level/hardware mark, or direct `pytest.mark.H100` | Add `core_model`/… plus `cpu`/`cuda`/`hardware_test(`; SKU marks only via helpers |
+| `check-test-ci-coverage` | Test file missing level or hardware mark | Add `core_model`/… plus `cpu`/`cuda`/`hardware_test(` |
 | `shellcheck` | Native `shellcheck` missing, or script warning | Install via apt/dnf/brew (or `shellcheck.exe` on Windows). See [Linting](../README.md#linting) |
 | `mypy-3.10` / `markdownlint-cli2` / `check-buildkite` | Types, docs markdown, or Buildkite YAML | See [Linting](../README.md#linting) |

@@ -170,10 +170,6 @@ because the user-visible output is diffusion media rather than a token stream.
 The
 [MiniMax-H3 recipe](https://github.com/vllm-project/vllm-omni/blob/main/recipes/MiniMaxAI/MiniMax-H3.md)
 contains the hardware-specific profiles and warmup requirements.
-The internal, opt-in
-[temporal VAE chunk callback](feature/minimax_h3_temporal_vae_chunks.md)
-documents the narrower released-source contract for publishing committed video
-frames during decode. It does not change the complete-output serving topology.

 ### Cosmos3: unified MoT reasoner and generator

@@ -1,260 +0,0 @@
-# MiniMax H3 temporal VAE chunk callback
-
-!!! warning "Experimental internal capability"
-
-    This feature is an opt-in model integration seam. The current MiniMax H3
-    pipeline still uses complete-output VAE decode, and no CLI option, serving
-    path, transport, or MP4 sink enables temporal chunks by default.
-
-## Status and scope
-
-MiniMax H3 decodes video latents in temporal windows, but the released VAE
-normally exposes only the complete reconstructed video. The temporal callback
-publishes frames after the released decoder has completed overlap blending and
-padding removal. This gives a future runtime consumer a stable point at which
-it can schedule representation conversion, D2H transfer, and encoding while
-the VAE continues decoding later windows.
-
-The implementation remains model-local:
-
- `vae.py` owns VAE topology and the thin
-  `decode_latent_with_chunks()` facade;
- `chunked_decode.py` owns the normalized callback contract, metadata
-  validation, and fixed VAE-group rendezvous points; and
- `temporal_chunks.py` owns the source-gated compatibility planner and the
-  released temporal assembly control flow.
-
-Crop policy, typed media, transport, backpressure, encoding, and public output
-remain runtime responsibilities. The model code does not schedule requests or
-select a presentation format.
-
-## Execution flow
-
-The default and opt-in paths are intentionally separate:
-
-```text
-Default:
-latent -> MiniMaxH3VideoVAE.decode_latent() -> complete RGB video
-
-Opt-in producer seam:
-latent
-  -> MiniMaxH3VideoVAE.decode_latent_with_chunks()
-  -> group-local preflight and callback-owner validation
-  -> released temporal decode / overlap blending / padding removal
-  -> borrowed decoder-domain chunk
-  -> normalize + clone into owned RGB float32 BCTHW [0, 1]
-  -> synchronous callback
-  -> complete RGB video is still returned
-```
-
-Calling `decode_latent_with_chunks()` is the opt-in. Supplying no callback on a
-VAE peer does not select the default decoder; every rank in a VAE patch group
-must enter the chunked method, and only the output rank receives the callback.
-
-## Producer contract
-
-The normalized callback has this structural signature:
-
-```python
-def callback(
-    frames: torch.Tensor,
-    /,
-    *,
-    chunk_index: int,
-    total_chunks: int,
-    frame_start: int,
-    is_final: bool,
-) -> None:
-    ...
-```
-
-For a successful decode:
-
- `frames` is owned, contiguous RGB float32 in `BCTHW` layout and `[0, 1]`;
- callback storage does not alias the returned complete video;
- the consumer may retain or mutate a chunk;
- `chunk_index` is zero-based, contiguous, and strictly increasing;
- `total_chunks` is positive and stable for the decode;
- `frame_start` equals the end of the previous published range;
- chunks are nonempty and contain only temporally committed frames;
- exactly one chunk has `is_final=True`; and
- concatenating chunks along time exactly reconstructs the returned video.
-
-The callback runs synchronously on the producer's current device stream. A
-consumer that reads the tensor on another stream must establish ordering, for
-example with an event, and retain the allocation for that stream's lifetime.
-A slow callback also blocks VAE production. An overlap-oriented runtime
-consumer should therefore acquire bounded capacity, enqueue side-stream work,
-and return promptly instead of encoding frames inside the callback.
-
-Chunks retain the decoded VAE canvas dimensions. A future runtime integration
-must apply the same requested-height and requested-width crop as the complete
-MiniMax H3 pipeline before declaring the media transport-ready.
-
-The raw compatibility callback is deliberately narrower. It receives borrowed
-decoder-domain tensors and must not retain or mutate them. The coordinator is
-the ownership boundary that normalizes and clones raw chunks before invoking
-the public callback.
-
-## Temporal planning and terminal publication
-
-The released configuration advances five latent tokens per temporal window and
-includes a two-token overlap, so each `_adaptive_decode` input normally contains
-seven tokens. After temporal scaling, pre-padding removal, and five-frame
-overlap blending, the normal committed unit is 17 output frames. The final
-remainder depends on the requested frame plan.
-
-Representative released frame plans include:
-
-| Output frames | Published chunks |
-| ---: | --- |
-| 124 | `7 x 17 + 5` |
-| 209 | `12 x 17 + 5` |
-| 243 | `14 x 17 + 5` |
-| 362 | `21 x 17 + 5` |
-
-The terminal callback is held until the complete frame plan passes its logical
-frame, padding, written-frame, chunk-count, and finality checks. Earlier chunks
-are semantically stable, but they do not imply that the overall request has
-succeeded; a consumer must discard or abort partial output if a later decoder
-failure terminates the request.
-
-## Source and configuration gate
-
-The released VAE has no public temporal-callback API. vLLM-Omni therefore owns
-a compatibility implementation for one exact released code contract. The
-reference revision identifies its provenance, but runtime acceptance is based
-on source content, configuration, and structure rather than repository or
-weight metadata:
-
- reference Hugging Face revision, not independently checked at runtime:
-  `42ed227ee7df40d41602854ae760620d6eb651fe`;
- sorted 15-file top-level `video_vae/*.py` manifest (excluding
-  `__init__.py`) SHA-256:
-  `30e64afbaa940696bf8bfe2c86003a4b1666c84b22fa3aa12403a3e6a03f705d`;
- clip length `17`;
- token chunk, overlap, and drop `5`, `2`, and `3`;
- temporal ratio `4`;
- frame pre-padding and overlap `3` and `5`; and
- no isolated first or last frame.
-
-The adapter does not verify Hugging Face revision metadata, repository identity,
-or model weights. The manifest pin detects Python compatibility drift; it is
-not a security boundary for remote-code execution. The source manifest, listed
-temporal attributes, structural contract, plan, and dtype are validated before
-the first `_adaptive_decode` call. A mismatch disables only the opt-in chunk path.
-`decode_latent()` remains available and performs no source fingerprint or chunk
-rendezvous.
-
-## Supported models
-
-Support in this table means the internal producer contract above, not public
-serving or end-to-end transport support.
-
-| Model or component | Source contract | Status | Validation |
-| --- | --- | --- | --- |
-| MiniMax H3 FL2VA video VAE | Released manifest/listed temporal attributes above (reference `42ed227e`) | Supported, opt-in | Same-environment full-output parity and L20X/B300 feasibility measured on the implementation through PR commit `60e9bbad` |
-| MiniMax H3 Ref2VA video VAE | Same released manifest/listed temporal attributes | Source-compatible; not separately benchmarked | Ref2VA and FL2VA use byte-identical video-VAE source and configuration in the reference snapshot |
-| MiniMax H3 with a different Python manifest or listed temporal attribute | Different enforced contract | Unsupported for chunk mode | Fails closed before decoder collectives; complete decode remains available |
-| Same manifest/temporal attributes with different weights or other component metadata | Not checked by this gate | Not independently qualified | The adapter does not inspect weights, repository identity, revision metadata, or arbitrary component configuration |
-| Other video models | Any | Unsupported | No generic temporal VAE producer ABI is declared by this feature |
-
-## Feature compatibility
-
-| Dimension or combination | Status | Constraint |
-| --- | --- | --- |
-| Existing complete-output decode | Supported and default | No source check, callback object, or callback rendezvous |
-| Explicit chunk decode | Supported internally | Caller invokes `decode_latent_with_chunks()` directly |
-| VAE patch parallel size 1 | Supported by the model seam | Caller supplies the callback only on the request output owner; no WORLD rendezvous is added |
-| VAE patch parallel size equal to the full DiT group | Supported when data parallel size is 1 | Every VAE-group rank calls; exactly group rank 0 owns the callback |
-| Partial VAE patch subgroup | Unsupported | MiniMax H3 currently accepts size 1 or the full DiT group only |
-| Data parallel size greater than 1 with VAE patch parallel size 1 | Model seam only | Per-replica callback ownership is possible, but no serving/runtime consumer is wired |
-| Data parallel size greater than 1 with VAE patch parallel size greater than 1 | Unsupported | Fails before decoder collectives because no replica-local VAE patch group is defined |
-| Too few spatial tiles for the VAE group | Correctness fallback | Decodes rank-locally and restores native tiling state; slower than patch parallel decode |
-| Device memory and full-output assembly | Not reduced or bounded | The compatibility path still assembles and returns the full decode while allocating owned callback chunks; no peak-HBM reduction is claimed |
-| DiT cache, attention, TP, SP, HSDP, LoRA, or quantization features | No direct callback dependency | This seam runs after denoising and does not independently qualify combinations beyond the MiniMax H3 model's own matrix |
-| Layerwise or distributed component offload | Not separately validated | Component residency must place the video VAE before decode |
-| Step execution or request batching | Not integrated | No pipeline caller binds a runtime chunk consumer |
-| Requested-size crop and device uint8 preparation | Not implemented here | Runtime-owned follow-up |
-| Pinned D2H, SHM/ZMQ, or cross-process transport | Not implemented here | Runtime-owned follow-up |
-| Complete MP4 or fragmented MP4 sink | Not implemented here | Existing serving behavior still waits for complete output |
-| Audio VAE chunks and incremental A/V composition | Not supported | Video chunks do not imply audio or request finality |
-| Cancellation and multi-output lease ownership | Not implemented here | Requires a bounded runtime lifecycle |
-| NVIDIA CUDA | Feasibility evidence | L20X and B300 measurements apply to the implementation through `60e9bbad`; the current coordination refactor was not reprofiled |
-| AMD ROCm | Not validated for this feature | Base MiniMax H3 support is documented separately and does not imply temporal-chunk validation |
-| Ascend, Intel, or other accelerators | No feature support claim | Consult the model support table independently; this design adds no platform qualification |
-
-## Distributed and failure semantics
-
-When VAE patch parallelism is active, all VAE-group ranks run the same
-preflight before entering checkpoint collectives. Source/planner errors and the
-resolved temporal concat dtype are reconciled at fixed group rendezvous points.
-
-Callback ownership is also a group decision. Rank `r` contributes `r + 1` when
-it owns a callback and zero otherwise. A group SUM equals one only when rank 0
-is the sole owner; every invalid owner configuration therefore fails
-identically on all ranks before decode.
-
-Callback exceptions are retained on the output rank. Further publication stops,
-but every rank completes the same remaining native decoder collective schedule.
-The original exception is then raised on the owner and
-`MiniMaxH3ChunkCallbackPeerError` on peers.
-
-Exceptions raised inside the checkpoint's native distributed decode cannot in
-general be reconciled by the adapter: a peer may already be blocked inside a
-checkpoint-owned collective. Resolving such failures requires native collective
-handling or supervisor-level timeout and worker-group teardown. The process
-group must not be assumed reusable after such a failure.
-
-## Validation evidence
-
-The contract suite covers ordinary and padded frame plans, blend boundaries,
-owned storage and mutation isolation, source-manifest/listed-temporal-attribute
-drift, callback failure, and healthy subsequent requests. Real Gloo processes cover
-rank-asymmetric preflight failures, dtype disagreement, two-rank callback
-failure propagation, three-rank wrong-owner rejection, insufficient-tile
-fallback, and group recovery for failures that occur at adapter-owned
-rendezvous points.
-
-Four-device L20X and independent four- and eight-device B300 experiments on the
-implementation through PR commit `60e9bbad` showed exact serial-versus-chunked
-frame/MP4 parity within each environment and lower VAE-to-complete-MP4 latency
-when an experimental consumer was attached. Hashes are not claimed to match
-across different hardware/software environments. The later coordination and
-module-boundary refactors were revalidated with CPU and real Gloo tests but were
-not reprofiled on GPUs. These are feasibility measurements, not a promise for
-default serving, because this change does not contain the transport or encoder
-consumer. See [RFC #6872](https://github.com/vllm-project/vllm-omni/issues/6872)
-and [PR #6885](https://github.com/vllm-project/vllm-omni/pull/6885) for the
-measurement provenance and limitations.
-
-## Generalization boundary and follow-ups
-
-This feature does not define a repository-wide `SupportsVideoChunkDecode`
-interface. H3 temporal blending, Wan causal caches, LTX temporal/spatial halos,
-and pipeline-level generation chunks have different commit and failure
-boundaries. A generic producer contract should be extracted only after a second
-model proves the same semantic event without `Any` arguments or model-name
-dispatch.
-
-The reusable future boundary is the emitted, committed video-media chunk, not a
-universal latent-to-VAE method. It should extend the typed media work in
-[RFC #6541](https://github.com/vllm-project/vllm-omni/issues/6541) and
-[PR #6615](https://github.com/vllm-project/vllm-omni/pull/6615), while keeping
-model-specific halo, cache, padding, blending, distributed, and source-gating
-logic inside each model adapter.
-
-Runtime follow-ups own:
-
- typed video-plus-audio media and request/output identity;
- requested-size crop and device preparation;
- bounded leases, backpressure, cancellation, and cleanup;
- D2H/IPC transport and observability; and
- complete-MP4 and fragmented-MP4 consumers.
-
-## Related documentation
-
- [Diffusion model integration](../module/diffusion/diffusion_model_integration.md)
- [VAE patch parallelism](vae_parallel.md)
- [Async diffusion output](async_diffusion_output.md)
- [MiniMax H3 recipe](https://github.com/vllm-project/vllm-omni/blob/main/recipes/MiniMaxAI/MiniMax-H3.md)
@@ -26,8 +26,6 @@ Legend: ✅ supported, ⚠️ compatibility path or limited validation, ❌ unsu
 | **TP > 1** | ⚠️ Ordinary TP-aware loader only; no direct checkpoint mmap. | ⚠️ Ordinary TP-aware loader only; no direct checkpoint mmap. |
 | **HSDP** | ❌ Rejected to avoid double-sharding parameters. | ⚠️ Limited end-to-end coverage. |
 | **Per-tensor online FP8 linears** | ✅ Ordinary loader finalizes weights and scales before DLO sharding. | ✅ Ordinary loader retains complete rank-local tensors. |
-| **Online INT8 linears** | ✅ Ordinary loader finalizes the int8 weight (contiguous or transposed view) and fp32 scale before DLO sharding. | ✅ Ordinary loader retains complete rank-local tensors. |
-| **Online MXFP8 linears** | ✅ Ordinary loader finalizes the fp8 weight (contiguous) and e8m0 block scale (contiguous or transposed view) before DLO sharding. | ✅ Ordinary loader retains complete rank-local tensors. |
 | **Other online quantization methods** | ❌ Rejected until runtime packing and scale layouts are validated. | ⚠️ Allowed through the ordinary loader; validation is method-specific. |
 | **Model-level or standard layerwise CPU offload** | ❌ Disabled because DLO takes priority. | ❌ Disabled because DLO takes priority. |
 | **Resident leading layers** | ❌ Rejected. | ✅ Requires eligible resident paths in the model's `OffloadPlan`. |
@@ -440,15 +438,11 @@ requirements.

 Direct checkpoint mmap can back either transfer path. It is currently limited
 to proven TP1, non-HSDP, non-online-quantized layouts. Other layouts use the
-ordinary loader. Per-tensor online FP8, online INT8, and online MXFP8 linears
-can use DLO AllGather after the ordinary loader finalizes their runtime
-weights and scales; DLO then shards and reconstructs those tensors with their
-recorded layouts (they all keep plain 1-byte dtypes over ordinary strided
-views: online INT8 has an int8 weight plus fp32 scale, online MXFP8 has an
-fp8 weight plus e8m0 block scale, each with a contiguous or transposed-view
-finalized tensor, all covered by the same physical-order packing as online
-FP8). Other online methods must use `--dlo-no-use-allgather` or disable
-online quantization until their runtime layouts are validated.
+ordinary loader. Per-tensor online FP8 linears can use DLO AllGather after the
+ordinary loader finalizes their runtime weights and scales; DLO then shards and
+reconstructs those tensors with their recorded layouts. Other online methods
+must use `--dlo-no-use-allgather` or disable online quantization until their
+runtime layouts are validated.

 The Host Weight Runtime representation, publication, and no-AllGather consumer
 contracts are merged; see
@@ -465,8 +459,8 @@ Current source-level validation includes:
 - HSDP + DLO without AllGather acceptance at configuration level;
 - loader preflight fallback for TP, HSDP, online quantization, unknown custom
  loaders, missing keys, and shape/dtype mismatches;
- ordinary-loader fallback for per-tensor online FP8, online INT8, and online
-  MXFP8 linears followed by DLO sharding of finalized weights and scales;
+- ordinary-loader fallback for per-tensor online FP8 linears followed by DLO
+  sharding of finalized weights and scales;
 - exact loader-to-backend plan transfer and ordinary-loader fallback;
 - ordered publication hashing, overlapped durability, unordered parallel
  checksum fallback, and node-local source-digest reuse/invalidation;
@@ -22,60 +22,59 @@ We use **Qwen-Image** as the reference implementation for decode parallel, and *

 **VAE Patch Parallelism** is an acceleration technique for both **encoding** and **decoding**. Instead of processing the entire tensor at once, the tensor is:

- Split into multiple spatial tiles
+ Split into multiple spatial tiles

- Distributed across multiple ranks
+ Distributed across multiple ranks

- Encoded/Decoded in parallel
+ Encoded/Decoded in parallel

- Merged to reconstruct the final output
+ Merged to reconstruct the final output

 This approach:

- Distributes computation across multiple devices
+ Distributes computation across multiple devices

- Reduces peak memory usage per device
+ Reduces peak memory usage per device

- Accelerates encoding/decoding latency
+ Accelerates encoding/decoding latency

 ### When to Use Encode vs Decode Parallel

-| Operation           | Use Case                     | Example                           |
-| ------------------- | ---------------------------- | --------------------------------- |
-| **Decode Parallel** | Text-to-Image, Text-to-Video | Latent → Image/Video              |
-| **Encode Parallel** | Image-to-Video (I2V)         | Image → Latent (for conditioning) |
+| Operation | Use Case | Example |
+|-----------|----------|---------|
+| **Decode Parallel** | Text-to-Image, Text-to-Video | Latent → Image/Video |
+| **Encode Parallel** | Image-to-Video (I2V) | Image → Latent (for conditioning) |

 ### Architecture
-
 We introduce **DistributedVaeExecutor** as the core component responsible for distributed VAE encoding/decoding.

 The executor is model-agnostic and accepts three function parameters:

- split – Partition the latent into tiles
+ split – Partition the latent into tiles

- exec – Decode a single tile
+ exec – Decode a single tile

- merge – Combine decoded tiles into the final output
+ merge – Combine decoded tiles into the final output

 #### Execution Flow

- Call split(z) to generate a list of TileTask and a GridSpec
+ Call split(z) to generate a list of TileTask and a GridSpec

- Dispatch tasks across ranks using workload-based balancing
+ Dispatch tasks across ranks using workload-based balancing

- Each rank executes exec(task) on its assigned tiles
+ Each rank executes exec(task) on its assigned tiles

- Gather decoded tile results to rank 0
+ Gather decoded tile results to rank 0

- Rank 0 performs merge(...)
+ Rank 0 performs merge(...)

- (Optional) Broadcast final result to all ranks
+ (Optional) Broadcast final result to all ranks

 This design separates:

- Distributed execution logic
+ Distributed execution logic

- Model-specific tiling and merging logic
+ Model-specific tiling and merging logic

 #### Why split / exec / merge is necessary?

@@ -83,23 +82,23 @@ The latent tensor cannot be arbitrarily partitioned.

 During decoding:

- Each output pixel may depend on neighboring pixels
+ Each output pixel may depend on neighboring pixels

- The receptive field is model-dependent
+ The receptive field is model-dependent

 Therefore:

- Tiles must include overlap
+ Tiles must include overlap

- Merge must perform blending to avoid seams
+ Merge must perform blending to avoid seams

 ## Step-by-Step Implementation (Decode)

 ### Step 1: Implement DistributedAutoencoderKLQwenImage
-
 `QwenImagePipeline` use `AutoencoderKLQwenImage` for vae, so implement a distributed version:

-```python
+
+```
 class DistributedAutoencoderKLQwenImage(AutoencoderKLQwenImage, DistributedVaeMixin):
    @classmethod
    def from_pretrained(cls, *args: Any, **kwargs: Any):
@@ -107,35 +106,28 @@ class DistributedAutoencoderKLQwenImage(AutoencoderKLQwenImage, DistributedVaeMi
        model.init_distributed()
        return model
 ```
-
 **Key points**:
-
- Inherit both AutoencoderKLQwenImage and DistributedVaeMixin
- Call init_distributed() after loading weights
+ Inherit both AutoencoderKLQwenImage and DistributedVaeMixin
+ Call init_distributed() after loading weights

 ### Step 2: Implement split/exec/merge
-
 Reuse `AutoencoderKLQwenImage.tiled_decode` logic and divide it into three stages. And we need return tiles with `GridSpec` and `TileTask`:
-
-```python
+```
 class GridSpec:
    split_dims: tuple[int, ...]  # Tensor dimensions being split (e.g., (2, 3) for (B, C, H, W))
    grid_shape: tuple[int, ...]  # Tile grid layout (num_rows, num_cols)
    tile_spec: dict = field(default_factory=dict) # Metadata required for merging
    output_dtype: torch.dtype | None = None # Final output dtype
 ```
-
-```python
+```
 class TileTask:
    tile_id: int # task id
    grid_coord: tuple[int, ...]  # Tile position in grid
    tensor: torch.Tensor | list[torch.Tensor]  # The tile tensor
    workload: int | float = 1 # Used for load balancing (e.g., tile area)
 ```
-
 And tiled base split/exec/merge as follow:
-
-```python
+```
 def tile_split(self, z: torch.Tensor) -> tuple[list[TileTask], GridSpec]:
    # mostly copy from AutoencoderKL
    _, _, num_frames, height, width = z.shape
@@ -218,14 +210,11 @@ def tile_merge(self, coord_tensor_map: dict[tuple[int, ...], torch.Tensor], grid
 ```

 ### Step 3: Override tiled_decode
-
 We need to override tiled_decode, the main logic is:
-
- check distributed is enabled
- select split/exec/merge
- Invoke self.distributed_executor.execute to decode
-
-```python
+ check distributed is enabled
+ select split/exec/merge
+ Invoke self.distributed_executor.execute to decode
+```
 def tiled_decode(self, z: torch.Tensor, return_dict: bool = True):
    if not self.is_distributed_enabled():
        return super().tiled_decode(z, return_dict=return_dict)
@@ -241,14 +230,11 @@ def tiled_decode(self, z: torch.Tensor, return_dict: bool = True):

    return DecoderOutput(sample=result)
 ```
-
 `broadcast_result` is set to True or False depending on the model; when enabled, the result will be used even on ranks other than 0.

 ### Step 4: Modify Pipeline
-
 Change vae model from AutoencoderKLQwenImage to DistributedAutoencoderKLQwenImage
-
-```diff
+```
 class YourModelPipeline(nn.Module):
    def __init__(
        self,
@@ -273,8 +259,8 @@ For models that require VAE encoding (e.g., Image-to-Video), you can also parall

 Similar to decode, split the input tensor into tiles. Key considerations:

- **Patchify handling**: If the model uses `patch_size`, scale tile parameters accordingly
- **Temporal chunking**: Video VAEs may have temporal compression (e.g., 4x)
+ **Patchify handling**: If the model uses `patch_size`, scale tile parameters accordingly
+ **Temporal chunking**: Video VAEs may have temporal compression (e.g., 4x)

 ```python
 def encode_tile_split(self, x: torch.Tensor) -> tuple[list[TileTask], GridSpec]:
@@ -420,28 +406,23 @@ def tiled_encode(self, x: torch.Tensor) -> torch.Tensor:
 **Key differences from decode parallel:**

 | Aspect | Decode Parallel | Encode Parallel |
-| -------- | ----------------- | ----------------- |
+|--------|-----------------|-----------------|
 | `broadcast_result` | Often `False` (only rank 0 needs output) | `True` (all ranks need latents for diffusion) |
 | Patchify | Applied in merge (unpatchify) | Handled by parent `_encode()` before `tiled_encode()` |
 | Temporal chunking | Frame-by-frame | Chunk-based (e.g., 1 + 4n frames) |

 ## Testing
-
 Verify numerical consistency between:
+ vae_patch_parallel_size = 1

- vae_patch_parallel_size = 1
- vae_patch_parallel_size = N
+ vae_patch_parallel_size = N

 Example:
-
-```python
 torch.allclose(output_1, output_n, atol=1e-5)
-```

 Testing requirements:
-
- Fix random seed
- Use identical tiling strategy
+ Fix random seed
+ Use identical tiling strategy

 ```python
 m = Omni(
@@ -453,7 +434,6 @@ m = Omni(
        ),
    )
 ```
-
 When vae_patch_parallel_size is larger than the DiT world size, it will automatically fall back to using the DiT world size instead.

 ## Reference Implementations
@@ -461,7 +441,7 @@ When vae_patch_parallel_size is larger than the DiT world size, it will automati
 Complete examples in the codebase:

 | Model | Path | Decode Parallel | Encode Parallel |
-| ------- | ------ | ----------------- | ----------------- |
+|-------|------|-----------------|-----------------|
 | **Z-Image** | `vllm_omni/diffusion/distributed/autoencoders/autoencoder_kl.py` | ✅ | ❌ |
 | **Wan2.2** | `vllm_omni/diffusion/distributed/autoencoders/autoencoder_kl_wan.py` | ✅ | ✅ |
 | **Qwen-Image** | `vllm_omni/diffusion/distributed/autoencoders/autoencoder_kl_qwenimage.py` | ✅ | ❌ |
@@ -476,7 +456,7 @@ The tile-parallel executor above assigns independent spatial **tiles** to ranks.
 ### How it differs from tile parallel

 | Aspect | Tile parallel (`"tile"`) | Spatially-sharded (`"spatial_shard_height"`/`"spatial_shard_width"`) |
-| -------- | -------------------------- | ----------------------------------------------- |
+|--------|--------------------------|-----------------------------------------------|
 | Unit of work | Independent overlapping tiles | A single global feature map sharded along H or W |
 | Cross-rank communication | Gather tiles to rank 0, stitch + blend | Per-conv **halo exchange** of boundary rows/cols (P2P) |
 | Output assembly | Blend overlapping tiles | All-gather shards on rank 0, trim padding (matches `broadcast_result=False`) |
@@ -501,7 +481,7 @@ serve.py (--vae-parallel-mode) / OmniEngineArgs
 ### Notes

 - The decoder is patched **in place** the first time spatial-shard decode runs and is bound to a single split dimension for the lifetime of the VAE instance.
- Numerical correctness vs. single-GPU decode is covered by `tests/diffusion/distributed/test_wan_spatial_shard.py::test_spatial_shard_decode_matches_reference` (multi-GPU, nightly `full_model` + `cards_2`).
+- Numerical correctness vs. single-GPU decode is covered by `tests/diffusion/distributed/test_wan_spatial_shard.py::test_spatial_shard_decode_matches_reference` (multi-GPU, nightly `full_model` + `distributed_cuda`).

 ---

@@ -25,10 +25,6 @@ implementation contract; it is not, by itself, a general support claim.
 - [Automatic Prefix Caching in Omni Models](feature/prefix_caching.md)
 - [Realtime AR-Diffusion Sessions](feature/realtime_ar_diffusion.md)

-### Model-specific capabilities
-
- [MiniMax H3 Temporal VAE Chunks](feature/minimax_h3_temporal_vae_chunks.md)
-
 ### Communication

 #### OmniConnector implementations
@@ -57,10 +57,3 @@ cross-stage routing.

 Test registry selection, checkpoint loading, minimal inference, input and output
 contracts, and every declared optional capability.
-
-## Model-specific capability documents
-
- [MiniMax H3 temporal VAE chunk callback](../../feature/minimax_h3_temporal_vae_chunks.md)
-  defines the source-gated producer seam and its boundary with future runtime
-  media transport. It is an internal opt-in capability, not a public serving
-  feature.
@@ -37,7 +37,6 @@ validation_paths:
  - tests/engine/test_async_omni_engine_input.py
  - tests/engine/test_async_omni_engine_outputs.py
  - tests/engine/test_async_omni_engine_abort.py
-  - tests/e2e/offline_inference/test_qwen3_omni.py
  - tests/engine/test_async_omni_engine_stage_init.py
  - tests/engine/test_orchestrator.py
  - tests/engine/test_orchestrator_error_handling.py
@@ -114,15 +113,6 @@ Numbers become append-only after normative promotion.

 Test routing, output ordering, queue correlation, cancellation, failure
 propagation, shutdown ordering, and representative multi-stage execution.
-AR abort of a final-stage LLM request must deliver a terminal output with
-`finish_reason="abort"` and the generated prefix so collocated training can
-resume; diffusion abort remains whole-sample retry. Aborting a parallel-sampling
-child must drop `parent_requests` once no children remain. EngineCore control
-RPCs (`sleep` / `wake_up` / `pause_scheduler` / `resume_scheduler`) must
-propagate worker exceptions rather than returning `{"supported": False}`.
-When a request has multiple final output stages, only the last abort message
-is request-terminal. Output-processor abort state is committed only after the
-physical EngineCore abort succeeds.
 Changes that cross into stage runtime or public error behavior require review
 from that contract's owners.

@@ -43,7 +43,6 @@ depends_on:
 validation_paths:
  - tests/entrypoints/test_omni_entrypoints.py
  - tests/entrypoints/test_async_omni.py
-  - tests/entrypoints/test_async_omni_pause_sleep_routing.py
  - tests/entrypoints/test_async_omni_duplex.py
  - tests/entrypoints/test_serve.py
  - tests/entrypoints/test_stream_finish_reason.py
@@ -120,13 +119,7 @@ promotion.

 Test request validation, protocol conversion, model-adapter routing,
 streaming, session identity, disconnect/cancellation, and error mapping for
-each affected offline or serving entrypoint. Sleep must wait for in-flight
-`generate()` admission before EngineCore offload; `wake_up` does not resume
-admission — callers must `resume_generation()`. Sleeping tags are tracked per
-stage so `wake_up(stage_ids=[0])` does not skip a later `wake_up(stage_ids=[1])`.
-Streaming input pumps take an admission slot immediately before each EngineCore
-ADD or update, not while waiting for the next client chunk. Frontend abort
-keeps `request_states` until `generate()` consumes the terminal output.
+each affected offline or serving entrypoint.

 ## Promotion gate

@@ -257,53 +257,6 @@ Lookup checks the marker both before validation and after constructing the
 lease so a weaker overlapping lookup cannot return a lease after a stronger
 validation has denied the artifact.

-If publishing the external deny marker fails, lookup returns a retryable
-storage failure and marks the artifact directory with a persistent invalidation
-mode before releasing the shared lock. Cooperating readers check that mode both
-while opening and immediately before returning a lease. Lookup then makes a
-bounded attempt to take the exclusive artifact lock and quarantine the entry;
-the quarantine transition removes the internal mode marker. If a cooperating
-builder replaces the marked inode while the locks are exchanged, containment
-preserves that new publication instead of quarantining it. If neither the
-external marker nor the directory-mode marker can be persisted, cross-process
-exclusion cannot be guaranteed on an unavailable filesystem and the domain
-must remain failed rather than treating the artifact as an ordinary invalid
-cache entry.
-
-New publications carry the internal invalidation mode through the staging
-rename. Hardening clears it and is the logical authority transition for a
-process crash; the temporary- and artifact-parent syncs then make the rename
-durable, with an artifact-inode sync first making the hardened mode durable. A
-process that dies before hardening leaves a persistently denied
-entry, while one that dies after successful hardening leaves an authoritative
-ready entry. Any synchronous failure after rename is marked invalid and
-contained in quarantine before `PUBLICATION_FAILED` is returned, so that
-nonretryable result cannot later resolve as a ready hit. If the filesystem
-refuses both the inode marker and the quarantine move, the store instead returns
-a retryable storage failure with `outcome_uncertain` details; as with deny-marker
-double failure, cross-process exclusion cannot be guaranteed until storage
-recovers. Once the artifact-parent sync succeeds, a later artifact- or
-build-lock release error is logged and lookup resolves the committed entry
-instead of reporting a contradictory publication failure. The same authority
-rule applies when a competing publisher already installed byte-identical
-semantic content; staging-cleanup errors are retried without downgrading that
-validated ready entry to publication failure.
-
-Before widening a hardened artifact for a lifecycle move, the store durably
-sets the same invalidation mode. A process crash before rename therefore leaves
-the writable source denied. After rename, the store re-hardens the destination
-but retains the marker until both parent-directory syncs make the move durable;
-only then does it durably clear the marker. An interrupted move therefore leaves
-a marked transition that the next exact-key lifecycle operation re-hardens.
-Explicit cleanup uses the locked move and retries exact-key
-`.cleanup.*` tombstones left by an interrupted removal; a later build performs
-the same tombstone reconciliation before producing replacement content.
-
-An operational failure while inspecting a noncooperative competing publication
-is a retryable storage failure. The competing entry remains authoritative and
-is not mislabeled as corrupt or quarantined merely because its manifest could
-not be read transiently.
-
 `inspect_domain()` and `inspect_artifact()` return typed, read-only snapshots.
 Inspection validation does not create a deny marker. Reported lock activity is
 point-in-time evidence only; kernel locks are not a persistent owner registry.
@@ -3,6 +3,4 @@
 vLLM-Omni's examples are split into two categories:

 - If you are using vLLM-Omni from within Python code, see the *Offline Inference* section.
- If you are using vLLM-Omni from an HTTP application or client, start with
-  the [API Server guide](../serving/README.md) to choose an endpoint, then use
-  the *Online Serving* section for model-specific examples.
+- If you are using vLLM-Omni from an HTTP application or client, see the *Online Serving* section.
@@ -34,7 +34,7 @@ vLLM-Omni is a Python library that supports the following GPU variants. The libr

 ### Pre-built wheels

-Note: Pre-built wheels are currently available for vLLM-Omni 0.11.0rc1, 0.12.0rc1, 0.14.0rc1, 0.14.0, 0.16.0, 0.18.0, 0.20.0, 0.21.0, 0.22.0, 0.23.0, 0.24.0, 0.25.0, 0.26.0, and 0.28.0. If you need a newer unreleased revision, please [build from source](https://docs.vllm.ai/projects/vllm-omni/en/latest/getting_started/installation/gpu/#build-wheel-from-source).
+Note: Pre-built wheels are currently available for vLLM-Omni 0.11.0rc1, 0.12.0rc1, 0.14.0rc1, 0.14.0, 0.16.0, 0.18.0, 0.20.0, 0.21.0, 0.22.0, 0.23.0, 0.24.0, 0.25.0, and 0.26.0; the 0.28.0 wheels are published together with the v0.28.0 release this branch produces. If you need a newer unreleased revision, please [build from source](https://docs.vllm.ai/projects/vllm-omni/en/latest/getting_started/installation/gpu/#build-wheel-from-source).

 === "NVIDIA CUDA"

@@ -5,7 +5,7 @@
 # --8<-- [end:requirements]
 # --8<-- [start:set-up-using-python]

-vLLM-Omni depends on the matching major/minor release of vLLM. The vLLM-Omni 0.28.x release line uses vLLM 0.28.x.
+vLLM-Omni depends on the matching major/minor release of vLLM. The vLLM-Omni 0.26.x release line uses vLLM 0.26.x.

 !!! note
    PyTorch installed via `conda` will statically link `NCCL` library, which can cause issues when vLLM tries to use `NCCL`. See <gh-issue:8420> for more details.
@@ -95,13 +95,6 @@ uv pip install --no-build-isolation --editable .

 vLLM-Omni offers an official docker image for deployment. These images are built on top of vLLM docker images and available on Docker Hub as [vllm/vllm-omni](https://hub.docker.com/r/vllm/vllm-omni/tags). The version of vLLM-Omni indicates which release of vLLM it is based on.

-Nightly CUDA images built from the latest `main` are also available:
-
- [`vllm/vllm-omni:nightly`](https://hub.docker.com/r/vllm/vllm-omni/tags?name=nightly) is the rolling multi-architecture tag.
- [`vllm/vllm-omni:nightly-<commit>`](https://hub.docker.com/r/vllm/vllm-omni/tags?name=nightly-) provides commit-pinned multi-architecture tags. The newest 14 are retained.
-
-Use a commit-pinned tag when you need a reproducible nightly environment.
-
 Here's an example deployment command that has been verified on 2 x H100's:
 ```bash
 docker run --runtime nvidia --gpus 2 \
@@ -23,19 +23,6 @@ vLLM-Omni currently recommends using the Docker image setup steps below.
 DOCKER_BUILDKIT=1 docker build -f docker/Dockerfile.xpu -t vllm-omni-xpu --shm-size=4g .
 ```

-This layers vLLM-Omni on top of the published `vllm/vllm-openai-xpu:<VLLM_VERSION>`
-base image, which Docker pulls automatically. To target a different vLLM release,
-pass `--build-arg VLLM_VERSION=<tag>`. If that tag has not been published yet, build
-the base from upstream's own Dockerfile first and point the build at it:
-
-```bash
-git clone --depth 1 --branch "$VLLM_VERSION" https://github.com/vllm-project/vllm /tmp/vllm
-DOCKER_BUILDKIT=1 docker build -t vllm-openai-xpu:local --target vllm-openai \
-  -f /tmp/vllm/docker/Dockerfile.xpu /tmp/vllm
-DOCKER_BUILDKIT=1 docker build -f docker/Dockerfile.xpu -t vllm-omni-xpu --shm-size=4g \
-  --build-arg VLLM_BASE=vllm-openai-xpu:local .
-```
-
 #### Launch the docker image

 ##### Launch with OpenAI API Server
@@ -9,8 +9,8 @@ The recommended way to use vLLM-Omni on NPU is through the vLLM-Ascend pre-built

 ```bash
 # vLLM-Ascend image
-# Atlas A2: quay.io/atlas-ci/vllm-ascend:v0.28.0
-# Atlas A3: quay.io/atlas-ci/vllm-ascend:v0.28.0-a3
+# Atlas A2: quay.io/atlas-ci/vllm-ascend:v0.26.0
+# Atlas A3: quay.io/atlas-ci/vllm-ascend:v0.26.0-a3
 docker run --rm \
    --name vllm-omni-npu \
    --shm-size=1g \
@@ -28,11 +28,11 @@ docker run --rm \
    -v /etc/ascend_install.info:/etc/ascend_install.info \
    -v /root/.cache:/root/.cache \
    -p 8000:8000 \
-    -it quay.io/atlas-ci/vllm-ascend:v0.28.0-a3 bash
+    -it quay.io/atlas-ci/vllm-ascend:v0.26.0-a3 bash

 # Inside the container, install vLLM-Omni from source
 cd /vllm-workspace
-git clone -b v0.28.0 https://github.com/vllm-project/vllm-omni.git
+git clone -b v0.26.0 https://github.com/vllm-project/vllm-omni.git
 cd vllm-omni
 pip install -v -e . --no-build-isolation
 # or VLLM_OMNI_TARGET_DEVICE=npu pip install -v -e .
@@ -50,11 +50,21 @@ We are keeping [issue #886](https://github.com/vllm-project/vllm-omni/issues/886

 # --8<-- [start:installation-main]

-You can also install vLLM-Omni from the latest main branch if you want to use the latest features or bug fixes. Use the aligned vLLM-Ascend v0.28.0 image described above, and check [issue #886](https://github.com/vllm-project/vllm-omni/issues/886) for the status of the latest vLLM-Omni commit on NPU.
+You can also build vLLM-Omni from the latest main branch if you want to use the latest features or bug fixes. (But sometimes it will break for a while. You can check [issue #886](https://github.com/vllm-project/vllm-omni/issues/886) for the status of the latest commit of vLLM-Omni main branch on NPU.)

 ```bash
-# Inside the vLLM-Ascend v0.28.0 container, install vLLM-Omni from main
-cd /vllm-workspace
+# Pin vLLM and vLLM-Ascend to the v0.26 release line
+git clone -b v0.26.0 https://github.com/vllm-project/vllm.git
+cd vllm
+VLLM_TARGET_DEVICE=empty pip install -v -e .
+cd ..
+
+git clone -b releases/v0.26.0rc https://github.com/vllm-project/vllm-ascend.git
+cd vllm-ascend
+pip install -v -e .
+cd ..
+
+# Install vLLM-Omni from the latest main branch
 git clone https://github.com/vllm-project/vllm-omni.git
 cd vllm-omni
 pip install -v -e . --no-build-isolation
@@ -78,8 +88,8 @@ Supported images as following.
 Here's an example deployment command that has been verified on 4 x NPUs:

 ```bash
-# Atlas A2: quay.io/ascend/vllm-omni:v0.28.0
-# Atlas A3: quay.io/ascend/vllm-omni:v0.28.0-a3
+# Atlas A2: quay.io/ascend/vllm-omni:v0.26.0
+# Atlas A3: quay.io/ascend/vllm-omni:v0.26.0-a3
 docker run --rm \
    --name vllm-omni-a3 \
    --shm-size=4g \
@@ -104,7 +114,7 @@ docker run --rm \
    -v /etc/ascend_install.info:/etc/ascend_install.info \
    -v ~/.cache:/root/.cache \
    -p 8091:8091 \
-    -it quay.io/ascend/vllm-omni:v0.28.0-a3 bash
+    -it quay.io/ascend/vllm-omni:v0.26.0-a3 bash
 ```

 !!! tip
@@ -91,7 +91,7 @@ if __name__ == "__main__":

 For more usages, please refer to [offline inference](../user_guide/examples/offline_inference/qwen2_5_omni.md)

-## Online Serving with the API Server
+## Online Serving with OpenAI-Completions API

 Text-to-image generation quickstart with vLLM-Omni:

@@ -100,16 +100,20 @@ vllm serve Tongyi-MAI/Z-Image-Turbo --omni --port 8091
 ```

 ```bash
-curl -s http://localhost:8091/v1/images/generations \
+curl -s http://localhost:8091/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
-    "prompt": "a cup of coffee on the table",
-    "size": "1024x1024",
-    "response_format": "b64_json",
-    "seed": 42
-  }' | jq -r '.data[0].b64_json' | base64 -d > coffee.png
+    "messages": [
+      {"role": "user", "content": "a cup of coffee on the table"}
+    ],
+    "extra_body": {
+      "height": 1024,
+      "width": 1024,
+      "num_inference_steps": 50,
+      "guidance_scale": 4.0,
+      "seed": 42
+    }
+  }' | jq -r '.choices[0].message.content[0].image_url.url' | cut -d',' -f2 | base64 -d > coffee.png
 ```

-See the [API Server guide](../serving/README.md) to choose an endpoint. For
-model-specific details, refer to the [text-to-image online serving
-example](../user_guide/examples/online_serving/text_to_image.md).
+For more details, please refer to [online serving](../user_guide/examples/online_serving/text_to_image.md).
@@ -21,20 +21,20 @@ th {
 |--------------|--------|-------------------|------------|---------|-----|-----------|--------|
 | `Qwen3OmniMoeForConditionalGeneration` | Qwen3-Omni | `Qwen/Qwen3-Omni-30B-A3B-Instruct` | ✅︎ | | | | [Repository](https://github.com/vllm-project/vllm-omni/blob/main/recipes/Qwen/Qwen3-Omni.md) |
 | `Qwen2_5OmniForConditionalGeneration` | Qwen2.5-Omni | `Qwen/Qwen2.5-Omni-7B`, `Qwen/Qwen2.5-Omni-3B` | ✅︎ | ✅︎ | ✅︎ | ✅︎ | — |
-| `MingFlashOmniForConditionalGeneration` + `MingImagePipeline` | Ming-flash-omni-2.0 (omni-speech + imagegen<sup>1</sup>) | `Jonathan1909/Ming-flash-omni-2.0` | ✅︎ | | | | — |
+| `MingFlashOmniForConditionalGeneration` + `MingImagePipeline` | Ming-flash-omni-2.0 (omni-speech + imagegen<sup>1</sup>) | `Jonathan1909/Ming-flash-omni-2.0` | ✅︎ |   |   |   | — |
 | `BagelForConditionalGeneration` | BAGEL (DiT-only) | `ByteDance-Seed/BAGEL-7B-MoT` | ✅︎ | ✅︎ | | ✅︎ | — |
 | `InternVLAA1Pipeline` | InternVLA-A1 | `InternRobotics/InternVLA-A1-3B` | ✅︎ | ✅︎ | | | — |
 | `Gr00tN1d7Pipeline` | GR00T N1.7 | `nvidia/GR00T-N1.7-3B` | ✅︎ | | | | — |
 | `HunyuanImage3ForCausalMM` | HunyuanImage3.0 (DiT-only) | `tencent/HunyuanImage-3.0`, `tencent/HunyuanImage-3.0-Instruct` | ✅︎ | ✅︎ | ✅︎ | ✅︎ | — |
-| `QwenImagePipeline` | Qwen-Image | `Qwen/Qwen-Image` | ✅︎ | | | ✅︎ | [Published](https://recipes.vllm.ai/Qwen/Qwen-Image) |
-| `QwenImagePipeline` | Qwen-Image-2512 | `Qwen/Qwen-Image-2512` | ✅︎ | | | ✅︎ | [Repository](https://github.com/vllm-project/vllm-omni/blob/main/recipes/Qwen/Qwen-Image-2512.md) |
+| `QwenImagePipeline` | Qwen-Image | `Qwen/Qwen-Image` | ✅︎ | | | | [Published](https://recipes.vllm.ai/Qwen/Qwen-Image) |
+| `QwenImagePipeline` | Qwen-Image-2512 | `Qwen/Qwen-Image-2512` | ✅︎ | | | | [Repository](https://github.com/vllm-project/vllm-omni/blob/main/recipes/Qwen/Qwen-Image-2512.md) |
 | `QwenImageEditPipeline` | Qwen-Image-Edit | `Qwen/Qwen-Image-Edit` | ✅︎ | ✅︎ | ✅︎ | ✅︎ | — |
 | `QwenImageEditPlusPipeline` | Qwen-Image-Edit-2509 | `Qwen/Qwen-Image-Edit-2509` | ✅︎ | ✅︎ | ✅︎ | ✅︎ | — |
 | `QwenImageLayeredPipeline` | Qwen-Image-Layered | `Qwen/Qwen-Image-Layered` | ✅︎ | ✅︎ | ✅︎ | ✅︎ | — |
 | `QwenImageEditPlusPipeline` | Qwen-Image-Edit-2511 | `Qwen/Qwen-Image-Edit-2511` | ✅︎ | ✅︎ | ✅︎ | ✅︎ | — |
 | `GlmImagePipeline` | GLM-Image | `zai-org/GLM-Image` | ✅︎ | ✅︎ | | | — |
 | `ZImagePipeline` | Z-Image | `Tongyi-MAI/Z-Image-Turbo` | ✅︎ | ✅︎ | ✅︎ | ✅︎ | — |
-| `Krea2Pipeline` | Krea 2 (Raw + Turbo) | `krea/Krea-2-Raw`, `krea/Krea-2-Turbo` | ✅︎ | | | ✅︎ | — |
+| `Krea2Pipeline` | Krea 2 (Raw + Turbo) | `krea/Krea-2-Raw`, `krea/Krea-2-Turbo` | ✅︎ | | | | — |
 | `WanPipeline` | Wan2.1-T2V, Wan2.2-T2V, Wan2.2-TI2V | `Wan-AI/Wan2.1-T2V-1.3B-Diffusers`, `Wan-AI/Wan2.1-T2V-14B-Diffusers`, `Wan-AI/Wan2.2-T2V-A14B-Diffusers`, `Wan-AI/Wan2.2-TI2V-5B-Diffusers` | ✅︎ | ✅︎ | ✅︎ | ✅︎ | — |
 | `WanImageToVideoPipeline` | Wan2.2-I2V | `Wan-AI/Wan2.2-I2V-A14B-Diffusers` | ✅︎ | ✅︎ | ✅︎ | ✅︎ | — |
 | `LingBotWorldCausalDMDPipeline` | LingBot-World 2.0 v1 (experimental) | `robbyant/lingbot-world-v2-14b-causal-fast-diffusers` | Experimental | | | | — |
@@ -42,7 +42,7 @@ th {
 | `SanaWmPipeline` | SANA-WM | `BBBBruce/SANA-WM_bidirectional-stage1-diffusers` | ✅︎ | | | | [Repository](https://github.com/vllm-project/vllm-omni/blob/main/recipes/NVIDIA/SANA-WM.md) |
 | `Wan22S2VPipeline` | Wan2.2-S2V | `Wan-AI/Wan2.2-S2V-14B` | ✅︎ | ✅︎ | ✅︎ | ✅︎ | — |
 | `Wan22VACEPipeline` | Wan2.1-VACE | `Wan-AI/Wan2.1-VACE-1.3B-diffusers`, `Wan-AI/Wan2.1-VACE-14B-diffusers` | ✅︎ | ✅︎ | ✅︎ | ✅︎ | — |
-| `Wan22VACEPipeline` | Wan2.2-VACE | `Pyros13/Wan2.2-VACE-Fun-A14B-Diffusers` | ✅︎ |   |   | ✅︎ | — |
+| `Wan22VACEPipeline` | Wan2.2-VACE | `Pyros13/Wan2.2-VACE-Fun-A14B-Diffusers` | ✅︎ |   |   |   | — |
 | `LTX2Pipeline` | LTX-2 / LTX-2.3 one-stage T2V and I2V | `Lightricks/LTX-2`, `diffusers/LTX-2.3-Diffusers` | ✅︎ | ✅︎ | | | — |
 | `LTX2TwoStagePipeline` | LTX-2 / LTX-2.3 ordinary two-stage T2V and I2V | `Lightricks/LTX-2`, `diffusers/LTX-2.3-Diffusers` + matching Lightricks LoRA and upsampler | ✅︎ | ✅︎ | | | — |
 | `LTX2DistilledOneStagePipeline` | LTX-2 / LTX-2.3 merged-distilled one-stage T2V and I2V | `rootonchair/LTX-2-19b-distilled`, `diffusers/LTX-2.3-Distilled-Diffusers` | ✅︎ | ✅︎ | | | — |
@@ -62,12 +62,12 @@ th {
 | `LongcatImagePipeline` | LongCat-Image | `meituan-longcat/LongCat-Image` | ✅︎ | ✅︎ | ✅︎ | ✅︎ | — |
 | `LongCatImageEditPipeline` | LongCat-Image-Edit | `meituan-longcat/LongCat-Image-Edit` | ✅︎ | ✅︎ | ✅︎ | ✅︎ | — |
 | `LongCatVideoAvatarPipeline` | LongCat-Video-Avatar-1.5 A2V/AI2V (native single-speaker and multi-speaker AI2V/AVC) | `meituan-longcat/LongCat-Video-Avatar-1.5` | ✅︎ | | | | [Repository](https://github.com/vllm-project/vllm-omni/blob/main/recipes/meituan-longcat/LongCat-Video-Avatar-1.5.md) |
-| `BooguImagePipeline` | Boogu-Image | `Boogu/Boogu-Image-0.1-Base`, `Boogu/Boogu-Image-0.1-Edit`, `Boogu/Boogu-Image-0.1-Edit-Turbo` | ✅︎ | | | ✅︎ | [Repository](https://github.com/vllm-project/vllm-omni/blob/main/recipes/Boogu/Boogu-Image.md) |
+| `BooguImagePipeline` | Boogu-Image | `Boogu/Boogu-Image-0.1-Base`, `Boogu/Boogu-Image-0.1-Edit` | ✅︎ | | | | — |
 | `StableDiffusionXLPipeline` | Stable-Diffusion-XL | `stabilityai/stable-diffusion-xl-base-1.0` | ✅︎ | ✅︎ | ✅︎ | ✅︎ | — |
 | `StableDiffusion3Pipeline` | Stable-Diffusion-3 | `stabilityai/stable-diffusion-3.5-medium` | ✅︎ | ✅︎ | | ✅︎ | — |
 | `CosyVoice3Model` | CosyVoice3 | `FunAudioLLM/Fun-CosyVoice3-0.5B-2512` | ✅︎ | ✅︎ | | ✅︎ | — |
-| `OmniVoiceModel` | OmniVoice | `k2-fsa/OmniVoice` | ✅︎ | | | ✅︎ | — |
-| `VoxCPM2TalkerForConditionalGeneration` | VoxCPM2 | `openbmb/VoxCPM2` | ✅︎ | | | ✅︎ | — |
+| `OmniVoiceModel` | OmniVoice | `k2-fsa/OmniVoice` | ✅︎ | | | | — |
+| `VoxCPM2TalkerForConditionalGeneration` | VoxCPM2 | `openbmb/VoxCPM2` | ✅︎ | | | | — |
 | `DotsTTSForConditionalGeneration` | dots.tts | `rednote-hilab/dots.tts-soar` | ✅︎ | | | | — |
 | `MammothModa2ForConditionalGeneration` | MammothModa2-Preview | `bytedance-research/MammothModa2-Preview` | ✅︎ | ✅︎ | | | — |
 | `MammothModa2ForConditionalGeneration` | MammothModa2-Dev (AR-only image understanding) | `bytedance-research/MammothModa2-Dev` | ✅︎ | | | | — |
@@ -78,10 +78,10 @@ th {
 | `OmniGen2Pipeline` | OmniGen2 | `OmniGen2/OmniGen2` | ✅︎ | ✅︎ | | ✅︎ | — |
 | `StableAudioPipeline` | Stable-Audio-Open | `stabilityai/stable-audio-open-1.0` | ✅︎ | ✅︎ | | ✅︎ | — |
 | `MiniMaxMusic3ForConditionalGeneration` | MiniMax Music 3 (text-to-music) | `MiniMaxAI/MiniMax-Music3` | ✅︎ | | | | — |
-| `Qwen3TTSForConditionalGeneration` | Qwen3-TTS-12Hz-1.7B-CustomVoice | `Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice` | ✅︎ | | | ✅︎ | [Published](https://recipes.vllm.ai/Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice) |
+| `Qwen3TTSForConditionalGeneration` | Qwen3-TTS-12Hz-1.7B-CustomVoice | `Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice` | ✅︎ | | | | [Published](https://recipes.vllm.ai/Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice) |
 | `Qwen3TTSForConditionalGeneration` | Qwen3-TTS-12Hz-1.7B-VoiceDesign | `Qwen/Qwen3-TTS-12Hz-1.7B-VoiceDesign` | ✅︎ | ✅︎ | ✅︎ | ✅︎ | — |
 | `Qwen3TTSForConditionalGeneration` | Qwen3-TTS-12Hz-1.7B-Base | `Qwen/Qwen3-TTS-12Hz-1.7B-Base` | ✅︎ | ✅︎ | ✅︎ | ✅︎ | — |
-| `MingTTSForConditionalGeneration` | Ming-omni-tts dense 0.5B | `inclusionAI/Ming-omni-tts-0.5B` | ✅︎ | | | ✅︎ | — |
+| `MingTTSForConditionalGeneration` | Ming-omni-tts dense 0.5B | `inclusionAI/Ming-omni-tts-0.5B` | ✅︎ | | | | — |
 | `GLMTTSForConditionalGeneration` | GLM-TTS | `zai-org/GLM-TTS` | ✅︎ | | | | — |
 | `MossTTSNanoForCausalLM` | MOSS-TTS-Nano | `OpenMOSS-Team/MOSS-TTS-Nano` | ✅︎ | | | | — |
 | `MossTTSDelayModel` | MOSS-TTS, MOSS-TTSD, MOSS-SoundEffect, MOSS-VoiceGenerator | `OpenMOSS-Team/MOSS-VoiceGenerator` | ✅︎ | | | | — |
@@ -95,16 +95,16 @@ th {
 | `MiMoV2ASRForCausalLM` | MiMo-V2.5-ASR | `XiaomiMiMo/MiMo-V2.5-ASR` | ✅︎ | ✅︎ | | | — |
 | `Flux2Pipeline` | FLUX.2-dev | `black-forest-labs/FLUX.2-dev` | ✅︎ | ✅︎ | | | — |
 | `FishSpeechSlowARForConditionalGeneration` | Fish Speech S2 Pro | `fishaudio/s2-pro` | ✅︎ | ✅︎ | | ✅︎ | — |
-| `SenseNovaU1Pipeline` | SenseNova-U1 (DiT-only), SenseNova-U1.5 | `SenseNova/SenseNova-U1-8B-MoT`, `sensenova/SenseNova-U1.5-8B-MoT` | ✅︎ | | | | [Repository](https://github.com/vllm-project/vllm-omni/blob/main/recipes/SenseNova/SenseNova-U1.5.md) |
+| `SenseNovaU1Pipeline` | SenseNova-U1 (DiT-only) | `SenseNova/SenseNova-U1-8B-MoT` | ✅︎ | | | | — |
 | `LancePipeline` | Lance | `bytedance-research/Lance` | ✅︎ | | | | — |
 | `HunyuanVideo15Pipeline` | HunyuanVideo-1.5-T2V | `hunyuanvideo-community/HunyuanVideo-1.5-Diffusers-480p_t2v`, `hunyuanvideo-community/HunyuanVideo-1.5-Diffusers-720p_t2v` | ✅︎ | ✅︎ | | | — |
 | `HunyuanVideo15ImageToVideoPipeline` | HunyuanVideo-1.5-I2V | `hunyuanvideo-community/HunyuanVideo-1.5-Diffusers-480p_i2v`, `hunyuanvideo-community/HunyuanVideo-1.5-Diffusers-720p_i2v` | ✅︎ | ✅︎ | | | — |
 | `VoxtralTTSForConditionalGeneration` | Voxtral TTS | `mistralai/Voxtral-4B-TTS-2603` | ✅︎ | ✅︎ | | | — |
 | `CovoAudioForConditionalGeneration` | Covo-Audio-Chat | `tencent/Covo-Audio-Chat` | ✅︎ | | | | — |
-|`DyninOmniForConditionalGeneration` | Dynin-Omni | `snu-aidas/Dynin-Omni` | ✅︎ | | | ✅︎ | — |
+|`DyninOmniForConditionalGeneration` | Dynin-Omni | `snu-aidas/Dynin-Omni` | ✅︎ | | | | — |
 | `MiniCPMO45OmniForConditionalGeneration` | MiniCPM-o 4.5 | `openbmb/MiniCPM-o-4_5` | ✅︎ | | ✅︎ | | — |
 | `ErnieImagePipeline` | ERNIE-Image | `baidu/ERNIE-Image`, `baidu/ERNIE-Image-Turbo` | ✅︎ | ✅︎ | ✅︎ | ✅︎ | — |
-| `GepardTalkerForConditionalGeneration` | Gepard-1.0 | `nineninesix/gepard-1.0` | ✅︎ | | | ✅︎ | — |
+| `GepardTalkerForConditionalGeneration` | Gepard-1.0 | `nineninesix/gepard-1.0` | ✅︎ | | | | — |
 |`HiDreamImagePipeline` | HiDream-I1-Full | `HiDream-ai/HiDream-I1-Full` | ✅︎ | ✅︎ | | | — |
 | `HiDreamO1ImagePipeline` | HiDream-O1-Image | `HiDream-ai/HiDream-O1-Image` | ✅︎ | | | | — |

@@ -1,187 +0,0 @@
-# API Server
-
-vLLM-Omni exposes OpenAI-compatible HTTP APIs plus vLLM-Omni extensions for
-image, audio, video, realtime, and robot-policy workloads. Use this page to
-choose an endpoint; follow the linked reference page for model-specific fields,
-wire protocols, and response details.
-
-Each server instance hosts one model. An endpoint is usable only when the
-loaded model supports that task.
-
-This guide focuses on vLLM-Omni's primary task endpoints. Compatible models
-may also expose standard routes inherited from vLLM, such as
-`POST /v1/completions`; those routes remain model- and task-dependent.
-
-## Start and Verify the Server
-
-Start a model with the shared serving command:
-
-```bash
-vllm serve <model> --omni --port 8091
-```
-
-Some models require additional flags or a deployment configuration. Use the
-model-specific guide when one is provided.
-
-After the server starts, check its health and served model name:
-
-```bash
-export VLLM_OMNI_BASE_URL=http://localhost:8091
-
-curl "$VLLM_OMNI_BASE_URL/health"
-curl "$VLLM_OMNI_BASE_URL/v1/models" | jq .
-```
-
-Use `http://localhost:8091/v1` as the `base_url` for an OpenAI SDK client. If
-the server was started with `--api-key`, include an `Authorization` header with
-`Bearer <api-key>` in requests.
-
-## Choose a Core HTTP Endpoint
-
-Prefer a task-specific endpoint when one matches your workload. Use Chat
-Completions for conversational or heterogeneous omni pipelines, rather than as
-the default wrapper for every generation task.
-
-| Task | Endpoint | Request | Response | Details |
-|------|----------|---------|----------|---------|
-| Conversation or multimodal understanding/generation | `POST /v1/chat/completions` | JSON | OpenAI-style JSON or SSE | [Chat Completions](chat_completions_api.md) |
-| Text-to-speech | `POST /v1/audio/speech` | JSON | Audio bytes or SSE | [Speech](speech_api.md) |
-| Sound, music, or ambient audio generation | `POST /v1/audio/generate` | JSON | Audio bytes | [Audio Generation](audio_generate_api.md) |
-| Text-to-image | `POST /v1/images/generations` | JSON | Base64 JSON or an image file | [Image Generation](image_generation_api.md) |
-| Image editing | `POST /v1/images/edits` | Multipart form | Base64 JSON or SSE | [Image Edit](image_edit_api.md) |
-| Video generation (recommended) | `POST /v1/videos` | Multipart form | Asynchronous job JSON | [Videos](videos_api.md) |
-| Video generation (blocking) | `POST /v1/videos/sync` | Multipart form | Video bytes | [Videos](videos_api.md#synchronous-response) |
-
-`POST /v1/videos/sync` is intended for tests and simple scripts. Use the
-asynchronous job API for production requests so clients can poll, download,
-and delete generated videos independently.
-
-## Minimal Requests
-
-The examples below assume a compatible model is already running. TTS voices,
-media inputs, generation controls, and output capabilities vary by model.
-
-=== "Chat"
-
-    ```bash
-    curl "$VLLM_OMNI_BASE_URL/v1/chat/completions" \
-      -H "Content-Type: application/json" \
-      -d '{
-        "messages": [{"role": "user", "content": "Describe vLLM-Omni briefly."}],
-        "modalities": ["text"]
-      }'
-    ```
-
-=== "Speech"
-
-    ```bash
-    curl "$VLLM_OMNI_BASE_URL/v1/audio/speech" \
-      -H "Content-Type: application/json" \
-      -d '{
-        "input": "Hello from vLLM-Omni.",
-        "voice": "vivian"
-      }' \
-      --output speech.wav
-    ```
-
-=== "Audio"
-
-    ```bash
-    curl "$VLLM_OMNI_BASE_URL/v1/audio/generate" \
-      -H "Content-Type: application/json" \
-      -d '{
-        "input": "Ocean waves on a quiet beach",
-        "audio_length": 5.0
-      }' \
-      --output audio.wav
-    ```
-
-=== "Image"
-
-    ```bash
-    curl "$VLLM_OMNI_BASE_URL/v1/images/generations" \
-      -H "Content-Type: application/json" \
-      -d '{
-        "prompt": "A small robot reading beside a window",
-        "size": "1024x1024",
-        "response_format": "file"
-      }' \
-      --output image.png
-    ```
-
-=== "Image edit"
-
-    ```bash
-    curl "$VLLM_OMNI_BASE_URL/v1/images/edits" \
-      -F "image=@input.png" \
-      -F "prompt=Turn the background into a snowy mountain" \
-      | jq -r '.data[0].b64_json' \
-      | base64 --decode > edited.png
-    ```
-
-=== "Video"
-
-    ```bash
-    curl "$VLLM_OMNI_BASE_URL/v1/videos" \
-      -F "prompt=A cinematic tracking shot of a mountain lake at sunrise" \
-      -F "size=1280x720" \
-      -F "seconds=5"
-    ```
-
-    The response contains a video job `id`. Poll and download it with:
-
-    ```bash
-    curl "$VLLM_OMNI_BASE_URL/v1/videos/<video_id>"
-    curl -L "$VLLM_OMNI_BASE_URL/v1/videos/<video_id>/content" --output video.mp4
-    ```
-
-## Choose a Streaming or Realtime Endpoint
-
-These WebSocket APIs have different event schemas and cannot be used
-interchangeably.
-
-| Workload | Endpoint | Interaction model | Details |
-|----------|----------|-------------------|---------|
-| Incremental text input for speech synthesis | `WS /v1/audio/speech/stream` | Send text events and receive audio | [Streaming Text to Speech](speech_api.md#streaming-text-input-websocket) |
-| Live video understanding | `WS /v1/video/chat/stream` | Send video frames and receive text/audio | [Streaming Video Input](video_stream_api.md) |
-| Turn-based realtime audio | `WS /v1/realtime` | Stream one audio input and receive transcript/audio events | [Realtime Audio](realtime_api.md) |
-| Continuous speech-to-speech interaction | `WS /v1/realtime?duplex=1` or `WS /v1/duplex` | Listen and speak concurrently with session control | [Full Duplex](full_duplex_api.md) |
-| Generated video chunks | `WS /v1/realtime/video` | Start a diffusion request and receive fragmented MP4 | [Streaming Video Output](streaming_video_output_api.md) |
-| Robot policy inference | `WS /v1/realtime/robot/openpi` | Send MessagePack observations and receive action arrays | [OpenPI Robot Policy](openpi_api.md) |
-
-All six routes are model- or configuration-dependent. In particular,
-`/v1/realtime` is not full duplex unless the client sets `duplex=1` and the
-deployment explicitly enables duplex sessions. Clients should also verify the
-duplex capability payload because the query-parameter form falls back to the
-ordinary realtime handler when duplex is unavailable.
-
-## Related Endpoints
-
-| Purpose | Endpoints | Reference |
-|---------|-----------|-----------|
-| Discovery and readiness | `GET /health`, `GET /v1/models` | This page |
-| Batched conversations | `POST /v1/chat/completions/batch` | [Batch requests](chat_completions_api.md#batch-requests) |
-| Batched speech | `POST /v1/audio/speech/batch` | [Batch speech generation](speech_api.md#batch-speech-generation) |
-| TTS voice management | `GET/POST /v1/audio/voices`, `DELETE /v1/audio/voices/{name}` | [Voices](speech_api.md#voices-endpoint) |
-| Video job lifecycle | `GET /v1/videos`, `GET/DELETE /v1/videos/{video_id}`, `GET /v1/videos/{video_id}/content` | [Video endpoints](videos_api.md#endpoints) |
-| Release and restore stage memory | `POST /v1/omni/sleep`, `POST /v1/omni/wakeup` | [Sleep Mode](../features/sleep_mode.md) |
-
-## Standalone Experimental Servers
-
-PersonaPlex also provides a Moshi-compatible WebSocket server, while JoyVL
-provides a stateful interaction orchestrator in front of another model server.
-These are separate processes with their own routes; they are not additional
-paths on every `vllm serve --omni` instance. See [Standalone Experimental
-Servers](standalone_servers.md) before deploying either one.
-
-## Compatibility Notes
-
- A registered path does not mean every model supports it. Unsupported tasks
-  return an OpenAI-style error or a WebSocket `error` event.
- Use JSON for Chat, Speech, Audio Generation, and Image Generation. Use
-  `multipart/form-data` when uploading images, videos, or other files.
- Generation can take minutes for large diffusion models. Configure client
-  timeouts accordingly; prefer asynchronous video jobs over long blocking
-  requests.
- Model-specific parameters are vLLM-Omni extensions. Check the endpoint
-  reference before assuming an OpenAI SDK exposes them directly.
@@ -4,14 +4,14 @@ vLLM-Omni provides an API for text-to-audio generation using diffusion-based mod

 Unlike the [Speech API](speech_api.md) which targets text-to-speech synthesis, the Audio Generate API is designed for general-purpose audio generation from text descriptions (sound effects, music, ambient soundscapes, etc.).

-Each server instance runs a single model (specified at startup via `vllm serve <model> --omni`).
+Each server instance runs a single model (specified at startup via `vllm-omni serve <model> --omni`).

 ## Quick Start

 ### Start the Server

 ```bash
-vllm serve stabilityai/stable-audio-open-1.0 \
+vllm-omni serve stabilityai/stable-audio-open-1.0 \
    --host 0.0.0.0 \
    --port 8091 \
    --gpu-memory-utilization 0.9 \
@@ -326,7 +326,7 @@ curl http://localhost:8091/health
 Enable debug logging:

 ```bash
-vllm serve stabilityai/stable-audio-open-1.0 \
+vllm-omni serve stabilityai/stable-audio-open-1.0 \
    --host 0.0.0.0 \
    --port 8091 \
    --gpu-memory-utilization 0.9 \
@@ -1,110 +1,78 @@
 # Chat Completions API

-Use `/v1/chat/completions` for conversational and multimodal pipelines that
-follow the OpenAI Chat API schema. Depending on the loaded model, a request can
-contain text, image, audio, or video input and can return text, audio, images,
-or other model-specific output.
-
-Some diffusion pipelines also support image generation and editing through
-Chat Completions. Prefer the dedicated [Image Generation
-API](image_generation_api.md) or [Image Edit API](image_edit_api.md) for
-task-specific clients; their request fields and response contracts are more
-direct.
-
-## Basic Request
-
-```bash
-curl http://localhost:8091/v1/chat/completions \
-  -H "Content-Type: application/json" \
-  -d '{
-    "messages": [
-      {"role": "user", "content": "Describe vLLM-Omni briefly."}
-    ],
-    "modalities": ["text"]
-  }'
-```
-
-Use `stream: true` for Server-Sent Events. Input media syntax, supported output
-`modalities`, and response choices depend on the model; see the
-[model-specific online serving examples](../user_guide/examples/online_serving/qwen3_omni.md).
+vLLM-Omni supports generating and editing images via the `/v1/chat/completions`
+endpoint using diffusion models. This page explains how to pass generation
+parameters (such as `num_inference_steps`, `height`, `width`) to diffusion
+models through this endpoint.

 !!! tip
-    Each server hosts one model. Query `GET /v1/models` for its served model
-    name when your client requires the `model` field.
+    For dedicated endpoints that accept generation parameters as top-level
+    fields, see [Image Generation API](image_generation_api.md) and
+    [Image Edit API](image_edit_api.md).

-## vLLM Extension Parameters
+## Passing Generation Parameters

-Send standard and vLLM-specific Chat parameters as top-level fields in a
-direct HTTP request. When using the OpenAI Python SDK, pass vLLM-specific
-fields through the SDK's `extra_body` keyword; the SDK merges them into the
-top-level JSON sent to the server.
+The `/v1/chat/completions` endpoint follows the OpenAI Chat API schema, which
+does not natively include diffusion-specific fields like `num_inference_steps`
+or `height`. How you pass these extra fields depends on your client.

-=== "curl"
+### curl / Python `requests`

-    ```bash
-    curl http://localhost:8091/v1/chat/completions \
-      -H "Content-Type: application/json" \
-      -d '{
-        "messages": [{"role": "user", "content": "Write a short haiku."}],
-        "top_k": 40
-      }'
-    ```
-
-=== "OpenAI Python SDK"
-
-    ```python
-    from openai import OpenAI
-
-    client = OpenAI(base_url="http://localhost:8091/v1", api_key="none")
-
-    response = client.chat.completions.create(
-        model="your-served-model",
-        messages=[{"role": "user", "content": "Write a short haiku."}],
-        extra_body={"top_k": 40},
-    )
-    ```
-
-??? note "Diffusion compatibility"
-
-    Diffusion models exposed through Chat Completions can accept supported
-    fields such as `num_inference_steps`, `seed`, `height`, and `width`. Send
-    them at the top level with direct HTTP, or in the SDK's `extra_body`
-    keyword argument.
-
-    A literal nested `"extra_body"` JSON object is accepted for compatibility,
-    but it is not recommended for new direct HTTP clients. Do not provide the
-    same parameter in multiple locations; the server returns a `400` error for
-    duplicate diffusion parameters. Prefer the dedicated [Image Generation
-    API](image_generation_api.md) or [Image Edit API](image_edit_api.md) when
-    either endpoint matches the task.
-
-## Batch Requests
-
-`POST /v1/chat/completions/batch` accepts the same shared generation fields,
-but `messages` is a list of conversations. The response contains one choice
-per conversation in input order.
+Wrap generation parameters inside an `"extra_body"` key in the JSON body:

 ```bash
-curl http://localhost:8091/v1/chat/completions/batch \
+curl -s http://localhost:8091/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "messages": [
-      [{"role": "user", "content": "Summarize vLLM in one sentence."}],
-      [{"role": "user", "content": "Summarize vLLM-Omni in one sentence."}]
+      {"role": "user", "content": "A beautiful landscape painting"}
    ],
-    "max_tokens": 64
+    "extra_body": {
+      "num_inference_steps": 50,
+      "seed": 42
+    }
  }'
 ```

-Batch Chat Completions does not support streaming, tools, beam search, or
-`n > 1`.
+### OpenAI Python SDK
+
+Use the `extra_body` **keyword argument**. The SDK automatically merges these
+fields into the top-level request body:
+
+```python
+response = client.chat.completions.create(
+    model="Qwen/Qwen-Image",
+    messages=[{"role": "user", "content": "A beautiful landscape painting"}],
+    extra_body={
+        "num_inference_steps": 50,
+        "seed": 42,
+    },
+)
+```
+
+!!! note "SDK `extra_body` vs. JSON `extra_body`"
+    These two `extra_body` usages look similar but work differently under the
+    hood. The SDK flattens the dict into the top-level request JSON, while the
+    curl/requests approach sends it as a nested `"extra_body"` key. Both are
+    handled correctly by the server.
+
+!!! note "About the `ignored fields` warning"
+    You may see a log message like:
+
+    ```
+    WARNING: The following fields were present in the request but ignored: {'height', 'width', ...}
+    ```
+
+    This is **harmless**. It is emitted by vLLM's request validation layer
+    because these fields are not part of the standard OpenAI
+    `ChatCompletionRequest` schema. The fields are still stored internally
+    and correctly forwarded to the diffusion pipeline.

 ## Model-Specific Examples

-For complete examples with model-specific inputs and outputs, see:
+For complete examples with full request/response details, see the model-specific
+guides:

- [Qwen3-Omni](../user_guide/examples/online_serving/qwen3_omni.md)
- [Qwen2.5-Omni](../user_guide/examples/online_serving/qwen2_5_omni.md)
 - [Text-to-Image (Qwen-Image)](../user_guide/examples/online_serving/text_to_image.md)
 - [Image-to-Image (Qwen-Image-Edit, Qwen-Image-Layered)](../user_guide/examples/online_serving/image_to_image.md)
 - [GLM-Image](../user_guide/examples/online_serving/glm_image.md)
@@ -1,116 +0,0 @@
-# Full-Duplex WebSocket API
-
-vLLM-Omni provides an experimental full-duplex runtime for models that can
-continue receiving speech while producing speech. It adds persistent session
-state, model-specific turn policy, overlap handling, playback acknowledgement,
-and optional session resume.
-
-Full duplex is distinct from the turn-based [Realtime Audio API](realtime_api.md).
-
-## Choose an Endpoint
-
-| Endpoint | Protocol | Recommended use |
-|----------|----------|-----------------|
-| `WS /v1/realtime?duplex=1` | OpenAI Realtime-style event projection | Applications and browser clients |
-| `WS /v1/duplex` | Native vLLM-Omni duplex events | Runtime integration and low-level testing |
-
-Both endpoints use the same duplex engine and require the same model-side
-adapter. Prefer the Realtime projection unless the native lifecycle events are
-specifically required.
-
-## Enable Full Duplex
-
-The route becomes usable only when the deployment configuration explicitly
-sets:
-
-```yaml
-session_mode: duplex
-```
-
-The selected model pipeline must also provide a duplex serving adapter.
-Model-native deployments configure the corresponding engine runtime extension
-and control plane as part of their registered pipeline.
-
-!!! warning
-
-    `WS /v1/duplex` fails with `Duplex API is not available` when duplex is not
-    enabled. By contrast, `/v1/realtime?duplex=1` falls back to the ordinary
-    turn-based realtime handler when no duplex handler exists. Confirm that
-    `session.created.session.capabilities` is present before treating the
-    connection as full duplex.
-
-The current unified-runtime integrations are:
-
- MiniCPM-o 4.5, using `vllm_omni/deploy/minicpmo_4_5_duplex.yaml`;
- PersonaPlex, whose default `vllm_omni/deploy/personaplex.yaml` enables duplex.
-
-JoyVL is a separate HTTP interaction orchestrator and does not use these
-WebSocket endpoints. See [Standalone Experimental Servers](standalone_servers.md).
-
-## MiniCPM-o Quick Start
-
-Start the duplex deployment:
-
-```bash
-vllm serve openbmb/MiniCPM-o-4_5 --omni \
-  --deploy-config vllm_omni/deploy/minicpmo_4_5_duplex.yaml \
-  --trust-remote-code \
-  --port 8091
-```
-
-Stream a mono, PCM16, 16 kHz WAV file with the provided client:
-
-```bash
-python examples/online_serving/minicpmo/realtime_duplex_demo.py \
-  --url 'ws://localhost:8091/v1/realtime?duplex=1' \
-  --model openbmb/MiniCPM-o-4_5 \
-  --input-wav input_16k_mono.wav \
-  --ref-audio reference_voice.wav \
-  --output-dir /tmp/minicpmo-duplex
-```
-
-## Realtime Event Lifecycle
-
-A typical `/v1/realtime?duplex=1` session follows this lifecycle:
-
-1. Send `session.update` with the model, modalities, audio formats, and session
-   options.
-2. Wait for `session.created`; inspect `session.capabilities` instead of
-   assuming every model supports the same controls.
-3. Send `input_audio_buffer.append` events while microphone audio arrives.
-4. Send `input_audio_buffer.commit` at a user-turn boundary when required by
-   the model policy.
-5. Consume `response.created`, transcript deltas, `response.audio.delta`, and
-   `response.done` or `response.listen` events.
-6. Send `playback.ack` after audio has been played when the session advertises
-   playback acknowledgement support.
-7. Send `session.close` and wait for `session.closed`.
-
-Unlike the turn-based realtime endpoint, input may continue while a response
-is active. The server can emit `overlap.decision` to describe whether input was
-deferred, treated as a short acknowledgement, or used to interrupt output.
-
-## Capabilities and Model Differences
-
-The `session.created` payload includes capability fields such as
-`supports_barge_in`, `supports_playback_ack`, `supports_multi_session`,
-`supports_session_resume`, and `chunk_period_ms`. Treat this payload as the
-runtime contract.
-
-For example, PersonaPlex supports native overlapping speech but currently
-advertises `supports_barge_in=false`; destructive output interruption and
-model-state rewind have not been validated for that integration. Capacity and
-session-resume behavior also depend on the selected deployment configuration.
-
-## Native Protocol
-
-`WS /v1/duplex` exposes lower-level lifecycle names including
-`session.create`, `input_audio_buffer.append`, `turn.signal`, `playback.ack`,
-and `session.close`. It returns native session, input, response, overlap, and
-error events. This protocol is experimental and may evolve with the runtime;
-applications should use the provided Realtime client where possible.
-
-See the [MiniCPM-o example](https://github.com/vllm-project/vllm-omni/tree/main/examples/online_serving/minicpmo),
-[PersonaPlex example](https://github.com/vllm-project/vllm-omni/tree/main/examples/online_serving/personaplex),
-and [full-duplex runtime design](https://github.com/vllm-project/vllm-omni/blob/main/vllm_omni/experimental/fullduplex/DESIGN.md)
-for model-specific validation and architecture details.
@@ -1,109 +0,0 @@
-# OpenPI Robot Policy WebSocket API
-
-Use `WS /v1/realtime/robot/openpi` for low-latency robot-policy inference with
-OpenPI-compatible clients. A persistent WebSocket carries observations to the
-loaded policy model and returns action arrays.
-
-Despite the `/realtime` prefix, this endpoint does not use the OpenAI Realtime
-event schema. It uses binary MessagePack frames with NumPy extensions.
-
-## Availability
-
-The endpoint is available only when the loaded diffusion policy configuration
-contains `policy_server_config`. The server sends that model-specific
-configuration to the client as the first binary frame.
-
-For example, DreamZero declares:
-
-```yaml
-policy_server_config:
-  image_resolution: [180, 320]
-  n_external_cameras: 2
-  needs_wrist_camera: true
-  needs_stereo_camera: false
-  needs_session_id: true
-  action_space: joint_position
-```
-
-Without this configuration, the WebSocket returns `Robot policy not available`
-and closes.
-
-## DreamZero Quick Start
-
-Start the bundled DreamZero deployment:
-
-```bash
-vllm serve GEAR-Dreams/DreamZero-DROID --omni --port 8091 \
-  --served-model-name dreamzero-droid \
-  --deploy-config vllm_omni/deploy/dreamzero_tp1_cfg2.yaml \
-  --enforce-eager --disable-log-stats
-```
-
-Install the optional client dependencies, download the sample camera inputs,
-and run the client:
-
-```bash
-pip install openpi-client websockets opencv-python
-
-hf download YangshenDeng/vllm-omni-dreamzero-assets \
-  --repo-type dataset \
-  --local-dir outputs/dreamzero/assets
-
-python examples/online_serving/dreamzero/openpi_client.py \
-  --host 127.0.0.1 \
-  --port 8091 \
-  --video-dir outputs/dreamzero/assets
-```
-
-## Protocol
-
-The connection is request-response after an initial server handshake:
-
-```text
-connect
-  <- msgpack(policy_server_config)
-
-infer
-  -> msgpack({"endpoint": "infer", "session_id": "...", ...observation})
-  <- msgpack(ndarray | dict[str, ndarray])
-
-reset
-  -> msgpack({"endpoint": "reset"})
-  <- msgpack({"status": "reset successful"})
-```
-
-If `endpoint` is omitted, the server treats the message as `infer`. Each
-inference message produces one action response after the engine request
-completes; action tokens or intermediate tensors are not streamed.
-
-NumPy arrays use the marker format implemented by `openpi-client`. The server
-also accepts the legacy vLLM NumPy marker representation on input. JSON text
-frames are not observation messages.
-
-## Sessions and Reset
-
- The API layer's current-session and first-call counters are scoped to each
-  WebSocket connection.
- `session_id` identifies model-side state across observations on that
-  connection. If omitted, it defaults to `default`.
- The first inference for a session is sent to the model with `reset=true`.
- Changing `session_id`, or sending the `reset` command, causes the next
-  inference to start with `reset=true`.
- The policy pipeline owns observation transforms and persistent model state,
-  normally keyed by `session_id`; the API layer forwards the raw observation
-  dictionary.
-
-## Limits and Errors
-
- Maximum inbound payload size is 64 MiB.
- The server closes an idle connection after 30 seconds.
- Invalid binary input returns a MessagePack
-  `{"type":"error","message":"Invalid request payload"}` response.
- Inference failures return a generic `Internal inference error` without
-  exposing an internal traceback.
- Unsupported NumPy object, structured, and complex dtypes are rejected.
-
-Observation keys, camera layout, state tensors, and action shapes are defined
-by the loaded policy rather than by this transport. Read the handshake before
-constructing observations. See the [DreamZero example](https://github.com/vllm-project/vllm-omni/tree/main/examples/online_serving/dreamzero)
-for a complete OpenPI client and DROID simulation loop.
@@ -1,82 +0,0 @@
-# Realtime Audio WebSocket API
-
-Use `WS /v1/realtime` to stream PCM audio into a compatible omni model and
-receive incremental transcription and synthesized response audio. This is a
-turn-based streaming API: it starts one generation over an audio stream and
-finishes after the client closes that input stream.
-
-For a model that listens while it is already speaking, use the experimental
-[Full-Duplex API](full_duplex_api.md) instead.
-
-## Quick Start
-
-Start a compatible model. The currently documented example uses Qwen3-Omni:
-
-```bash
-vllm serve Qwen/Qwen3-Omni-30B-A3B-Instruct --omni --port 8091
-```
-
-Run the provided client with a mono, 16-bit PCM, 16 kHz WAV file:
-
-```bash
-python examples/online_serving/qwen3_omni/openai_realtime_client.py \
-  --url ws://localhost:8091/v1/realtime \
-  --model Qwen/Qwen3-Omni-30B-A3B-Instruct \
-  --input-wav input_16k_mono.wav \
-  --output-wav response.wav \
-  --output-text response.txt
-```
-
-The client dependency is `websockets`.
-
-## Protocol
-
-Messages in both directions are JSON text frames. Audio carried inside an
-event is base64-encoded raw PCM16.
-
-| Direction | Event | Purpose |
-|-----------|-------|---------|
-| Server to client | `session.created` | Confirms that the WebSocket connection is ready |
-| Client to server | `session.update` | Selects and validates the served model |
-| Client to server | `input_audio_buffer.commit` with `final: false` | Starts generation over the incoming stream |
-| Client to server | `input_audio_buffer.append` | Appends base64 PCM16 audio |
-| Client to server | `input_audio_buffer.commit` with `final: true` | Marks the end of input |
-| Server to client | `transcription.delta` | Carries incremental response text |
-| Server to client | `transcription.done` | Carries final text and token usage |
-| Server to client | `response.audio.delta` | Carries incremental PCM16 response audio |
-| Server to client | `response.audio.done` | Marks the end of response audio |
-| Server to client | `error` | Reports an invalid event, model, or audio payload |
-
-A minimal client sends events in this order:
-
-```json
-{"type":"session.update","model":"Qwen/Qwen3-Omni-30B-A3B-Instruct"}
-{"type":"input_audio_buffer.commit","final":false}
-{"type":"input_audio_buffer.append","audio":"<base64-pcm16>"}
-{"type":"input_audio_buffer.commit","final":true}
-```
-
-The initial non-final commit intentionally starts generation before all audio
-has arrived. Continue sending `input_audio_buffer.append` events while the
-engine is consuming the stream.
-
-## Audio Handling
-
- Input is mono PCM16 at 16 kHz for the Qwen3-Omni example.
- `response.audio.delta.audio` contains base64-encoded PCM16 bytes.
- Read `sample_rate_hz` from each audio event instead of assuming an output
-  rate. Qwen3-Omni output is typically 24 kHz.
- Concatenate audio deltas in receive order to construct the output waveform.
- Set `input_audio_buffer.commit.final` to `true` after the last input chunk so
-  the server can terminate the streaming request.
-
-## Availability and Limitations
-
-The path is registered on the unified API server, but it is usable only when
-the loaded pipeline implements realtime audio input and produces compatible
-audio output. Unsupported deployments return an `error` event. The endpoint
-does not provide duplex session resume, playback acknowledgement, overlap
-policy, or barge-in controls.
-
-See the [Qwen3-Omni online serving example](https://github.com/vllm-project/vllm-omni/tree/main/examples/online_serving/qwen3_omni)
-for concurrency options, chunk pacing, and per-delta audio debugging.
@@ -2,7 +2,7 @@

 vLLM-Omni provides an OpenAI-compatible API for text-to-speech (TTS) generation. Supported TTS models include:

- **Qwen3-TTS** (`Qwen/Qwen3-TTS-12Hz-*`) -- Qwen3-based TTS with CustomVoice, VoiceDesign, and Base (voice cloning) task types. Native output: 24 kHz; API output can be resampled to 8 kHz.
+- **Qwen3-TTS** (`Qwen/Qwen3-TTS-12Hz-*`) -- Qwen3-based TTS with CustomVoice, VoiceDesign, and Base (voice cloning) task types. Output: 24 kHz.
 - **Fish Speech S2 Pro** (`fishaudio/s2-pro`) -- Dual-AR TTS with DAC codec. Supports text-to-speech and voice cloning via reference audio. Output: 44.1 kHz.
 - **Voxtral TTS** (`mistralai/Voxtral-4B-TTS-2603`) -- AR + FlowMatching TTS with preset voices. Output: 24 kHz.
 - **CosyVoice3** (`FunAudioLLM/Fun-CosyVoice3-0.5B-2512`) -- 2-stage talker + flow-matching code2wav. Voice cloning via `ref_audio` + `ref_text` (no presets). Output: 24 kHz.
@@ -116,7 +116,6 @@ Content-Type: application/json

 | Parameter | Type | Default | Description |
 |-----------|------|---------|-------------|
-| `sample_rate` | integer | model native | Target output sample rate. Qwen3-TTS supports 8000 or 24000 Hz; the model remains native 24 kHz internally and is resampled before encoding. |
 | `task_type` | string | "CustomVoice" | TTS task type: CustomVoice, VoiceDesign, or Base |
 | `language` | string | "Auto" | Language (see supported languages below) |
 | `instructions` | string | "" | Voice style/emotion instructions |
@@ -1,109 +0,0 @@
-# Standalone Experimental Servers
-
-Most vLLM-Omni endpoints run on the unified server started with
-`vllm serve <model> --omni`. The experimental full-duplex package also contains
-two separate user-facing server processes. Their routes are not added to the
-unified API server.
-
-| Server | Transport | Role |
-|--------|-----------|------|
-| PersonaPlex compatibility server | WebSocket | Hosts the official PersonaPlex browser protocol and a raw-PCM alternative |
-| JoyVL interaction server | HTTP | Adds state, memory, proactive decisions, and delegation in front of a separate OpenAI-compatible model backend |
-
-## PersonaPlex Compatibility Server
-
-Use this server when a client requires the official Moshi/PersonaPlex wire
-protocol or the bundled browser UI. For the engine-owned unified duplex path,
-use [`WS /v1/realtime?duplex=1`](full_duplex_api.md) instead.
-
-Start it with:
-
-```bash
-HF_TOKEN=... CUDA_VISIBLE_DEVICES=0 python -m \
-  vllm_omni.experimental.fullduplex.personaplex.serving.server \
-  --port 8091
-```
-
-| Route | Purpose |
-|-------|---------|
-| `GET /` | Official PersonaPlex browser client, when its assets are available |
-| `GET /health` | Readiness check |
-| `WS /api/chat` | Moshi-compatible tagged binary protocol with Opus audio |
-| `WS /v1/audio/duplex` | Simpler float32 raw-PCM protocol |
-
-### `/api/chat` Protocol
-
-The first byte of every binary frame is a message tag:
-
-| Direction | Tag | Payload |
-|-----------|-----|---------|
-| Server to client | `0x00` | Ready handshake |
-| Client to server | `0x01` | Opus microphone audio at 24 kHz |
-| Server to client | `0x01` | Opus response audio |
-| Server to client | `0x02` | UTF-8 inner-monologue text |
-
-The optional `text_prompt` and `voice_prompt` query parameters select a persona
-and voice. The default server admits one conversation; use `--batch-size N` to
-configure multiple lockstep slots. Connections beyond capacity close with
-WebSocket code 1013.
-
-### `/v1/audio/duplex` Protocol
-
-Send `{"type":"open","persona":"...","voice":"..."}` as a JSON text frame,
-then exchange binary float32 PCM frames. The server returns generated float32
-audio as binary frames and text as `{"type":"text","text":"..."}`. Send
-`{"type":"close"}` and wait for `{"type":"done"}` to flush and finish.
-
-See the [PersonaPlex example](https://github.com/vllm-project/vllm-omni/tree/main/examples/online_serving/personaplex)
-for the browser and headless clients.
-
-## JoyVL Interaction Server
-
-JoyVL is an orchestration layer, not a model-serving engine. It calls a
-separate OpenAI-compatible backend and adds per-session frame history, memory,
-persona policy, and optional delegation.
-
-Start the model backend and orchestrator separately:
-
-```bash
-vllm serve jdopensource/JoyAI-VL-Interaction-Preview \
-  --served-model-name JoyAI-VL-Interaction-Preview \
-  --port 8092 \
-  --max-model-len 131072 \
-  --enable-prefix-caching \
-  --limit-mm-per-prompt '{"image":256,"video":1}'
-
-python -m vllm_omni.experimental.fullduplex.joyvl.serving.server \
-  --port 8091 \
-  --main-backend-url http://127.0.0.1:8092/v1 \
-  --main-model JoyAI-VL-Interaction-Preview
-```
-
-| Method and route | Purpose |
-|------------------|---------|
-| `GET /health` | Readiness check |
-| `GET /v1/models` | Reports the configured interaction model |
-| `POST /v1/chat/completions` | Processes one frame or interaction tick |
-| `POST /reset` | Resets a session |
-| `POST /v1/streaming/reset` | Alias for `/reset` |
-| `POST /v1/streaming/persona` | Changes the session persona |
-
-Send one multimodal Chat Completions request per video frame, normally around
-one frame per second, and identify the session with `x-session-id`. Responses
-include an `interaction` block whose action is `silence`, `response`, or
-`delegate`.
-
-```bash
-curl http://127.0.0.1:8091/v1/chat/completions \
-  -H 'content-type: application/json' \
-  -H 'x-session-id: demo' \
-  -d '{
-    "messages": [{"role": "user", "content": [
-      {"type": "text", "text": "Alert me if a fire appears"},
-      {"type": "image_url", "image_url": {"url": "data:image/jpeg;base64,..."}}
-    ]}]
-  }'
-```
-
-See the [JoyVL recipe](https://github.com/vllm-project/vllm-omni/blob/main/recipes/JD/JoyAI-VL-Interaction.md)
-for memory, personas, delegation backends, and the browser UI.
@@ -1,92 +0,0 @@
-# Streaming Video Output WebSocket API
-
-Use `WS /v1/realtime/video` to receive fragmented MP4 chunks while a compatible
-diffusion video model is generating. This endpoint streams generated video
-output; it is unrelated to [`WS /v1/video/chat/stream`](video_stream_api.md),
-which streams live video input into a conversational model.
-
-## Quick Start
-
-Start a model that supports step execution and enable diffusion streaming
-output:
-
-```bash
-vllm serve BestWishYsh/Helios-Distilled --omni \
-  --diffusion-streaming-output \
-  --port 8091
-```
-
-Install the client dependency and run the provided client:
-
-```bash
-pip install av websockets
-
-python examples/online_serving/streaming_video_generation/streaming_video_client.py \
-  --host 127.0.0.1 \
-  --port 8091 \
-  --model BestWishYsh/Helios-Distilled \
-  --prompt "A serene lakeside sunrise with mist over the water" \
-  --output output.mp4
-```
-
-## Protocol
-
-Control events are JSON text frames. Each `video.chunk_metadata` event is
-immediately followed by one binary WebSocket frame containing fragmented MP4
-bytes.
-
-| Direction | Event or frame | Purpose |
-|-----------|----------------|---------|
-| Client to server | `session.start` | Starts generation with a prompt and video parameters |
-| Server to client | `video.start` | Confirms the request ID, format, and accepted configuration |
-| Server to client | `video.chunk_metadata` | Describes the next binary media frame |
-| Server to client | Binary frame | Carries one fragmented MP4 chunk |
-| Client to server | `session.interaction` | Queues an optional prompt update during generation |
-| Server to client | `session.interaction.queued` | Confirms that an interaction was queued |
-| Client to server | `session.ping` | Refreshes the stall clock |
-| Server to client | `session.pong` | Acknowledges a ping |
-| Client to server | `session.stop` | Cancels the active generation |
-| Server to client | `session.done` | Reports completion or cancellation and the chunk count |
-| Server to client | `error` | Reports validation, generation, or timeout failures |
-
-Start a minimal session with:
-
-```json
-{
-  "type": "session.start",
-  "model": "BestWishYsh/Helios-Distilled",
-  "prompt": "A serene lakeside sunrise",
-  "format": "m4s",
-  "width": 640,
-  "height": 384,
-  "fps": 16,
-  "num_frames": 99
-}
-```
-
-Only `m4s` output is currently accepted. Concatenate the binary frames in
-receive order, then remux the fragmented stream if the target player requires
-a progressive MP4 file.
-
-## Mid-Generation Interaction
-
-Models that implement interaction updates can change the active prompt:
-
-```json
-{
-  "type": "session.interaction",
-  "interaction": {
-    "event_id": "sunlight-update",
-    "event": {"prompt": "Sunlight breaks through the mist"},
-    "transition_chunks": 2
-  }
-}
-```
-
-This feature is pipeline-dependent. An enabled endpoint does not imply that
-the loaded model supports midway prompt updates.
-
-The default start timeout is 10 seconds and the server reports a stall after
-about 60 seconds without generation progress or a `session.ping`. See the
-[streaming video generation example](https://github.com/vllm-project/vllm-omni/tree/main/examples/online_serving/streaming_video_generation)
-for prompt-update scheduling and browser playback.
@@ -17,8 +17,6 @@ globally.
 | --- | --- | --- |
 | `sparsity` | finite, `[0, 1]` | Nominal dropped-key-block fraction; default `0.8`; `0` disables sparsity |
 | `start_step` | integer, `>= 0` | Number of early denoise steps kept dense |
-| `end_step` | integer, `>= 0` | Number of final denoise steps kept dense (tail fallback) |
-| `precision` | `"bf16"`, `"fp8"`, `"mix"` | Kernel precision mode; default `"bf16"`. Requires MindIE-SD with `sparse_attention(precision=...)` |
 | `skip_layers` | selector such as `"0-3,38"` | DiT blocks kept dense |

 ```bash
@@ -54,12 +52,6 @@ platform raises. It is incompatible with ring sequence parallelism because
 the kernel needs the complete key sequence for ranking; use Ulysses sequence
 parallelism with `ring_degree=1`.

-The `precision` knob requires a MindIE-SD release whose `sparse_attention`
-accepts `precision=`; older releases accept it through `**kwargs` but
-silently ignore it and run the BF16 path. `RAINFUSION_ATTN` raises at runtime
-when a non-`bf16` precision is requested without that support, so pin a
-compatible MindIE-SD release when using `precision="mix"` or `"fp8"`.
-
 ## Geometry handling

 RainFusion handles arbitrary video grids by rearranging video spatially and,
@@ -12,17 +12,15 @@ LoRA adapters are lightweight, model-specific fine-tuning weights that can be ap
 ## LoRA Adapter Format

 ### PEFT (Parameter-Efficient Fine-Tuning) format (default)
-
 A typical PEFT-format LoRA adapter directory structure:

-```text
+```
 lora_adapter/
 ├── adapter_config.json
 └── adapter_model.safetensors
 ```

 The `adapter_config.json` file contains metadata about the LoRA adapter, including:
-
 - `r`: LoRA rank
 - `lora_alpha`: LoRA alpha scaling factor
 - `target_modules`: List of module names to apply LoRA to
@@ -69,7 +67,7 @@ For distilled few-step LoRAs, pass `lora_backend="distill"` together with one or
 For Qwen-Image and Wan pipelines, the distill backend calls `pipeline.load_lora_weights(...)` during worker initialization.

 | Pipeline | Supported distilled LoRA repo | Notes |
-| ---------- | -------------------------------- | ------- |
+|----------|--------------------------------|-------|
 | `QwenImagePipeline` | `lightx2v/Qwen-Image-2512-Lightning` | Used with Qwen-Image-2512 Lightning-style few-step inference. |
 | `Wan22Pipeline` | `lightx2v/Wan2.1-Distill-Loras`, `lightx2v/Wan2.2-Distill-Loras` | Wan2.1 uses one LoRA file. For dual-transformer Wan2.2 MoE, pass high-noise then low-noise LoRA files. |
 | `Wan22I2VPipeline` | `lightx2v/Wan2.2-Distill-Loras` | For dual-transformer Wan2.2 MoE, pass high-noise then low-noise LoRA files. |
@@ -242,37 +240,6 @@ Notes:
 - Output quality and speed depend on the replacement checkpoints and sampling params you choose.
 - If you only need to fuse distilled LoRAs into a Wan2.2 checkpoint at load time (without the full LightX2V convert + assemble pipeline), you can instead pass them directly via `--lora-backend distill --lora-path <high>.safetensors <low>.safetensors`. See the [Distill backend](#distill-backend-fuse-distilled-lora-at-init) section above.

-## MiniMax-H3 adapter-declared schedules
-
-MiniMax-H3 supports two few-step mechanisms that must not be conflated:
-
- **Checkpoint-pinned schedule**: a merged release writes `base_schedule` into
-  `model_index.json`. Requests must pass `num_inference_steps` as the interval
-  count (for example `4` for `[1.0, 0.7, 0.4, 0.15, 0.0]`).
- **Runtime Turbo LoRA**: LightX2V Turbo artifacts request five sigma points and
-  enforce `flow_shift=6`, `audio_flow_shift=3`.
- **Runtime native LoRA**: FlashGen-style artifacts declare
-  `key_format=minimax-h3-native` and embed `base_schedule` in safetensors
-  metadata. When active, the adapter schedule overrides the base checkpoint and
-  requests must use the interval-count contract (`num_inference_steps=4`).
-  Request-mode generation may omit the field and take the count from the adapter
-  schedule; step execution requires it explicitly, because the step scheduler
-  derives the total step count from the request before the adapter schedule is
-  known.
-
-Native artifacts also declare `qkv_layout=grouped`. The H3 loader reorders fused
-`qkv_proj` LoRA rows with the same `_reorder_grouped_qkv_to_qkv` helper used for
-base-weight loading, then binds the packed Q/K/V slices through the legacy PEFT
-manager without modifying `DiffusionLoRAManager`.
-
-Both runtime adapters run with distributed layerwise offload in request-mode
-generation, where the manager keeps the LoRA A/B buffers resident on the compute
-device while DLO streams the base blocks. MiniMax-H3 step execution still
-rejects DLO, so `--step-execution` cannot be combined with
-`--enable-distributed-layerwise-offload`. Model-level CPU offload and standard
-layerwise offload are rejected in every mode, because the dynamic LoRA tensors
-are neither parameters nor registered buffers and therefore do not participate
-in those weight lifecycles.

 ## See Also

@@ -138,7 +138,7 @@ The following tables show which models support each feature:
 | **Qwen-Image-Edit**      |     ✅     |     ✅      |           ✅           |       ✅        |         ✅         |          ❌          |   ✅    |             ✅             |      ✅ (decode)      |       ❌        |        ❌         |
 | **Qwen-Image-Edit-2509** |     ✅     |     ✅      |           ✅           |       ✅        |         ✅         |          ❌          |   ✅    |        ✅ (decode)         |          ✅           |       ❌        |        ❌         |
 | **Qwen-Image-Layered**   |     ✅     |     ✅      |           ✅           |       ✅        |         ✅         |          ❌          |   ✅    |             ✅             |      ✅ (decode)      |       ❌        |        ❌         |
-| **SenseNova-U1 / U1.5**  |     ❌     |     ✅      |           ❌           |       ✅        |         ✅         |          ❌          |   ❌    |             ✅             |          ❌           |       ❌        |        ❌         |
+| **SenseNova-U1**         |     ❌     |     ✅      |           ❌           |       ✅        |         ✅         |          ❌          |   ❌    |             ✅             |          ❌           |       ❌        |        ❌         |
 | **Stable-Diffusion-XL**  |     ❌     |     ❌      |           ✅           |       ✅        |         ✅         |          ❌          |   ✅    |             ✅             |      ✅ (decode)      |       ❌        |        ❌         |
 | **Stable-Diffusion3.5**  |     ❌     |     ✅      |           ❌           |       ✅        |         ✅         |          ❌          |   ❌    |             ✅             |      ✅ (decode)      |       ❌        |        ❌         |
 | **Z-Image**              |     ✅     |     ✅      |           ✅           |       ❓        |   ✅ (TP=2 only)   |          ❌          |   ✅    |             ❌             |      ✅ (decode)      |       ✅        |        ❌         |
@@ -1,5 +1,5 @@
 # SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM project

 """
 Image-to-Video generation example using Wan2.2 I2V/TI2V models, LTX2/LTX-2.3,
@@ -111,13 +111,6 @@ def build_image_to_video_prompt(
    return result


-def _validate_video_output_type(output_type: str) -> None:
-    if output_type not in {"image", "video"}:
-        raise ValueError(
-            f"Unexpected output type '{output_type}', expected 'video' or legacy 'image' for video generation."
-        )
-
-
 def parse_args() -> argparse.Namespace:
    parser = argparse.ArgumentParser(
        description=(
@@ -672,7 +665,10 @@ def main():
        frames = frames[0] if frames else None

    if isinstance(frames, OmniRequestOutput):
-        _validate_video_output_type(frames.final_output_type)
+        if frames.final_output_type != "image":
+            raise ValueError(
+                f"Unexpected output type '{frames.final_output_type}', expected 'image' for video generation."
+            )
        if frames.multimodal_output and "audio" in frames.multimodal_output:
            audio = frames.multimodal_output["audio"]
            audio_sample_rate = frames.multimodal_output.get("audio_sample_rate", audio_sample_rate)
@@ -1,5 +1,5 @@
 # SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM project

 import argparse
 import json
@@ -174,13 +174,6 @@ def build_text_to_video_prompt(prompt: str, negative_prompt: str | None) -> dict
    return result


-def _validate_video_output_type(output_type: str) -> None:
-    if output_type not in {"image", "video"}:
-        raise ValueError(
-            f"Unexpected output type '{output_type}', expected 'video' or legacy 'image' for video generation."
-        )
-
-
 def _normalize_float_tensor(tensor: torch.Tensor, source_range: str) -> torch.Tensor:
    """Normalize decoded frames according to the pipeline's output contract."""
    if not tensor.is_floating_point():
@@ -627,7 +620,10 @@ def main():
        frames = frames[0] if frames else None

    if isinstance(frames, OmniRequestOutput):
-        _validate_video_output_type(frames.final_output_type)
+        if frames.final_output_type != "image":
+            raise ValueError(
+                f"Unexpected output type '{frames.final_output_type}', expected 'image' for video generation."
+            )
        if frames.multimodal_output and "audio" in frames.multimodal_output:
            audio = frames.multimodal_output["audio"]
            audio_sample_rate = frames.multimodal_output.get("audio_sample_rate", audio_sample_rate)
@@ -49,7 +49,7 @@ scenario below for live barge-in validation on the target GPU.
 Default:

 ```bash
-vllm serve openbmb/MiniCPM-o-4_5 \
+vllm-omni serve openbmb/MiniCPM-o-4_5 \
    --omni \
    --deploy-config vllm_omni/deploy/minicpmo_4_5.yaml \
    --trust-remote-code \
@@ -159,33 +159,6 @@ python examples/online_serving/minicpmo/realtime_duplex_demo.py \
    --output-dir /tmp/minicpmo_realtime_duplex_demo
 ```

-Video input uses the same session. PyAV demuxes JPEG frames at `--video-fps`
-(default 1.0) and vLLM `load_audio` extracts a 16 kHz mono WAV unless
-`--input-wav` overrides the soundtrack. Frames keep their capture size
-(`--frame-max-side 0`); the server `process_image` normalizes at 448.
-
-`--stack-frames N` raises the visual refresh rate the way official duplex does:
-each 1 s unit also samples the `N-1` sub-frames captured inside it and tiles
-them into a single composite image sent next to that unit's base frame. The
-audio cadence never changes — a unit is always one second — and the wire always
-carries 2 images per unit however large `N` is, because the sub-frames share one
-composite. Official uses 5 for high refresh rate mode:
-
-```bash
-python examples/online_serving/minicpmo/realtime_duplex_demo.py \
-    --url ws://localhost:8099/v1/realtime?duplex=1 \
-    --model openbmb/MiniCPM-o-4_5 \
-    --input-video /path/to/clip.mp4 \
-    --stack-frames 5 \
-    --ref-audio /path/to/MiniCPM-o-Demo/assets/ref_audio/ref_minicpm_signature.wav \
-    --output-dir /tmp/minicpmo_realtime_duplex_video_demo
-```
-
-Detail inside a composite is capped by `scale_resolution=448` at
-`max_slice_nums=1`: official suggests HD slicing (`max_slice_nums=[2, 1]`) for
-stacked frames, which the duplex adapter does not implement yet. Reading small
-text or digits out of a wide scene is limited by that, not by frame timing.
-
 ## Open the experimental browser client

 The browser UI serves the page and proxies the same-origin Realtime WebSocket to
@@ -12,9 +12,7 @@ from __future__ import annotations
 import argparse
 import asyncio
 import base64
-import io
 import json
-import math
 import sys
 import time
 from pathlib import Path
@@ -24,8 +22,8 @@ if str(REPO_ROOT) not in sys.path:
    sys.path.insert(0, str(REPO_ROOT))

 from vllm_omni.experimental.fullduplex.client import (  # noqa: E402
-    PCM16_BYTES_PER_SAMPLE,
-    PCM16_SAMPLE_RATE,
+    PCM16_BYTES_PER_SAMPLE,  # noqa: F401 - compatibility export
+    PCM16_SAMPLE_RATE,  # noqa: F401 - compatibility export
    RealtimeDuplexClient,
    RealtimeEventCollector,
    build_realtime_url,
@@ -40,10 +38,6 @@ from vllm_omni.experimental.fullduplex.client import (  # noqa: E402
 from vllm_omni.experimental.fullduplex.client import (  # noqa: E402
    reference_audio_data_url as _ref_audio_data_url,
 )
-from vllm_omni.experimental.fullduplex.video_stacking import (  # noqa: E402
-    concat_frames_b64,
-    unit_subframe_offsets,
-)


 class _StreamingOutputWriter:
@@ -155,23 +149,6 @@ def _response_in_progress(events: list[dict[str, object]]) -> bool:
    )


-async def _commit_input_after_playback_checkpoint(client: RealtimeDuplexClient) -> float:
-    """Commit input only after preserving already-played assistant history.
-
-    Native duplex can finish several responses while the user is still
-    streaming.  Under the ``ack_only`` playback policy, acknowledging those
-    responses after the final input commit is intentionally rejected: it could
-    append an old assistant message after the newly committed user message.
-    Checkpoint playback first so the server commits each assistant item at its
-    original history position.  A second acknowledgement after the commit can
-    then safely advance response-local playback that drained in the meantime.
-    """
-    await client.acknowledge_playback()
-    commit_sent_at_s = time.monotonic()
-    await client.commit()
-    return commit_sent_at_s
-
-
 def _event_count_after(
    events: list[dict[str, object]],
    event_type: str,
@@ -182,258 +159,14 @@ def _event_count_after(
    return sum(event.get("type") == event_type for event in events[index + 1 :])


-def _probe_video_frames_and_fps(video_path: Path) -> tuple[int, float]:
-    """Frame count and average FPS via PyAV (same approach as Daily-Omni)."""
-    import av
-
-    with av.open(str(video_path)) as container:
-        stream = container.streams.video[0]
-        avg_fps = float(stream.average_rate) if stream.average_rate else 0.0
-        num_frames = stream.frames or 0
-        if num_frames <= 0:
-            duration: float | None = None
-            if stream.duration is not None and stream.time_base is not None:
-                duration = float(stream.duration * stream.time_base)
-            elif container.duration is not None:
-                duration = container.duration / av.time_base
-            if duration and avg_fps:
-                num_frames = int(math.floor(duration * avg_fps))
-        if num_frames <= 0:
-            stream.thread_type = "AUTO"
-            num_frames = sum(1 for _ in container.decode(stream))
-    if num_frames <= 0:
-        raise ValueError(f"No decodable video frames in {video_path}")
-    return num_frames, avg_fps or 1.0
-
-
-def _upright(image, rotation: int):
-    """Apply a container display-matrix rotation, in degrees, to ``image``.
-
-    PyAV hands back the stored raster and leaves the display matrix for the
-    caller, unlike the ffmpeg CLI which autorotates. Phone clips routinely
-    carry +/-90, so skipping this feeds the model a sideways frame: arrows and
-    digits stop being readable and the model mirrors the motion it reports.
-    """
-    from PIL import Image
-
-    quarter_turns = {
-        90: Image.Transpose.ROTATE_90,
-        180: Image.Transpose.ROTATE_180,
-        270: Image.Transpose.ROTATE_270,
-    }
-    turn = quarter_turns.get(int(rotation) % 360)
-    return image.transpose(turn) if turn is not None else image
-
-
-def _decode_video_frames_rgb(video_path: Path, frame_idx: list[int]) -> list:
-    """Decode presentation-order ``frame_idx`` into upright RGB ``PIL.Image``s."""
-    import av
-
-    wanted = sorted({int(i) for i in frame_idx})
-    if not wanted:
-        return []
-
-    decoded: dict[int, object] = {}
-    with av.open(str(video_path)) as container:
-        stream = container.streams.video[0]
-        stream.thread_type = "AUTO"
-        cursor = 0
-        for position, frame in enumerate(container.decode(stream)):
-            if wanted[cursor] > position:
-                continue
-            image = _upright(frame.to_image(), getattr(frame, "rotation", 0) or 0)
-            while cursor < len(wanted) and wanted[cursor] <= position:
-                decoded[wanted[cursor]] = image
-                cursor += 1
-            if cursor >= len(wanted):
-                break
-
-    if not decoded:
-        raise ValueError(f"Decoded no frames from {video_path} for indices {wanted[:8]}")
-    last_image = decoded[max(decoded)]
-    return [decoded.get(int(i), last_image) for i in frame_idx]
-
-
-def _resize_max_side(image, max_side: int):
-    width, height = image.size
-    longest = max(width, height)
-    if longest <= max_side or max_side <= 0:
-        return image
-    scale = max_side / float(longest)
-    new_size = (max(1, int(round(width * scale))), max(1, int(round(height * scale))))
-    return image.resize(new_size)
-
-
-def _jpeg_bytes(image, *, quality: int = 95) -> bytes:
-    buf = io.BytesIO()
-    image.convert("RGB").save(buf, format="JPEG", quality=quality)
-    return buf.getvalue()
-
-
-def _to_pcm16_bytes(audio_np) -> bytes:
-    import numpy as np
-
-    arr = np.asarray(audio_np).reshape(-1)
-    if arr.dtype != np.int16:
-        arr = (np.clip(arr, -1.0, 1.0) * 32767.0).astype(np.int16)
-    return arr.tobytes()
-
-
-def _pcm16_wav_duration_s(path: Path) -> float:
-    pcm16 = read_pcm16_wav(path)
-    return len(pcm16) / float(PCM16_SAMPLE_RATE * PCM16_BYTES_PER_SAMPLE)
-
-
-def _extract_video_frames(
-    video_path: Path,
-    work_dir: Path,
-    *,
-    fps: float,
-    max_side: int,
-    duration_s: float,
-    stack_frames: int = 1,
-) -> tuple[list[str], list[str | None]]:
-    """Demux JPEG frames only. Frame count follows ``duration_s``, not soundtrack.
-
-    Returns the base frames (one per model unit) and the parallel stacked track.
-    With ``stack_frames=N`` each unit also samples its ``N-1`` interior
-    sub-frames and tiles them into one composite, which is how official duplex
-    raises the visual refresh rate while keeping one second of audio per unit.
-    A unit whose interior falls past the end of the clip stacks nothing.
-    """
-    num_frames, avg_fps = _probe_video_frames_and_fps(video_path)
-    sample_fps = max(float(fps), 1e-6)
-    unit_s = 1.0 / sample_fps
-    timeline_s = max(float(duration_s), unit_s)
-    num_samples = max(1, int(math.ceil(timeline_s * sample_fps - 1e-9)))
-    offsets = unit_subframe_offsets(stack_frames, unit_s=unit_s)
-
-    def frame_at(seconds: float) -> int:
-        # Official get_video_frame_audio_segments uses truncating int(), not
-        # round(), so 29.97 fps lands on the same presentation index.
-        return min(int(seconds * avg_fps), num_frames - 1)
-
-    base_idx = [frame_at(unit * unit_s) for unit in range(num_samples)]
-    sub_times = [
-        [unit * unit_s + offset for offset in offsets if unit * unit_s + offset < timeline_s]
-        for unit in range(num_samples)
-    ]
-    sub_idx = [frame_at(seconds) for times in sub_times for seconds in times]
-    images = _decode_video_frames_rgb(video_path, base_idx + sub_idx)
-    sub_images = iter(images[len(base_idx) :])
-
-    frame_dir = work_dir / "frames"
-    frame_dir.mkdir(parents=True, exist_ok=True)
-    for stale in frame_dir.glob("*.jpg"):
-        stale.unlink()
-    frames_b64: list[str] = []
-    stacked_b64: list[str | None] = []
-    for index, (image, times) in enumerate(zip(images[: len(base_idx)], sub_times), start=1):
-        jpeg = _jpeg_bytes(_resize_max_side(image, max_side))
-        (frame_dir / f"frame_{index:05d}.jpg").write_bytes(jpeg)
-        frames_b64.append(base64.b64encode(jpeg).decode("ascii"))
-        interior = [_resize_max_side(next(sub_images), max_side) for _ in times]
-        if not interior:
-            stacked_b64.append(None)
-            continue
-        composite = concat_frames_b64(interior)
-        (frame_dir / f"stack_{index:05d}.jpg").write_bytes(base64.b64decode(composite))
-        stacked_b64.append(composite)
-    return frames_b64, stacked_b64
-
-
-def _extract_video_soundtrack(video_path: Path, work_dir: Path) -> Path:
-    """Write a 16 kHz mono WAV from the video soundtrack."""
-    try:
-        from vllm.multimodal.media.audio import load_audio
-    except ImportError as exc:  # pragma: no cover
-        raise SystemExit("--input-video needs vLLM audio helpers (load_audio)") from exc
-
-    audio_np, sr = load_audio(str(video_path), sr=PCM16_SAMPLE_RATE, mono=True)
-    work_dir.mkdir(parents=True, exist_ok=True)
-    audio_path = work_dir / "input_16k.wav"
-    write_pcm16_wav(audio_path, _to_pcm16_bytes(audio_np), sample_rate_hz=int(sr))
-    return audio_path
-
-
-def _extract_video_input(
-    video_path: Path,
-    work_dir: Path,
-    *,
-    fps: float,
-    max_side: int,
-    stack_frames: int,
-) -> tuple[Path, list[str], list[str | None]]:
-    """Demux a video into duplex inputs: 16 kHz mono WAV plus JPEG frames.
-
-    Use this only when the video soundtrack is the audio source. When the
-    caller already has ``--input-wav``, extract frames against that WAV
-    duration instead so a silent clip cannot fail ``load_audio``.
-    """
-    audio_path = _extract_video_soundtrack(video_path, work_dir)
-    frames_b64, stacked_b64 = _extract_video_frames(
-        video_path,
-        work_dir,
-        fps=fps,
-        max_side=max_side,
-        duration_s=_pcm16_wav_duration_s(audio_path),
-        stack_frames=stack_frames,
-    )
-    return audio_path, frames_b64, stacked_b64
-
-
-def _resolve_duplex_av_inputs(
-    *,
-    input_wav: str | None,
-    input_video: str | None,
-    work_dir: Path,
-    fps: float,
-    max_side: int,
-    stack_frames: int = 1,
-) -> tuple[str, list[str], list[str | None]]:
-    """Return ``(wav_path, frames_b64, stacked_b64)``. External WAV wins over video audio."""
-    if input_video:
-        video_path = Path(input_video)
-        if input_wav:
-            frames, stacked = _extract_video_frames(
-                video_path,
-                work_dir,
-                fps=fps,
-                max_side=max_side,
-                duration_s=_pcm16_wav_duration_s(Path(input_wav)),
-                stack_frames=stack_frames,
-            )
-            return input_wav, frames, stacked
-        extracted_wav, frames, stacked = _extract_video_input(
-            video_path,
-            work_dir,
-            fps=fps,
-            max_side=max_side,
-            stack_frames=stack_frames,
-        )
-        return str(extracted_wav), frames, stacked
-    if not input_wav:
-        raise SystemExit("provide --input-wav, --input-video, or both")
-    return input_wav, [], []
-
-
 async def run_demo(args: argparse.Namespace) -> dict[str, object]:
-    output_dir = Path(args.output_dir)
-    stream_writer = _StreamingOutputWriter(output_dir)
-
-    input_wav, video_frames, stacked_video_frames = _resolve_duplex_av_inputs(
-        input_wav=args.input_wav,
-        input_video=args.input_video,
-        work_dir=output_dir / "video_input",
-        fps=args.video_fps,
-        max_side=args.frame_max_side,
-        stack_frames=args.stack_frames,
-    )
-
-    input_pcm16 = read_pcm16_wav(Path(input_wav))
+    input_pcm16 = read_pcm16_wav(Path(args.input_wav))
    if not input_pcm16:
        raise ValueError("input WAV has no audio")

+    output_dir = Path(args.output_dir)
+    stream_writer = _StreamingOutputWriter(output_dir)
+
    url = build_realtime_url(
        args.url,
        args.model,
@@ -451,12 +184,10 @@ async def run_demo(args: argparse.Namespace) -> dict[str, object]:
            timeout_s=args.timeout_s,
        )
        stream_event_cursor = len(client.events.events)
-        frames_sent = await client.stream_pcm16(
+        await client.stream_pcm16(
            input_pcm16,
            chunk_ms=args.chunk_ms,
            realtime=not args.no_realtime_pacing,
-            video_frames=video_frames,
-            stacked_video_frames=stacked_video_frames,
        )
        commit_event_cursor = len(client.events.events)
        stream_decision = _latest_model_decision(client.events.events, stream_event_cursor)
@@ -465,7 +196,8 @@ async def run_demo(args: argparse.Namespace) -> dict[str, object]:
            chunk_period_ms=_chunk_period_ms(client.events.events),
        )
        wait_for_post_commit_decision = False
-        commit_sent_at_s = await _commit_input_after_playback_checkpoint(client)
+        commit_sent_at_s = time.monotonic()
+        await client.commit()
        wait_error: str | None = None
        committed_index: int | None = None
        post_commit_decision: str | None = None
@@ -578,10 +310,6 @@ async def run_demo(args: argparse.Namespace) -> dict[str, object]:
                    committed_index,
                ),
            },
-            "input_video": args.input_video,
-            "video_frames_extracted": len(video_frames),
-            "video_frames_sent": frames_sent,
-            "video_stacked_frames_extracted": sum(frame is not None for frame in stacked_video_frames),
            "audio_bytes": len(audio),
            "audio_chunk_count": len(stream_writer.audio_chunk_paths),
            "audio_chunk_files": [str(path) for path in stream_writer.audio_chunk_paths],
@@ -628,48 +356,7 @@ def parse_args() -> argparse.Namespace:
    parser.add_argument("--url", default="ws://localhost:8099/v1/realtime?duplex=1")
    parser.add_argument("--model", default="openbmb/MiniCPM-o-4_5")
    parser.add_argument("--session-id")
-    parser.add_argument(
-        "--input-wav",
-        help="Mono 16 kHz PCM16 clip to stream. Optional when --input-video carries the audio track.",
-    )
-    parser.add_argument(
-        "--input-video",
-        help=(
-            "Video to stream as omni duplex input. PyAV demuxes frames and "
-            "vLLM load_audio extracts a 16 kHz mono WAV (unless --input-wav overrides the audio)."
-        ),
-    )
-    parser.add_argument(
-        "--video-fps",
-        type=float,
-        default=1.0,
-        help="Frame extraction rate. The wire contract carries ~1 frame per second of audio.",
-    )
-    parser.add_argument(
-        "--frame-max-side",
-        type=int,
-        default=0,
-        help=(
-            "Optional client-side downscale of the longest side, in pixels. "
-            "0 (default) keeps the capture resolution so the server "
-            "process_image can normalize at scale_resolution=448 from a "
-            "higher-quality source. The web demo never resizes on the client."
-        ),
-    )
-    parser.add_argument(
-        "--stack-frames",
-        type=int,
-        default=1,
-        help=(
-            "Frames sampled per 1 s audio unit. 1 (default) sends the unit's "
-            "frame alone. N > 1 also samples the N-1 sub-frames inside the unit "
-            "and tiles them into one composite sent next to the base frame, so "
-            "motion (arrows, doors, digits changing) is visible at N fps while "
-            "the audio stays 1 s per unit. Official uses 5 for high refresh "
-            "rate mode; the wire carries 2 images per unit at any N because the "
-            "sub-frames share one composite."
-        ),
-    )
+    parser.add_argument("--input-wav", required=True)
    parser.add_argument(
        "--ref-audio",
        required=True,
@@ -685,7 +372,7 @@ def parse_args() -> argparse.Namespace:
        "--temperature",
        type=float,
        default=None,
-        help="Stage 0 sampling temperature; omit to preserve the model default.",
+        help="Stage0 sampling temperature; response-required soft-interrupt uses 0.0.",
    )
    parser.add_argument("--no-realtime-pacing", action="store_true")
    parser.add_argument("--require-audio", action="store_true")
@@ -37,7 +37,7 @@ During generation the client normally sends only `session.start` and then receiv
 ## Install Client Dependency

 ```bash
-pip install av websockets
+pip install websockets
 # For the Gradio demo:
 pip install 'vllm-omni[demo]' websockets
 ```
@@ -8,7 +8,7 @@ the received bytes to disk when the session finishes or is interrupted.
 Image/reference input is intentionally not supported in this example yet.

 Requirements:
-    pip install av websockets
+    pip install websockets
 """

 import argparse
@@ -1,6 +1,3 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
-
 """OpenAI-compatible client for Qwen3-TTS via /v1/audio/speech endpoint.

 This script demonstrates how to use the OpenAI-compatible speech API
@@ -77,8 +74,6 @@ def run_tts_generation(args) -> None:
        "voice": args.speaker,
        "response_format": args.response_format,
    }
-    if args.sample_rate is not None:
-        payload["sample_rate"] = args.sample_rate

    # Add optional parameters
    if args.instructions:
@@ -251,13 +246,6 @@ def parse_args():
        choices=["wav", "mp3", "flac", "pcm", "aac", "opus"],
        help="Audio output format (default: wav)",
    )
-    parser.add_argument(
-        "--sample-rate",
-        type=int,
-        default=None,
-        choices=[8000, 24000],
-        help="Output sample rate in Hz (default: model native)",
-    )
    parser.add_argument(
        "--output",
        "-o",
@@ -71,11 +71,11 @@ dev = [
    "voxcpm==2.0.3", # for VoxCPM2 path-locating in tests
    "torchdiffeq==0.2.5", # for covo_audio Test
    "FlagEmbedding==1.4.0", # for hunyuan_image3 tests
+    "kernels==0.14.1", # for accuracy tests comparison with diffusers output (0.15.0 incompatible with diffusers 0.38.0)
    "s3tokenizer==0.3.0", # for tts test
    # Provides top-level cosyvoice2; do not co-install stepaudio2-minicpmo,
    # whose top-level stepaudio2 package conflicts with this distribution.
    "step-audio2==1.0.0", # Step-Audio2 / MiniCPM-o 4.5 Token2Wav (flashcosyvoice / hyperpyyaml)
-    "silero-vad==6.1", # Realtime server-VAD duplex E2E
 ]

 demo = [
@@ -85,10 +85,6 @@ demo = [
    "cn2an>=0.5.22",  # Chinese number normalization for TTS (e.g. "1次" -> "一次")
 ]

-server-vad = [
-    "silero-vad==6.1",
-]
-
 forced-aligner = [
    "qwen-asr>=0.0.6",  # faithful Japanese/Korean word segmentation for the TTS forced aligner
 ]
@@ -248,11 +244,11 @@ addopts = [
    "--dist=loadgroup"
 ]
 markers = [
-    # ci/cd required — tag with [ci-level] for tests.helpers.mark.get_level_markers().
-    "core_model: [ci-level] L1&L2 tests (run in each PR)",
-    "advanced_model: [ci-level] L3 level tests (run on each merge)",
-    "full_model: [ci-level] L4 level tests (run nightly)",
-    "local_model: [ci-level] tests (run locally)",
+    # ci/cd required
+    "core_model: L1&L2 tests (run in each PR)",
+    "advanced_model: L3 level tests (run on each merge)",
+    "full_model: L4 level tests (run nightly)",
+    "local_model: tests (run locally)",
    # function module markers
    "diffusion: Diffusion model tests",
    "omni: Omni model tests",
@@ -262,41 +258,38 @@ markers = [
    "sp: Sequence parallelism tests",
    "example: Doc example code tests",
    # platform markers
-    # Tag with [hardware-platform] for tests.helpers.mark.get_supported_platforms().
-    # Add [gpu] on accelerator platforms that should also get pytest.mark.gpu.
-    # cpu/gpu are pytest marks; hardware_test(res=...) rejects them as keys.
-    "cpu: [hardware-platform] Tests that run on CPU",
-    "gpu: [hardware-platform] Tests that run on GPU (auto-added)",
-    "cuda: [hardware-platform] [gpu] Tests that run on CUDA (auto-added)",
-    "rocm: [hardware-platform] [gpu] Tests that run on AMD/ROCm (auto-added)",
-    "xpu: [hardware-platform] [gpu] Tests that run on XPU (auto-added)",
-    "npu: [hardware-platform] Tests that run on NPU/Ascend (auto-added)",
-    "musa: [hardware-platform] [gpu] Tests that run on MUSA/Moore Threads (auto-added)",
+    "cpu: Tests that run on CPU",
+    "gpu: Tests that run on GPU (auto-added)",
+    "cuda: Tests that run on CUDA (auto-added)",
+    "rocm: Tests that run on AMD/ROCm (auto-added)",
+    "xpu: Tests that run on XPU (auto-added)",
+    "npu: Tests that run on NPU/Ascend (auto-added)",
+    "musa: Tests that run on MUSA/Moore Threads (auto-added)",
    # specified computation resources marks (auto-added)
-    # [hardware-resource] + [cuda]/[rocm]/[xpu]/[npu]/[musa] for hardware_test(res=...).
-    # Do not put [gpu] on SKU lines: that would make gpu a res key.
-    "H100: [hardware-resource] [cuda] Tests that require H100 GPU",
-    "H800: [hardware-resource] [cuda] Tests that require H800 GPU",
-    "H200: [hardware-resource] [cuda] Tests that require H200 GPU",
-    "L4: [hardware-resource] [cuda] Tests that require L4 GPU",
-    "B200: [hardware-resource] [cuda] Tests that require B200 GPU",
-    "MI325: [hardware-resource] [rocm] Tests that require MI325 GPU (AMD/ROCm)",
-    "B60: [hardware-resource] [xpu] Tests that require Intel Arc Pro B60 XPU",
-    "S5000: [hardware-resource] [musa] Tests that require S5000 GPU (Moore Threads/MUSA)",
-    "A2: [hardware-resource] [npu] Tests that require A2 NPU",
-    "A3: [hardware-resource] [npu] Tests that require A3 NPU",
-    # Tag with [hardware-cards] for tests.helpers.mark.get_supported_card_counts().
-    "cards_1: [hardware-cards] Tests that require 1 accelerator card",
-    "cards_2: [hardware-cards] Tests that require 2 accelerator cards",
-    "cards_3: [hardware-cards] Tests that require 3 accelerator cards",
-    "cards_4: [hardware-cards] Tests that require 4 accelerator cards",
-    "cards_5: [hardware-cards] Tests that require 5 accelerator cards",
-    "cards_6: [hardware-cards] Tests that require 6 accelerator cards",
-    "cards_7: [hardware-cards] Tests that require 7 accelerator cards",
-    "cards_8: [hardware-cards] Tests that require 8 accelerator cards",
+    # Tag with [hardware-resource] for tests.helpers.mark.get_hardware_mark_list().
+    "H100: [hardware-resource] Tests that require H100 GPU",
+    "H800: [hardware-resource] Tests that require H800 GPU",
+    "H200: [hardware-resource] Tests that require H200 GPU",
+    "L4: [hardware-resource] Tests that require L4 GPU",
+    "B200: [hardware-resource] Tests that require B200 GPU",
+    "MI325: [hardware-resource] Tests that require MI325 GPU (AMD/ROCm)",
+    "B60: [hardware-resource] Tests that require Intel Arc Pro B60 XPU",
+    "S5000: [hardware-resource] Tests that require S5000 GPU (Moore Threads/MUSA)",
+    "A2: [hardware-resource] Tests that require A2 NPU",
+    "A3: [hardware-resource] Tests that require A3 NPU",
+    "distributed_cuda: Tests that require multi cards on CUDA platform",
+    "distributed_rocm: Tests that require multi cards on ROCm platform",
+    "distributed_xpu: Tests that require multi cards on XPU platform",
+    "distributed_npu: Tests that require multi cards on NPU platform",
+    "distributed_musa: Tests that require multi cards on MUSA platform",
+    "skipif_cuda: Skip if the num of CUDA cards is less than the required",
+    "skipif_rocm: Skip if the num of ROCm cards is less than the required",
+    "skipif_xpu: Skip if the num of XPU cards is less than the required",
+    "skipif_npu: Skip if the num of NPU cards is less than the required",
+    "skipif_musa: Skip if the num of MUSA cards is less than the required",
    # more detailed markers
    "sp: Sequence parallelism tests (multi-GPU)",
-    "slow: [ci-level] Slow tests (may skip in quick CI)",
+    "slow: Slow tests (may skip in quick CI)",
    "benchmark: Benchmark tests",
    "xdist: xdist-parallelizable tests",
 ]
@@ -1,14 +1,11 @@
 # Boogu-Image

-> Text-to-image and image-editing online serving
-(Boogu-Image-0.1-Base / -Edit / -Edit-Turbo)
+> Text-to-image and image-editing online serving (Boogu-Image-0.1-Base / -Edit)

 ## Summary

 - Vendor: Boogu
- Model: `Boogu/Boogu-Image-0.1-Base` (text-to-image),
-  `Boogu/Boogu-Image-0.1-Edit` (image editing), and
-  `Boogu/Boogu-Image-0.1-Edit-Turbo` (four-step image editing)
+- Model: `Boogu/Boogu-Image-0.1-Base` (text-to-image), `Boogu/Boogu-Image-0.1-Edit` (image editing)
 - Task: Text-to-image generation and text-guided image editing (TI2I)
 - Mode: Online serving with the OpenAI-compatible API
 - Maintainer: Community
@@ -16,23 +13,19 @@
 ## When to use this recipe

 Use this recipe when you want a known-good starting point for serving
-`Boogu/Boogu-Image-0.1-Base`, `Boogu/Boogu-Image-0.1-Edit`, or
-`Boogu/Boogu-Image-0.1-Edit-Turbo` with vLLM-Omni's native pipeline (no
+`Boogu/Boogu-Image-0.1-Base` with vLLM-Omni's native pipeline (no
 `--diffusion-load-format diffusers`, and the upstream `boogu` package is not
 required).

 Boogu-Image-0.1 is an Apache-2.0 unified image generation and editing model
-family. The Base text-to-image checkpoint pairs a Qwen3-VL multimodal encoder
-with a Diffusion Transformer (DiT) and a flow-match Euler scheduler with
-time-shift. It handles photorealistic generation and Chinese/English text
-rendering. The Edit and Edit-Turbo checkpoints use the same native
-`BooguImagePipeline` for image editing.
+family. This recipe covers the Base text-to-image checkpoint, which pairs a
+Qwen3-VL multimodal encoder with a Diffusion Transformer (DiT) and a flow-match
+Euler scheduler with time-shift. It handles photorealistic generation and
+Chinese/English text rendering.

 ## References

 - Upstream model card: <https://huggingface.co/Boogu/Boogu-Image-0.1-Base>
- Edit model card: <https://huggingface.co/Boogu/Boogu-Image-0.1-Edit>
- Edit-Turbo 1K hotfix: <https://huggingface.co/Boogu/Boogu-Image-0.1-Edit-Turbo/tree/hotfix-1k-20260708>
 - Project page: <https://boogu.org>
 - GitHub: <https://github.com/boogu-project/Boogu-Image>
 - Related example: [`examples/online_serving/text_to_image/`](../../examples/online_serving/text_to_image/README.md)
@@ -100,7 +93,7 @@ curl -s http://localhost:8091/v1/chat/completions \
 - **Memory usage:** ~34.6 GiB on GPU. Use `--vae-use-slicing --vae-use-tiling`
  to trim peak VRAM if needed.
 - **Key flags:**
-    - `--omni` — enables vLLM-Omni diffusion serving.
+  - `--omni` — enables vLLM-Omni diffusion serving.
 - **Guidance:** Boogu-Image uses `guidance_scale` (mapped to the upstream
  `text_guidance_scale`); the default is `4.0`. Classifier-free guidance is
  active whenever `guidance_scale > 1.0`.
@@ -118,13 +111,13 @@ the image-editing (TI2I) path activates automatically when a request carries a
 reference image. The Base text-to-image path is unaffected (no reference image
 is sent).

-### Command
+#### Command

 ```bash
 vllm serve Boogu/Boogu-Image-0.1-Edit --omni --port 8091
 ```

-### Verification
+#### Verification

 Edit an image with `/v1/images/edits` (the model-card example — change a photo
 to a colored-pencil drawing). Diffusion parameters are plain multipart form
@@ -164,14 +157,14 @@ curl -s http://localhost:8091/v1/chat/completions \
  }' | jq -r '.choices[0].message.content[0].image_url.url' | cut -d',' -f2- | base64 -d > edited.png
 ```

-### Notes
+#### Notes

 - **Single reference image:** only one input image is supported for now (the
  upstream "Only support 1 reference image for now" limit).
 - **Guidance semantics:**
-    - `guidance_scale` = text guidance (upstream `text_guidance_scale`, default
+  - `guidance_scale` = text guidance (upstream `text_guidance_scale`, default
    `4.0`); `> 1.0` enables text CFG. Editing typically uses `5.0`.
-    - `guidance_scale_2` = image guidance (upstream `image_guidance_scale`,
+  - `guidance_scale_2` = image guidance (upstream `image_guidance_scale`,
    default `1.0` = off). Setting it `> 1.0` enables the double-guidance path
    (3 model predictions per step), steering more strongly toward the reference
    image.
@@ -185,69 +178,3 @@ curl -s http://localhost:8091/v1/chat/completions \
  supports Boogu-Image-Edit directly
  (`--model Boogu/Boogu-Image-0.1-Edit --guidance-scale 5.0`, optional
  `--guidance-scale-2`).
-
-## Fast image editing (Boogu-Image-0.1-Edit-Turbo)
-
-Edit-Turbo is the distilled image-editing checkpoint. Its `model_index.json`
-declares `BooguImagePipeline`, so it uses the same native TI2I path as Edit.
-The upstream 1K hotfix is recommended for stable results and is pinned below
-to avoid silently loading the older checkpoint from the repository's default
-revision.
-
-### Command
-
-```bash
-vllm serve Boogu/Boogu-Image-0.1-Edit-Turbo \
-  --omni \
-  --revision hotfix-1k-20260708 \
-  --port 8091
-```
-
-### Verification
-
-Edit an image with the distilled four-step settings:
-
-```bash
-curl -s http://localhost:8091/v1/images/edits \
-  -F model="Boogu/Boogu-Image-0.1-Edit-Turbo" \
-  -F image="@input.png" \
-  -F prompt="Change the style to a colored pencil drawing." \
-  -F num_inference_steps=4 \
-  -F guidance_scale=1.0 \
-  -F seed=42 \
-  | jq -r '.data[0].b64_json' | base64 -d > edited-turbo.png
-```
-
-Or via chat completions:
-
-```bash
-curl -s http://localhost:8091/v1/chat/completions \
-  -H "Content-Type: application/json" \
-  -d '{
-    "messages": [
-      {"role": "user", "content": [
-        {"type": "image_url", "image_url": {"url": "data:image/png;base64,<BASE64>"}},
-        {"type": "text", "text": "Change the style to a colored pencil drawing."}
-      ]}
-    ],
-    "extra_body": {
-      "num_inference_steps": 4,
-      "guidance_scale": 1.0,
-      "seed": 42
-    }
-  }' | jq -r '.choices[0].message.content[0].image_url.url' | cut -d',' -f2- | base64 -d > edited-turbo.png
-```
-
-### Notes
-
- **Pinned revision:** use `hotfix-1k-20260708`; upstream recommends the 1K
-  hotfix over the 1.5K variant for more stable results.
- **Recommended settings:** `num_inference_steps=4` and
-  `guidance_scale=1.0`. Edit-Turbo is guidance-distilled, meaning the guidance
-  behavior is already baked into the distilled checkpoint. A scale of `1.0`
-  avoids applying additional classifier-free guidance; unlike the regular Edit
-  checkpoint, it should not be increased to `5.0`.
- **Reference images:** the same single-reference and `align_res` behavior as
-  the regular Edit checkpoint applies.
- **Known limitations:** the same single-GPU limitations as Base and Edit apply;
-  CPU offload, Cache-DiT, and multi-GPU parallelism are not yet validated.
@@ -10,22 +10,14 @@ checkpoint for T2V and first-frame I2V. Every output includes synchronized
 48 kHz stereo audio.

 The raw [`Lightricks/LTX-2.5`](https://huggingface.co/Lightricks/LTX-2.5)
-repository is not directly loadable with `--model`. It supplies the canonical
-Diffusion VAE checkpoint and the official upsampler and LoRA sidecars used by
-the Full/SFT two-stage pipeline, so accept both model licenses and authenticate
-before first use.
-
-The vLLM-Omni integration code is provided under the project's
-[Apache-2.0 license](../../LICENSE). The LTX-2.5 model weights are separate
-artifacts covered by the
-[LTX-2.x Community License](https://github.com/Lightricks/LTX-2/blob/main/LICENSE.md),
-not the vLLM-Omni license. Users are responsible for reviewing and complying
-with the model license before downloading or using the weights.
+repository is not directly loadable with `--model`. It supplies the official
+upsampler and LoRA sidecars used by the Full/SFT two-stage pipeline, so accept
+both model licenses and authenticate before first use.

 ## Choose a pipeline

 | Pipeline | Mode | Default size | Schedule |
-| --- | --- | ---: | ---: |
+|---|---|---:|---:|
 | `LTX2Pipeline` | Full/SFT one-stage | 960x544 | 30 steps |
 | `LTX2TwoStagePipeline` | Full/SFT two-stage | 1920x1088 | 30 + 3 steps |
 | `LTX2DistilledOneStagePipeline` | Distilled one-stage | 960x544 | 8 steps |
@@ -36,40 +28,6 @@ latent upsampler, and run a three-step refinement stage. Select the class with
 `--model-class-name`; no `--task-type` flag is required. Supplying one initial
 image selects I2V, while omitting it selects T2V.

-## Video decoder
-
-LTX-2.5 uses the canonical Native Diffusion VAE decoder by default. To opt in
-to the legacy convolutional VAE, set the startup-only stage-0 model extra
-`ltx2_use_conv_vae: true`. The choice applies to all requests and all four
-pipeline classes above. DiffVAE loads its NATTEN kernel from Hugging Face Hub
-during startup. `decode_timestep` and `decode_noise_scale` condition only the
-legacy ConvVAE and have no effect when DiffVAE is selected.
-
-For offline Python usage:
-
-```python
-omni = Omni(
-    model="Lightricks/LTX-2.5-Diffusers",
-    model_class_name="LTX2Pipeline",
-    stage_overrides='{"0":{"extras":{"ltx2_use_conv_vae":true}}}',
-)
-```
-
-For online serving:
-
-```bash
-vllm serve Lightricks/LTX-2.5-Diffusers \
-  --omni \
-  --model-class-name LTX2Pipeline \
-  --stage-overrides '{"0":{"extras":{"ltx2_use_conv_vae":true}}}'
-```
-
-Both decoders are untiled by default. Set `vae_use_tiling` for memory-saving
-serial tiling; DiffVAE tiles only above 80 frames or 768 pixels in either spatial
-dimension. DiffVAE also supports distributed VAE decode; see the
-[VAE Parallelism Guide](../../docs/user_guide/diffusion/parallelism/vae_parallelism.md).
-DiffVAE is decoder-only, so I2V still uses the convolutional VAE encoder.
-
 ## Prerequisites

 ```bash
@@ -84,7 +42,7 @@ Install matching vLLM and vLLM-Omni versions, and ensure `ffmpeg` and
 ## Hardware

 | GPU | Status | Recommended scope |
-| --- | --- | --- |
+|---|---|---|
 | NVIDIA B300 | Verified | All four canonical pipelines |
 | NVIDIA B200 or H200 | Capacity-based recommendation; not yet verified | All four canonical pipelines |
 | NVIDIA GB200 or GB300 | Capacity-based recommendation; not yet verified | All four canonical pipelines |
@@ -1,82 +0,0 @@
-# MiniMax H3 disaggregated text encoder
-
-This opt-in topology runs the Qwen3-VL text encoder as a vLLM stage and sends
-its hidden states and token-role metadata to an encoder-free diffusion stage.
-The standard MiniMax H3 recipes remain single-stage and continue to load the
-text encoder inside the diffusion pipeline.
-
-## Start the server
-
-Choose the topology explicitly and load its deployment defaults:
-
-```bash
-vllm-omni serve MiniMaxAI/MiniMax-H3 \
-  --omni \
-  --deploy-config vllm_omni/deploy/minimax_h3_disaggregated.yaml
-```
-
-The default deployment assigns stage 0 to GPUs 0-1 with tensor parallel size
-2 and `max_num_seqs: 1`. Stage 1 uses GPUs 2-5 with diffusion tensor parallel
-size 1, Ulysses degree 4, and VAE patch parallel size 4. Adjust the
-`devices`, `tensor_parallel_size`, and stage 1 `parallel_config` values in a
-deployment override for the available hardware. Diffusion quantization,
-layerwise offload, distributed layerwise offload, VAE parallelism, and USP
-settings use the same stage 1 options documented in [MiniMax-H3.md](MiniMax-H3.md).
-
-For example, this five-GPU topology assigns one GPU to the encoder and four
-to the diffusion stage. `--stage-overrides` keeps placement and parallelism
-scoped to the owning stage rather than broadcasting an override to both:
-
-```bash
-CUDA_VISIBLE_DEVICES=0,1,2,3,4 \
-vllm-omni serve MiniMaxAI/MiniMax-H3 \
-  --omni \
-  --deploy-config vllm_omni/deploy/minimax_h3_disaggregated.yaml \
-  --stage-overrides '{"0":{"devices":"0","tensor_parallel_size":1},"1":{"devices":"1,2,3,4","tensor_parallel_size":1,"ulysses_degree":4,"vae_patch_parallel_size":4}}'
-```
-
-For memory-constrained deployments, start from the CPU-offload or distributed
-layerwise-offload profiles in [MiniMax-H3.md](MiniMax-H3.md). Apply memory and
-quantization options only to Stage 1 with `--stage-overrides`; retain the
-encoder's BF16 configuration and the video/audio VAEs' FP32 precision. Select
-one offload strategy per deployment:
-
-```bash
-# Stage 1 model-level CPU offload.
--stage-overrides '{"1":{"enable_cpu_offload":true}}'
-
-# Stage 1 distributed layerwise offload. Tune resident layers for available RAM.
--stage-overrides '{"1":{"enable_distributed_layerwise_offload":true,"dlo_use_allgather":false,"dlo_resident_layers":20}}'
-
-# Stage 1 online FP8 quantization of the DiT only.
--stage-overrides '{"1":{"diffusion_quantization_config":"{\"transformer\":{\"method\":\"fp8\"}}"}}'
-```
-
-The Stage 1 VAE patch-parallel options remain independent of offload and
-quantization. See [MiniMax-H3.md](MiniMax-H3.md) for memory requirements and
-hardware-qualified profiles before combining these options.
-
-Stage 1 sets `model_loaded.text_encoder: false`; it must not load or download
-text-encoder weights. This H3 topology explicitly keeps its single-replica
-diffusion stage inline, avoiding serialization of decoded video through a
-subprocess. It expects both stages to run in one deployment; cross-node payload
-transport is outside this configuration.
-
-The `/v1/videos` request schema and `extra_params.task` values (`t2va`,
-`fl2va`, and `ref2va`) are unchanged from the single-stage recipe.
-
-## Turbo LoRA
-
-MiniMax-H3 Turbo uses five sigma points, `flow_shift=6`, and
-`audio_flow_shift=3`. Start Turbo deployments with the dedicated defaults so
-requests that omit sampling controls do not inherit the 50-step base schedule:
-
-```bash
-vllm-omni serve MiniMaxAI/MiniMax-H3 \
-  --omni \
-  --lora-path /path/to/MiniMax-H3-Turbo \
-  --deploy-config vllm_omni/deploy/minimax_h3_disaggregated_turbo.yaml
-```
-
-The base deployment intentionally retains 50 inference steps for non-LoRA
-quality. Turbo LoRA supports T2VA and FL2VA requests, not Ref2VA.
@@ -69,12 +69,11 @@ reference-video preparation and MP4 output.

 For CUDA and ROCm deployments, non-streaming MP4 responses are encoded on the
 host CPU through PyAV/libx264 after generation. The response encoder selects the
-path automatically at runtime. The server-owned parallel converter accepts
-supported frame shapes and dtypes with either per-channel-contiguous or strided
-RGB planes, including interleaved arrays materialized by output transport.
-Standalone callers without a parallel converter retain the legacy fallback for
-strided planes. No CLI flag, model declaration, or user configuration is
-required. Streaming fMP4 output is unchanged.
+path automatically at runtime: inputs with a supported frame shape, common dtype,
+and RGB per-channel contiguous layout use direct planar frames; other inputs
+fall back to the legacy path before the PyAV container is opened. No CLI flag,
+model declaration, or user configuration is required. Streaming fMP4 output is
+unchanged.

 A community benchmark on 2x Xeon 8480C reported the following comparison
 between the legacy and direct planar paths
@@ -861,129 +860,6 @@ fixed HBM usage. Model-level and standard layerwise offload remain unsupported.
 The five requested sigma points produce the four denoiser evaluations expected
 by the Turbo artifact.

-### FlashGen native LoRA
-
-The FlashGen 4-step T2VA artifact uses the native MiniMax-H3 module layout and
-declares its distilled sigma schedule in safetensors metadata. It is published on
-[ModelScope](https://modelscope.cn/models/FlashGen/Minimax-H3-4step-lora-flashgen):
-
-```text
-FlashGen/Minimax-H3-4step-lora-flashgen/minimax_h3_t2va_flashgen_4step_v1.0_768p_bf16.safetensors
-```
-
-Download only that file:
-
-```bash
-python -m pip install modelscope
-export FLASHGEN_DIR=/path/to/minimax-h3-flashgen-lora
-export FLASHGEN_FILE=minimax_h3_t2va_flashgen_4step_v1.0_768p_bf16.safetensors
-modelscope download FlashGen/Minimax-H3-4step-lora-flashgen \
-  --local_dir "${FLASHGEN_DIR}" \
-  --include "${FLASHGEN_FILE}"
-export FLASHGEN_LORA="${FLASHGEN_DIR}/${FLASHGEN_FILE}"
-```
-
-Start from a non-offloaded or DLO FL2VA server command and add
-`--task-type fl2va --lora-backend peft --lora-path "${FLASHGEN_LORA}"`.
-Each request must use T2VA and the distilled interval-count contract:
-
-```bash
-F 'num_inference_steps=4' \
-F 'extra_params={"task":"t2va","duration":5.2}' \
-F "lora={\"name\":\"h3-flashgen-v1.0\",\"path\":\"${FLASHGEN_LORA}\",\"scale\":1.0}"
-```
-
-This path rejects Ref2VA and checkpoints that already pin `base_schedule` in
-`model_index.json`. The adapter metadata carries
-`base_schedule=1.0,0.7,0.4,0.15,0.0`, so `num_inference_steps=4` means four
-denoiser evaluations, not five sigma points. Request-mode generation may omit
-the field and take the count from the adapter schedule; `--step-execution`
-requires it explicitly, because the step scheduler reads the total step count
-off the request at admission, before the adapter schedule is known.
-
-DLO is supported in request-mode generation on the same terms as the Turbo
-adapter: the request-switchable LoRA A/B buffers stay resident on the
-accelerator while DLO streams only the base blocks, so budget for that
-additional fixed HBM usage. The native artifact is rank 64 over 259 target
-modules, and its packed `qkv_proj` and `fc1` layers reuse the full-input A
-tensor per slice while B carries slice-local output rows, so the resident
-footprint exceeds the on-disk payload; measure it for your parallel layout
-rather than assuming the checkpoint size. Pure Ulysses replicates the adapter
-on every rank, while DiT tensor parallelism shards the B buffers. Model-level
-and standard layerwise offload remain unsupported, and `--step-execution`
-cannot be combined with `--enable-distributed-layerwise-offload`.
-
-To validate a deployment, post the same fixed-seed T2VA request twice with the
-adapter and twice without it, then compare the four output digests. The adapter
-is bound and deterministic when each pair matches internally and the two pairs
-differ from each other.
-
-### FastH3 adapter
-
-[FastH3](https://haoailab.com/blogs/fasth3-preview/) is FastVideo's four-step
-DMD2 student of H3-Base. It reuses H3's text encoder, VAEs, tokenizers, and
-schedulers unchanged, replacing 49 denoiser evaluations with four. It is fused
-into the checkpoint at load time rather than switched per request, because it
-carries full-rank deltas that no LoRA layer can express.
-
-The bundle publishes four variants, so download one and point `--lora-path` at
-it; the repository root is refused rather than guessed at:
-
-```bash
-export FASTH3_DIR=/path/to/fasth3
-hf download FastVideo/FastVideo-FastH3-4-step-Preview-v1-LoRA \
-  dense-datafree/adapter_model.safetensors --local-dir "${FASTH3_DIR}"
-export FASTH3_LORA="${FASTH3_DIR}/dense-datafree/adapter_model.safetensors"
-```
-
-Add `--task-type fl2va --lora-path "${FASTH3_LORA}"` to a non-offloaded server
-command. T2VA is served by the FL2VA partition, so `--task-type fl2va` is
-correct even though FastH3 preview v1 distills T2VA only. Because the adapter is
-fused, `--lora-backend` does not apply and a request carrying a `lora=` field is
-rejected rather than served without the adapter it asked for.
-
-```bash
-F 'num_inference_steps=4' \
-F 'extra_params={"task":"t2va","duration":4.4}'
-```
-
-Requests must ask for `num_inference_steps=4` and `task=t2va`: the release's five
-sigma points bound four denoiser evaluations, and that count is what the step
-scheduler admits a request on. The server denoises on the release's own ladder,
-keeping H3's per-modality shifts at the checkpoint values, so a request that
-overrides `flow_shift` or `audio_flow_shift` is rejected - it would sample the
-student at noise levels it was never distilled at.
-
-Only a release that identifies itself as FastH3 is fused; any other
-`fastvideo-lora-v2` adapter stays on the dynamic LoRA route. A claimed artifact
-is then held to its own metadata: one that misdeclares its tensor counts or
-leaves a transformer block unedited is refused at startup instead of serving
-mostly base H3 weights on a four-step schedule. Offload is refused for the same
-reason - `--enable-cpu-offload`, `--enable-layerwise-offload` and
-`--enable-distributed-layerwise-offload` all bypass the fusion, so they fail fast.
-
-> [!NOTE]
-> Only the **Dense / Data-Free** variant is supported today. The three VSA
-> variants carry `.set_weight` compression gates for modules that do not exist
-> in vLLM-Omni's dense H3 attention, and the release states dense attention is
-> not a supported substitute for them, so they are refused at startup rather
-> than silently run wrong. Video Sparse Attention for H3's packed
-> `[text | cond | audio | video]` sequence is follow-up work.
-
-Measured on 8x NVIDIA B300 with USP8, VAE patch-parallel 8, `TRTLLM_ATTN`, at
-1344x768, 4.4 s, seed 1101, one warmup excluded and two runs recorded:
-
-| Adapter | Steps | End-to-end | Diffusion engine |
-| --- | ---: | ---: | ---: |
-| none (base H3) | 50 | 25.8 / 26.4 s | 16.22 / 16.28 s |
-| FastH3 Dense | 4 (5 sigma points) | 11.7 / 11.8 s | 2.37 / 2.36 s |
-
-The denoising speedup is 6.9x. End-to-end is 2.2x because text encoding, VAE
-decoding and muxing are a fixed cost that dominates a clip this short; longer
-generations move the end-to-end figure toward the denoising one. Fusing the
-adapter does not measurably change startup: weight loading took 77.3 s with it
-against 85.8 s without.
-
 ## Key parameters

 | Parameter | Recommended value | Notes |
@@ -1142,4 +1018,3 @@ vllm serve "${MODEL_ROOT}/FL2VA" \
 - [Supported models](../../docs/models/supported_models.md)
 - [Video API](../../docs/serving/videos_api.md)
 - [Diffusion parallelism](../../docs/user_guide/diffusion/parallelism/overview.md)
- [Temporal VAE chunk callback design](../../docs/design/feature/minimax_h3_temporal_vae_chunks.md)
@@ -18,11 +18,11 @@ examples in this repository.

 Qwen3-TTS supports three task types, each backed by a dedicated model checkpoint:

-| Task Type     | Model                                    | Description                                                   |
-| ------------- | ---------------------------------------- | ------------------------------------------------------------- |
-| `CustomVoice` | `Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice`   | Predefined speaker voices with optional style/emotion control |
-| `VoiceDesign` | `Qwen/Qwen3-TTS-12Hz-1.7B-VoiceDesign`   | Generate speech from a natural language voice description     |
-| `Base`        | `Qwen/Qwen3-TTS-12Hz-1.7B-Base`          | Voice cloning from reference audio + transcript               |
+| Task Type     | Model                                    | Description                                                  |
+| ------------- | ---------------------------------------- | ------------------------------------------------------------ |
+| `CustomVoice` | `Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice`  | Predefined speaker voices with optional style/emotion control |
+| `VoiceDesign` | `Qwen/Qwen3-TTS-12Hz-1.7B-VoiceDesign`  | Generate speech from a natural language voice description     |
+| `Base`        | `Qwen/Qwen3-TTS-12Hz-1.7B-Base`         | Voice cloning from reference audio + transcript              |

 Smaller 0.6B variants are also available for `CustomVoice` and `Base`.

@@ -166,15 +166,6 @@ python examples/offline_inference/text_to_speech/qwen3_tts/end2end.py --query-ty
 - Key flags: `--omni` is required. `--deploy-config` points to the bundled two-stage pipeline config.
 - Async chunking: Enabled by default in `qwen3_tts.yaml` for streaming-friendly first-audio latency. Raw audio streaming requires `stream=true`, `stream_format="audio"`, and `response_format="pcm"`.
 - Task/model matching: Each task type requires its matching model checkpoint. Using a CustomVoice model for a Base (voice clone) request will fail.
- Base codec termination: Base requests without an explicit `max_new_tokens`
-  use a text-scaled safety ceiling (at least 192 codec frames and no more than
-  the configured model limit). If the Talker reaches that ceiling without
-  codec EOS, non-streaming serving discards the incomplete audio and retries
-  once with a fresh seed; an explicit `seed` or `max_new_tokens` disables the
-  retry. SSE and WebSocket clients receive structured errors and must discard
-  previously emitted audio when the error contains `"action":"discard"`;
-  raw PCM streams terminate with a connection error after any partial bytes
-  already sent.
 - Known limitations: The server serves one model variant at a time. To switch task types (e.g., CustomVoice to Base), restart the server with the corresponding model.

 ## Hardware Support
@@ -78,7 +78,6 @@ recipes/
 | [`Qwen/Qwen3-Omni.md`](./Qwen/Qwen3-Omni.md) | Online serving for multimodal chat | 1x A100 80GB |
 | [`Qwen/Qwen3-TTS.md`](./Qwen/Qwen3-TTS.md) | Text-to-speech serving (CustomVoice / VoiceDesign / Base) | 1x H100/A100 80GB / 1x AMD MI300X |
 | [`SenseNova/SenseNova-U1.md`](./SenseNova/SenseNova-U1.md) | Unified image generation and understanding | 1x or 2x H200 / 1x AMD MI300X |
-| [`SenseNova/SenseNova-U1.5.md`](./SenseNova/SenseNova-U1.5.md) | Unified image generation and understanding, with an 8-step distilled LoRA | 1x A800 80GB |
 | [`StabilityAI/Stable-Audio-Open.md`](./StabilityAI/Stable-Audio-Open.md) | Offline + online text-to-audio generation (Stable Audio Open) | 1x RTX 4090 24GB / 1x AMD MI300X |
 | [`Tencent/Covo-Audio-Chat.md`](./Tencent/Covo-Audio-Chat.md) | Online serving for audio chat | 1x A100 80GB |
 | [`Tencent/HunyuanImage-3.0-Instruct.md`](./Tencent/HunyuanImage-3.0-Instruct.md) | DiT-only text-to-image serving and benchmark, including ModelOpt mixed FP8/NVFP4 | 4x H100/H800 80GB / 2x B200 |
@@ -1,159 +0,0 @@
-# SenseNova-U1.5
-
-> Unified image generation and understanding, with an official 8-step distilled LoRA
-
-## Summary
-
- Vendor: SenseNova
- Model: `sensenova/SenseNova-U1.5-8B-MoT`
- LoRA: `sensenova/SenseNova-U1.5-8B-MoT-LoRAs` (`SenseNova-U1.5-8B-MoT-LoRA-8step.safetensors`)
- Task: text2img, img2img, img2text (visual understanding), text2text (chat)
- Mode: Offline inference, Online serving (OpenAI-compatible API)
- Maintainer: Community
-
-## When to use this recipe
-
-U1.5 runs on the same `SenseNovaU1Pipeline` as U1 — the checkpoint keeps
-`model_type: neo_chat`, so it resolves without `--model-class-name`. Relative to U1 the
-config flips two fields, `use_pixel_head` to `true` (the flow-matching head becomes a
-`ConvDecoder`) and `noise_scale_max_value` from 8.0 to 16.0; both are already read by
-`SenseNovaU1Config`. The checkpoint is 13 shards / 50.2 GB on disk, 30.3 GB of that fp32, and
-loads to roughly 34 GB in bf16.
-
-Use this recipe for U1.5 specifically, including its distilled few-step LoRA. For U1 see
-[SenseNova-U1](SenseNova-U1.md).
-
-## Hardware Support
-
-### GPU
-
-#### 1x A800 80GB
-
- 1x NVIDIA A800 80GB PCIe with an Intel Xeon Gold 6336Y, driver 595.84, Linux
- Python 3.12.13, vLLM 0.28.0, torch 2.13.0+cu130, CUDA 13.0, diffusers 0.40.0,
-  transformers 5.14.1, flashinfer 0.6.16.post3, BF16, TP=1
- Peak GPU memory: 34.3 GB at 1024x1024, 36.4 GB at 1536x2720
-
-##### Text-to-Image
-
-```bash
-python examples/offline_inference/text_to_image/text_to_image.py \
-    --model sensenova/SenseNova-U1.5-8B-MoT \
-    --prompt "Close portrait of an elderly woman by a farmhouse window, warm natural light." \
-    --width 1024 --height 1024 \
-    --seed 42 --num-inference-steps 50 --cfg-scale 4.0 \
-    --extra-body '{"think": false, "cfg_norm": "none", "timestep_shift": 3.0, "t_eps": 0.02}' \
-    --output sensenova_u15_t2i.png
-```
-
-Think mode (`"think": true`) is recommended for higher image quality.
-
-##### Image-to-Image Editing
-
-```bash
-python examples/offline_inference/image_to_image/image_edit.py \
-    --model sensenova/SenseNova-U1.5-8B-MoT \
-    --prompt "Turn this into an oil painting" \
-    --image input.png --resolution 1024 \
-    --seed 42 --num-inference-steps 25 --cfg-scale 4.0 \
-    --extra-args '{"think": true, "img_cfg_scale": 1.0, "cfg_norm": "none", "timestep_shift": 3.0}' \
-    --output sensenova_u15_edit.png
-```
-
-##### 8-step distilled LoRA
-
-The distilled LoRA is fused into the generation tower at load time:
-
-```bash
-python examples/offline_inference/text_to_image/text_to_image.py \
-    --model sensenova/SenseNova-U1.5-8B-MoT \
-    --lora-path SenseNova-U1.5-8B-MoT-LoRA-8step.safetensors --lora-backend distill \
-    --prompt "Close portrait of an elderly woman by a farmhouse window, warm natural light." \
-    --width 1024 --height 1024 \
-    --seed 42 --num-inference-steps 8 --cfg-scale 1.0 \
-    --extra-body '{"think": false, "cfg_norm": "none", "timestep_shift": 3.0, "t_eps": 0.02}' \
-    --output sensenova_u15_lora8.png
-```
-
-**Use `--cfg-scale 1.0` with this LoRA.** It is distilled with DMD and runs without
-classifier-free guidance; the default `4.0` applies guidance twice and produces a
-blown-out, posterised image.
-
-##### Online serving
-
-```bash
-vllm serve sensenova/SenseNova-U1.5-8B-MoT --omni --port 8091
-
-python examples/online_serving/sensenova_u1/openai_chat_client.py \
-    -s http://127.0.0.1:8091 -m img2text -i input.png -p "Describe this image."
-```
-
-`-s` takes the base URL; the client appends `/v1` itself.
-
-#### Measured latency (1x A800 80GB, 25 steps, median of 3 after a warmup)
-
-| Resolution | Step latency | Total |
-| --- | --- | --- |
-| 1024x1024 | 209.5 ms | 5.24 s |
-| 1536x1536 | 458.5 ms | 11.46 s |
-
-At 1536x2720 with 50 steps, end-to-end is 43.0 s.
-
-#### Measured latency (1x NVIDIA H200 139GB, single run, no warmup)
-
-Reported by @hsliuustc0106 against PR head `29c090d`, on one reserved CUDA device.
-
- Python 3.12.13, vLLM 0.28.0, torch 2.13.0+cu130, CUDA 13.0, diffusers 0.40.0,
-  transformers 5.14.1, flashinfer 0.6.16.post3, BF16, TP=1
- `sensenova/SenseNova-U1.5-8B-MoT` with `SenseNova-U1.5-8B-MoT-LoRA-8step.safetensors`,
-  seed 42, 1024x1024
-
-| Case | Stage latency | Peak GPU memory |
-| --- | --- | --- |
-| 50 steps, cfg 4.0, think off | 9017 ms | 34384 MiB |
-| 8 steps, cfg 1.0, no LoRA control | 624.00 ms, 77.85 ms/step | 34376 MiB |
-| 8 steps, cfg 1.0, distilled LoRA | 623.85 ms, 77.98 ms/step | 34364 MiB |
-| 50 steps, cfg 4.0, think on, `VLLM_OMNI_SENSENOVA_PAGED_DECODE=1` | 81622 ms | 34594 MiB |
-
-Each row runs the Text-to-Image command above with `--num-inference-steps` and
-`--cfg-scale` as listed and `think` set through `--extra-body`; the LoRA row adds
-`--lora-path` and `--lora-backend distill`.
-
-Single runs without a preceding warmup, so the think-on row carries the first-request
-compile rather than a steady state. The distilled LoRA fused into 168 parameters, and its
-image differs from the same-seed no-LoRA control in 1,048,574 of 1,048,576 pixels, MAE 58.62.
-
-```bash
-pytest -m "cpu and not cuda" tests/diffusion/models/sensenova_u1/ tests/diffusion/lora/test_loader.py -q
-pytest -m "cpu and not cuda" tests/config/test_environment_variables.py tests/diffusion/test_diffusion_worker.py -q
-pytest -m cuda tests/diffusion/models/sensenova_u1/test_sensenova_u1_attention_gqa.py \
-    tests/diffusion/models/sensenova_u1/test_sensenova_u1_paged_decode.py -q --tb=short
-```
-
-71, 13 and 9 passed. Serving reached `/health`; `img2text` returned a 1921-character
-description and `text2text` the expected answer.
-
-#### Verification
-
-```bash
-pytest -q tests/diffusion/models/sensenova_u1/
-```
-
-## Notes
-
- Both `use_pixel_head` and `noise_scale_max_value` come from `config.json`; no flag is needed.
- The checkpoint carries no `configuration_neo_chat.py`, so the loader logs a few
-  "does not appear to have a file named configuration_neo_chat.py" errors during startup and
-  then proceeds on the in-tree config. Generation is unaffected.
- The LoRA targets the generation tower only (`*_mot_gen`); understanding-tower weights are
-  untouched.
- Autoregressive decode (think, text-to-text and image-to-text) runs on a paged K/V cache under
-  a captured CUDA graph. It falls back to the ordinary cache when the device or the bundled
-  `flash_attn_varlen_func` cannot support it; set `VLLM_OMNI_SENSENOVA_PAGED_DECODE=0` to force
-  that fallback.
- The first request after startup costs about 0.7 s more than the steady state whether the paged
-  path is on or off. Measured on one A800 with the inductor, triton and vLLM compile caches all
-  cleared, median of three runs: 718 ms above steady with the path on, 679 ms with it off.
- Each request captures its own graphs, and a think request captures twice because the sequence
-  grows past the 512 bucket, so the capture cost is paid per request rather than once at
-  startup.
@@ -13,8 +13,6 @@ transformers >= 5.10.1, < 5.15
 av>=14.0.0
 omegaconf>=2.3.0
 diffusers==0.40.0
-# Hub kernels, including LTX-2.5 NATTEN
-kernels==0.14.1
 safetensors>=0.8.0
 accelerate==1.12.0
 soundfile>=0.13.1
@@ -1,5 +1,5 @@
 # SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
 """Benchmark legacy and automatic MP4 response encoding on synthetic input."""

 from __future__ import annotations
@@ -11,16 +11,13 @@ import statistics
 import time
 from collections.abc import Callable
 from dataclasses import dataclass
-from functools import partial
 from pathlib import Path
-from typing import cast

 import numpy as np

 from vllm_omni.entrypoints.openai.video_api_utils import (
    _encode_video_bytes,
    _encode_video_bytes_legacy,
-    _PlanarFrameConverter,
 )


@@ -45,19 +42,12 @@ def _build_inputs(
    fps: int,
    audio_sample_rate: int,
    seed: int,
-    input_layout: str = "planar-backed",
 ) -> tuple[np.ndarray, np.ndarray]:
    rng = np.random.default_rng(seed)
    planar = rng.random((3, frames, height, width), dtype=np.float32)
    video = planar.transpose(1, 2, 3, 0)
-    if input_layout == "interleaved":
-        video = np.ascontiguousarray(video)
-        if not video.flags.c_contiguous:
-            raise RuntimeError("synthetic video is not C-contiguous")
-    elif input_layout == "planar-backed" and not all(video[..., channel].flags.c_contiguous for channel in range(3)):
+    if not all(video[..., channel].flags.c_contiguous for channel in range(3)):
        raise RuntimeError("synthetic video does not expose contiguous channel planes")
-    elif input_layout != "planar-backed":
-        raise ValueError(f"unsupported input layout: {input_layout}")

    sample_count = max(1, round(frames / fps * audio_sample_rate))
    timeline = np.arange(sample_count, dtype=np.float32) / audio_sample_rate
@@ -96,8 +86,8 @@ def _measure(

 def _summarize(records: list[dict[str, object]], label: str) -> dict[str, object]:
    selected = [record for record in records if record["label"] == label]
-    wall_values = [cast(float, record["wall_ms"]) for record in selected]
-    cpu_values = [cast(float, record["process_cpu_ms"]) for record in selected]
+    wall_values = [float(record["wall_ms"]) for record in selected]
+    cpu_values = [float(record["process_cpu_ms"]) for record in selected]
    return {
        "runs": len(selected),
        "wall_ms": {
@@ -123,16 +113,6 @@ def _parse_args() -> argparse.Namespace:
    parser.add_argument("--warmup", type=int, default=1)
    parser.add_argument("--rounds", type=_positive_int, default=3)
    parser.add_argument("--seed", type=int, default=20260817)
-    parser.add_argument(
-        "--input-layout",
-        choices=("planar-backed", "interleaved"),
-        default="planar-backed",
-    )
-    parser.add_argument(
-        "--frame-conversion-workers",
-        type=_positive_int,
-        default=8,
-    )
    parser.add_argument("--output-json", type=Path)
    args = parser.parse_args()
    if args.warmup < 0:
@@ -149,41 +129,33 @@ def main() -> None:
        fps=args.fps,
        audio_sample_rate=args.audio_sample_rate,
        seed=args.seed,
-        input_layout=args.input_layout,
    )
    baseline = BenchmarkVariant("legacy", _encode_video_bytes_legacy)
-    converter = _PlanarFrameConverter(max_workers=args.frame_conversion_workers)
-    candidate = BenchmarkVariant(
-        "automatic",
-        partial(_encode_video_bytes, frame_converter=converter),
-    )
+    candidate = BenchmarkVariant("automatic", _encode_video_bytes)

-    try:
-        for _ in range(args.warmup):
-            for variant in (baseline, candidate):
-                _measure(
-                    variant,
-                    video=video,
-                    audio=audio,
-                    fps=args.fps,
-                    audio_sample_rate=args.audio_sample_rate,
-                )
+    for _ in range(args.warmup):
+        for variant in (baseline, candidate):
+            _measure(
+                variant,
+                video=video,
+                audio=audio,
+                fps=args.fps,
+                audio_sample_rate=args.audio_sample_rate,
+            )

-        records: list[dict[str, object]] = []
-        for round_index in range(args.rounds):
-            order = (baseline, candidate) if round_index % 2 == 0 else (candidate, baseline)
-            for variant in order:
-                _, record = _measure(
-                    variant,
-                    video=video,
-                    audio=audio,
-                    fps=args.fps,
-                    audio_sample_rate=args.audio_sample_rate,
-                )
-                record["round"] = round_index + 1
-                records.append(record)
-    finally:
-        converter.shutdown()
+    records: list[dict[str, object]] = []
+    for round_index in range(args.rounds):
+        order = (baseline, candidate) if round_index % 2 == 0 else (candidate, baseline)
+        for variant in order:
+            _, record = _measure(
+                variant,
+                video=video,
+                audio=audio,
+                fps=args.fps,
+                audio_sample_rate=args.audio_sample_rate,
+            )
+            record["round"] = round_index + 1
+            records.append(record)

    output_hashes = {str(record["output_sha256"]) for record in records}
    if len(output_hashes) != 1:
@@ -193,8 +165,8 @@ def main() -> None:
    paired_savings_percent = []
    for round_index in range(args.rounds):
        pair = [record for record in records if record["round"] == round_index + 1]
-        baseline_ms = cast(float, next(record["wall_ms"] for record in pair if record["label"] == baseline.label))
-        candidate_ms = cast(float, next(record["wall_ms"] for record in pair if record["label"] == candidate.label))
+        baseline_ms = float(next(record["wall_ms"] for record in pair if record["label"] == baseline.label))
+        candidate_ms = float(next(record["wall_ms"] for record in pair if record["label"] == candidate.label))
        paired_savings_ms.append(baseline_ms - candidate_ms)
        paired_savings_percent.append((baseline_ms - candidate_ms) / baseline_ms * 100)

@@ -208,8 +180,6 @@ def main() -> None:
            "warmup": args.warmup,
            "rounds": args.rounds,
            "seed": args.seed,
-            "input_layout": args.input_layout,
-            "frame_conversion_workers": args.frame_conversion_workers,
            "video_shape": list(video.shape),
            "video_strides": list(video.strides),
            "codec_options": {"preset": "ultrafast", "threads": "0"},
@@ -279,79 +279,17 @@ def test_unmeasured_tpot_stays_missing_after_tokenizer_fallback():
    assert metrics.request_goodput == 0.0


-def test_zero_itl_does_not_create_zero_tpot():
+def test_measured_zero_itl_is_not_treated_as_missing():
    output = _make_output(100, output_tokens=3)
    output.itl = [0.0, 0.0]
    output.duplex_session_metrics = {"mean_ttft_ms": 100.0}

    metrics = _calculate_test_metrics([output], {"tpot": 1.0})

-    assert (metrics.num_tpot_samples, metrics.num_itl_samples) == (0, 2)
-    assert math.isnan(metrics.mean_tpot_ms)
+    assert (metrics.num_tpot_samples, metrics.num_itl_samples) == (1, 2)
+    assert metrics.mean_tpot_ms == 0.0
    assert metrics.mean_itl_ms == 0.0
-    assert metrics.request_goodput == 0.0
-
-
-def test_stage_output_tokens_without_client_timing_omit_tpot(capsys):
-    output = _make_output(100, output_tokens=5)
-    output.itl = []
-    output.text_latency = 0.0
-    output.ttft = 0.1
-
-    metrics, _ = calculate_metrics(
-        input_requests=[],
-        outputs=[output],
-        dur_s=1.0,
-        tokenizer=None,
-        selected_percentiles=[50.0, 99.0],
-        goodput_config_dict={},
-        task_type=TaskType.GENERATION,
-        selected_percentile_metrics=["tpot"],
-        max_concurrency=None,
-        request_rate=float("inf"),
-        benchmark_duration=1.0,
-    )
-
-    printed = capsys.readouterr().out
-    assert metrics.num_tpot_samples == 0
-    assert math.isnan(metrics.mean_tpot_ms)
-    assert "Time per Output Token" not in printed
-    assert "Mean TPOT" not in printed
-
-
-def test_consistent_client_latency_can_supply_tpot_fallback():
-    output = _make_output(100, output_tokens=10)
-    output.itl = []
-    output.ttft = 0.1
-    output.text_latency = 1.0
-
-    metrics = _calculate_test_metrics([output])
-
-    assert metrics.num_tpot_samples == 1
-    assert metrics.mean_tpot_ms == pytest.approx(100.0)
-
-
-def test_single_token_responses_do_not_report_zero_tpot(capsys):
-    output = _make_output(100, output_tokens=1)
-
-    metrics, _ = calculate_metrics(
-        input_requests=[],
-        outputs=[output],
-        dur_s=1.0,
-        tokenizer=None,
-        selected_percentiles=[50.0],
-        goodput_config_dict={},
-        task_type=TaskType.GENERATION,
-        selected_percentile_metrics=["tpot"],
-        max_concurrency=None,
-        request_rate=float("inf"),
-        benchmark_duration=1.0,
-    )
-
-    printed = capsys.readouterr().out
-    assert metrics.num_tpot_samples == 0
-    assert math.isnan(metrics.mean_tpot_ms)
-    assert "Time per Output Token" not in printed
+    assert metrics.request_goodput == 1.0


 def test_duplex_goodput_does_not_pair_measurements_from_different_requests():
@@ -1,5 +1,5 @@
 # SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM project

 """
 Unit tests for patch.py
@@ -9,98 +9,22 @@ import asyncio
 import base64
 import json
 import time
-from argparse import Namespace
 from types import SimpleNamespace

 import pytest
 from pytest_mock import MockerFixture
 from vllm.benchmarks.lib.endpoint_request_func import RequestFuncInput

-from vllm_omni.benchmarks.data_modules.seed_tts_dataset import (
-    SeedTTSSampleRequest,
-    SeedTTSTextSampleRequest,
-)
 from vllm_omni.benchmarks.patch.patch import (
    MixRequestFuncOutput,
-    _apply_stage0_token_timings,
-    _attach_seed_tts_to_request_func_input,
    async_request_openai_chat_omni_completions,
    async_request_openai_realtime_duplex,
-    should_request_stage_metrics,
 )
 from vllm_omni.experimental.fullduplex.client import RealtimeEventCollector

 pytestmark = [pytest.mark.core_model, pytest.mark.benchmark, pytest.mark.cpu]


-def _seed_tts_request_func_input() -> RequestFuncInput:
-    return RequestFuncInput(
-        model="test-model",
-        model_name="test-model",
-        prompt="target text",
-        api_url="http://test.com/v1/chat/completions",
-        prompt_len=2,
-        output_len=20,
-        extra_body={"modalities": ["text", "audio"]},
-    )
-
-
-def test_seed_tts_chat_request_carries_reference_audio_once() -> None:
-    reference_audio_url = "data:audio/wav;base64,AAAA"
-    sample = SeedTTSSampleRequest(
-        prompt="target text",
-        prompt_len=2,
-        expected_output_len=20,
-        multi_modal_data=None,
-        seed_tts_system_prompt="Clone the supplied voice.",
-        seed_tts_speech_extra={
-            "ref_audio": reference_audio_url,
-            "ref_text": "reference text",
-            "task_type": "Base",
-        },
-    )
-    request_func_input = _seed_tts_request_func_input()
-
-    _attach_seed_tts_to_request_func_input(sample, request_func_input)
-
-    assert request_func_input.omni_chat_messages == [
-        {
-            "role": "system",
-            "content": [{"type": "text", "text": "Clone the supplied voice."}],
-        },
-        {"role": "user", "content": [{"type": "text", "text": "target text"}]},
-    ]
-    assert request_func_input.extra_body["ref_audio"] == reference_audio_url
-    serialized_request = json.dumps(
-        {"messages": request_func_input.omni_chat_messages, **request_func_input.extra_body},
-    )
-    assert serialized_request.count(reference_audio_url) == 1
-
-
-def test_seed_tts_text_chat_messages_do_not_add_reference_audio() -> None:
-    sample = SeedTTSTextSampleRequest(
-        prompt="target text",
-        prompt_len=2,
-        expected_output_len=20,
-        multi_modal_data=None,
-        seed_tts_system_prompt="Use the configured voice.",
-        seed_tts_speech_extra=None,
-    )
-    request_func_input = _seed_tts_request_func_input()
-
-    _attach_seed_tts_to_request_func_input(sample, request_func_input)
-
-    assert getattr(request_func_input, "seed_tts_row", False) is True
-    assert request_func_input.omni_chat_messages == [
-        {
-            "role": "system",
-            "content": [{"type": "text", "text": "Use the configured voice."}],
-        },
-        {"role": "user", "content": [{"type": "text", "text": "target text"}]},
-    ]
-    assert "ref_audio" not in request_func_input.extra_body
-
-
 class MockResponse:
    """Mock aiohttp response for testing"""

@@ -178,24 +102,7 @@ async def test_seed_tts_realtime_duplex_exports_per_request_metrics(monkeypatch)
                received_at_s=now + 0.03,
            )
            self.events.add(
-                {
-                    "type": "response.done",
-                    "response": {
-                        "id": response_id,
-                        "metadata": {
-                            "vllm_omni": {
-                                "stage_metrics": {
-                                    "0": {
-                                        "num_tokens_out": 3,
-                                        "vllm_ttft_ms": 20.0,
-                                        "vllm_tpot_ms": 10.0,
-                                        "vllm_itls_ms": [10.0, 10.0],
-                                    }
-                                }
-                            }
-                        },
-                    },
-                },
+                {"type": "response.done", "response": {"id": response_id}},
                received_at_s=now + 0.04,
            )

@@ -236,7 +143,6 @@ async def test_seed_tts_realtime_duplex_exports_per_request_metrics(monkeypatch)
    assert client.configure_kwargs["native_duplex"] is False
    assert client.configure_kwargs["extra_body"] == {
        "ref_audio": "data:audio/wav;base64,AAAA",
-        "return_stage_metrics": True,
    }
    assert client.sent == [
        event
@@ -261,10 +167,6 @@ async def test_seed_tts_realtime_duplex_exports_per_request_metrics(monkeypatch)
    assert output.audio_ttfp > output.ttft
    assert output.audio_rtf > 0
    assert output.latency > 0
-    assert output.output_tokens == 12
-    assert output.itl == [0.01, 0.01] * 4
-    assert output.text_latency == pytest.approx(output.ttft + 0.08)
-    assert output.tpot_measured is True
    assert output.tts_turn_pcm_bytes == [b"\x00\x00" * 2400] * 4
    assert output.tts_output_pcm_bytes == b"\x00\x00" * 9600
    session_id = output.duplex_request_metrics[0]["session_id"]
@@ -296,160 +198,11 @@ async def test_seed_tts_realtime_duplex_exports_per_request_metrics(monkeypatch)
    }


-def test_stage0_token_timings_use_weighted_tpot_when_itls_are_incomplete():
-    output = MixRequestFuncOutput(ttft=0.1)
-
-    measured = _apply_stage0_token_timings(
-        output,
-        [
-            {"output_token_count": 3, "itls_ms": [], "tpot_ms": 10.0},
-            {"output_token_count": 2, "itls_ms": [], "tpot_ms": 40.0},
-        ],
-        expected_output_tokens=5,
-    )
-
-    assert measured is True
-    assert output.itl == []
-    assert output.tpot_measured is True
-    assert (output.text_latency - output.ttft) / 4 == pytest.approx(0.02)
-
-
-def test_stage0_zero_itls_fall_back_to_positive_tpot():
-    output = MixRequestFuncOutput(ttft=0.1)
-
-    measured = _apply_stage0_token_timings(
-        output,
-        [{"output_token_count": 3, "itls_ms": [0.0, 0.0], "tpot_ms": 25.0}],
-        expected_output_tokens=3,
-    )
-
-    assert measured is True
-    assert output.itl == []
-    assert output.tpot_measured is True
-    assert (output.text_latency - output.ttft) / 2 == pytest.approx(0.025)
-
-
-def test_stage0_token_count_without_timing_is_not_measured():
-    output = MixRequestFuncOutput(ttft=0.1)
-
-    measured = _apply_stage0_token_timings(
-        output,
-        [{"output_token_count": 3, "itls_ms": [], "tpot_ms": None}],
-        expected_output_tokens=3,
-    )
-
-    assert measured is False
-    assert output.itl == []
-    assert output.text_latency == output.ttft
-    assert output.tpot_measured is False
-
-
 def create_sse_chunk(data_dict):
    """Helper to create SSE formatted chunk"""
    return f"data: {json.dumps(data_dict)}\n\n".encode()


-def test_chat_text_timing_metrics_request_stage_metrics():
-    args = Namespace(
-        backend="openai-chat-omni",
-        percentile_metrics="ttft,tpot,itl,e2el",
-        print_stage=False,
-        extra_body={},
-    )
-
-    assert should_request_stage_metrics(args) is True
-
-
-@pytest.mark.asyncio
-async def test_bundled_first_text_chunk_uses_stage0_token_timings(mocker: MockerFixture):
-    """Engine timings recover TPOT when every text token arrives together."""
-    request_input = RequestFuncInput(
-        model="test-model",
-        model_name="test-model",
-        prompt="test prompt",
-        api_url="http://test.com/v1/chat/completions",
-        prompt_len=10,
-        output_len=20,
-    )
-    chunks = [
-        create_sse_chunk(
-            {
-                "choices": [{"delta": {"content": "ABCD"}}],
-                "modality": "text",
-                "usage": {"prompt_tokens": 10, "completion_tokens": 4, "total_tokens": 14},
-                "metrics": {
-                    "stage_metrics": {
-                        "0": {
-                            "num_tokens_out": 4,
-                            "vllm_itls_ms": [10.0, 11.0, 12.0],
-                            "vllm_tpot_ms": 11.0,
-                        }
-                    }
-                },
-            }
-        ),
-        b"data: [DONE]\n\n",
-    ]
-    mock_response = MockResponse(200, chunks)
-    mock_session = mocker.AsyncMock()
-    mock_session.post = mocker.MagicMock(return_value=mock_response)
-
-    output = await async_request_openai_chat_omni_completions(request_input, mock_session)
-
-    assert output.success is True
-    assert output.output_tokens == 4
-    assert output.itl == pytest.approx([0.010, 0.011, 0.012])
-    assert output.text_latency - output.ttft == pytest.approx(0.033)
-    assert output.tpot_measured is True
-
-
-@pytest.mark.asyncio
-async def test_positive_client_text_timings_take_precedence_over_stage0(mocker: MockerFixture):
-    request_input = RequestFuncInput(
-        model="test-model",
-        model_name="test-model",
-        prompt="test prompt",
-        api_url="http://test.com/v1/chat/completions",
-        prompt_len=10,
-        output_len=20,
-    )
-    chunks = [
-        create_sse_chunk(
-            {
-                "choices": [{"delta": {"content": "A"}}],
-                "modality": "text",
-                "usage": {"prompt_tokens": 10, "completion_tokens": 1, "total_tokens": 11},
-            }
-        ),
-        create_sse_chunk(
-            {
-                "choices": [{"delta": {"content": "B"}}],
-                "modality": "text",
-                "usage": {"prompt_tokens": 10, "completion_tokens": 2, "total_tokens": 12},
-                "metrics": {
-                    "stage_metrics": {
-                        "0": {
-                            "num_tokens_out": 2,
-                            "vllm_itls_ms": [1000.0],
-                            "vllm_tpot_ms": 1000.0,
-                        }
-                    }
-                },
-            }
-        ),
-        b"data: [DONE]\n\n",
-    ]
-    mock_response = MockResponse(200, chunks, delay_between_chunks=0.01)
-    mock_session = mocker.AsyncMock()
-    mock_session.post = mocker.MagicMock(return_value=mock_response)
-
-    output = await async_request_openai_chat_omni_completions(request_input, mock_session)
-
-    assert output.success is True
-    assert len(output.itl) == 1
-    assert 0.0 < output.itl[0] < 0.1
-
-
 # ============================================================================
 # output_tokens Tests
 # ============================================================================
@@ -1,6 +1,3 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
-
 """Tests for Daily-Omni ``--daily-omni-pack-mode`` packing recipes.

 The ``minicpm-interleave`` mode is an accuracy-critical protocol: MiniCPM-o only reaches
@@ -30,8 +27,6 @@ _MODULE_NAME = "vllm_omni.benchmarks.data_modules.daily_omni_dataset"

 if _MODULE_NAME not in sys.modules:
    _spec = importlib.util.spec_from_file_location(_MODULE_NAME, _MODULE_PATH)
-    if _spec is None or _spec.loader is None:
-        raise ImportError(f"Cannot load {_MODULE_NAME} from {_MODULE_PATH}")
    _mod = importlib.util.module_from_spec(_spec)
    sys.modules[_MODULE_NAME] = _mod
    _spec.loader.exec_module(_mod)
@@ -92,9 +87,7 @@ def _stub_extract(monkeypatch, dataset: DailyOmniDataset, counter: list[int]) ->
    def fake_extract(video_path, *, audio_path, include_audio, **_kw):
        counter.append(1)
        frames = [Image.new("RGB", (16, 16), color=(i, i, i)) for i in range(_NUM_FRAMES)]
-        segments: list[np.ndarray] = (
-            [np.zeros(1600, dtype=np.float32) for _ in range(_NUM_FRAMES)] if include_audio else []
-        )
+        segments = [np.zeros(1600, dtype=np.float32) for _ in range(_NUM_FRAMES)] if include_audio else []
        return frames, segments

    monkeypatch.setattr(dataset, "_extract_minicpm_frame_audio_segments", fake_extract)
@@ -585,7 +585,7 @@ async def test_adapter_requires_and_forwards_exact_prepared_payload(tmp_path: Pa

@pytest.mark.asyncio
@pytest.mark.parametrize(
-    ("request_metrics", "expected_itl", "expected_text_latency", "expected_tpot_measured"),
+    ("request_metrics", "expected_itl", "expected_text_latency"),
    [
        (
            [
@@ -594,7 +594,6 @@ async def test_adapter_requires_and_forwards_exact_prepared_payload(tmp_path: Pa
            ],
            [0.01, 0.02, 0.03],
            0.16,
-            True,
        ),
        (
            [
@@ -603,7 +602,6 @@ async def test_adapter_requires_and_forwards_exact_prepared_payload(tmp_path: Pa
            ],
            [],
            0.18,
-            True,
        ),
        (
            [
@@ -612,7 +610,6 @@ async def test_adapter_requires_and_forwards_exact_prepared_payload(tmp_path: Pa
            ],
            [],
            0.1,
-            False,
        ),
    ],
 )
@@ -622,7 +619,6 @@ async def test_adapter_reports_exact_or_weighted_token_timing(
    request_metrics: list[dict[str, object]],
    expected_itl: list[float],
    expected_text_latency: float,
-    expected_tpot_measured: bool,
 ):
    case, options = _case(tmp_path), _session_options(tmp_path)
    request = _request(case, options)
@@ -647,7 +643,7 @@ async def test_adapter_reports_exact_or_weighted_token_timing(
    assert output.success
    assert output.itl == expected_itl
    assert output.text_latency == pytest.approx(expected_text_latency)
-    assert output.tpot_measured is expected_tpot_measured
+    assert output.tpot_measured is True


 def test_batch_finalization_publishes_only_measured_results(tmp_path: Path, monkeypatch: pytest.MonkeyPatch):
@@ -1,6 +1,3 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
-
 """Tests for SeedTTSTextDataset, SeedTTSTextSampleRequest, SeedTTSDesignDataset,
 and SeedTTSDesignSampleRequest.

@@ -237,6 +234,34 @@ def test_seed_tts_design_dataset_rejects_missing_description(seed_tts_design_roo
        assert req.seed_tts_utterance_id == "ok"


+def test_attach_sets_seed_tts_row_even_without_extra_body():
+    """seed_tts_row=True must be set for SeedTTSTextSampleRequest (no extra body)."""
+    from vllm_omni.benchmarks.data_modules.seed_tts_dataset import SeedTTSTextSampleRequest
+
+    req = SeedTTSTextSampleRequest(
+        prompt="hello world",
+        prompt_len=2,
+        expected_output_len=100,
+        multi_modal_data=None,
+        request_id="test-0",
+        seed_tts_speech_extra=None,
+        seed_tts_ref_wav_path="",
+    )
+    assert req.seed_tts_speech_extra is None
+    assert req.seed_tts_ref_wav_path == ""
+    # The fix ensures that even with speech_extra=None, the function
+    # sets seed_tts_row=True. We verify the source code has the fix.
+    import inspect
+
+    import vllm_omni.benchmarks.patch.patch as patch_mod
+
+    src = inspect.getsource(patch_mod._attach_seed_tts_to_request_func_input)
+    # seed_tts_row must be set BEFORE the 'if not ex: return' check
+    row_pos = src.index("seed_tts_row")
+    not_ex_pos = src.index("if not ex:")
+    assert row_pos < not_ex_pos, "seed_tts_row must be set before 'if not ex: return'"
+
+
 def test_seed_tts_whisper_transcribe_passes_attention_mask(monkeypatch):
    from vllm_omni.benchmarks.data_modules import seed_tts_eval

@@ -1,42 +0,0 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
-
-from pathlib import Path
-from shlex import split
-
-import pytest
-import yaml
-
-pytestmark = [pytest.mark.core_model, pytest.mark.cpu]
-
-AMD_MERGE_PIPELINE = Path(".buildkite/amd/test-amd-merge.yml")
-
-
-def _find_step(label: str) -> dict:
-    pipeline = yaml.safe_load(AMD_MERGE_PIPELINE.read_text(encoding="utf-8"))
-
-    def walk(steps: list[dict]) -> dict | None:
-        for step in steps:
-            if step.get("label") == label:
-                return step
-            if nested := walk(step.get("steps", [])):
-                return nested
-        return None
-
-    step = walk(pipeline.get("steps", []))
-    assert step is not None, f"missing AMD pipeline step: {label}"
-    return step
-
-
-def test_qwen3_tts_base_preserves_advanced_model_arguments() -> None:
-    step = _find_step("Qwen3-TTS Base E2E Test")
-    commands = step["commands"]
-
-    assert all("bash -c" not in command for command in commands)
-    pytest_command = next(command for command in commands if "pytest" in command)
-    argv = split(pytest_command)
-
-    marker_index = argv.index("-m")
-    run_level_index = argv.index("--run-level")
-    assert argv[marker_index + 1] == "advanced_model and cuda"
-    assert argv[run_level_index + 1] == "advanced_model"
@@ -34,18 +34,12 @@ from vllm_omni.config.stage_config import (
    build_stage_runtime_overrides,
    load_deploy_config,
    merge_pipeline_deploy,
-    normalize_pipeline_cli_overrides,
    pipeline_cfg_resolver,
 )
 from vllm_omni.engine.arg_utils import SHARED_FIELDS, internal_blacklist_keys

 pytestmark = [pytest.mark.core_model, pytest.mark.cpu]

-SINGLE_STAGE_PIPE_CFG = (StagePipelineConfig(stage_id=0, model_stage="a", final_output=True),)
-# Validation strings to match against in post init
-NO_STAGES_MATCH_STR = "no stages"
-NO_TERMINAL_STAGE_MATCH_STR = "No terminal stage"
-

@pytest.fixture(autouse=True)
 def _stable_test_platform(monkeypatch):
@@ -435,20 +429,6 @@ class TestStageConfigFactory:
        assert cfg["final_output"] is True
        assert cfg["final_output_type"] == "image"

-    @pytest.mark.parametrize(
-        "model_class_name",
-        [
-            "MiniMaxH3Pipeline",
-            "WanPipeline",
-            "LTX2Pipeline",
-            "Cosmos3OmniDiffusersPipeline",
-        ],
-    )
-    def test_default_video_diffusion_output_type(self, model_class_name):
-        configs = StageConfigFactory.create_default_diffusion({"model_class_name": model_class_name})
-
-        assert configs[0]["final_output_type"] == "video"
-
    def test_default_diffusion_with_parallel_config(self):
        """Test diffusion config calculates devices from parallel_config."""

@@ -623,7 +603,6 @@ class TestPipelineDiscovery:
        p = PipelineConfig(
            model_type="custom_collide",
            hf_architectures=("SomeCollidingArch",),
-            stages=SINGLE_STAGE_PIPE_CFG,
        )
        assert p.hf_architectures == ("SomeCollidingArch",)

@@ -645,63 +624,33 @@ class TestStagePipelineConfig:


 class TestPipelineConfigNew:
-    def test_simple_init(self):
-        # Ensure pipeline config validates in post init.
-        PipelineConfig(
-            model_type="t",
-            model_arch="A",
-            stages=SINGLE_STAGE_PIPE_CFG,
-        )
-
    def test_frozen(self):
-        p = PipelineConfig(
-            model_type="t",
-            model_arch="A",
-            stages=SINGLE_STAGE_PIPE_CFG,
-        )
+        p = PipelineConfig(model_type="t", model_arch="A")
        with pytest.raises(AttributeError):
            p.model_type = "changed"

-    def test_validate_no_stages(self):
-        with pytest.raises(ValueError, match=NO_STAGES_MATCH_STR):
-            PipelineConfig(model_type="t", model_arch="A")
-
-    def test_validate_no_terminal_stage(self):
-        """A pipeline with no ``final_output`` stage can never emit a result."""
-        with pytest.raises(ValueError, match=NO_TERMINAL_STAGE_MATCH_STR):
-            PipelineConfig(
-                model_type="t",
-                model_arch="A",
-                stages=(
-                    StagePipelineConfig(stage_id=0, model_stage="a"),
-                    StagePipelineConfig(stage_id=1, model_stage="b", input_sources=(0,)),
-                ),
-            )
-
-    def test_pipeline_can_have_final_output_in_any_stage(self):
-        """Ensure any stage may carry ``final_output``."""
-        PipelineConfig(
+    def test_validate_valid(self):
+        p = PipelineConfig(
            model_type="t",
            model_arch="A",
            stages=(
-                StagePipelineConfig(stage_id=0, model_stage="a", final_output=True),
+                StagePipelineConfig(stage_id=0, model_stage="a"),
                StagePipelineConfig(stage_id=1, model_stage="b", input_sources=(0,)),
            ),
        )
+        assert p.validate() == []
+
+    def test_validate_no_stages(self):
+        p = PipelineConfig(model_type="t", model_arch="A")
+        assert any("no stages" in e.lower() for e in p.validate())


 class TestPipelineRegistration:
    def test_resolve_pipeline_prefers_deploy_pipeline_key(self, clean_pipeline_registry, tmp_path):
        deploy_key = "deploy_selected_pipeline"
        model_type_key = "hf_model_type_pipeline"
-        deploy_pipe = PipelineConfig(
-            model_type=deploy_key,
-            stages=SINGLE_STAGE_PIPE_CFG,
-        )
-        model_type_pipe = PipelineConfig(
-            model_type=model_type_key,
-            stages=SINGLE_STAGE_PIPE_CFG,
-        )
+        deploy_pipe = PipelineConfig(model_type=deploy_key)
+        model_type_pipe = PipelineConfig(model_type=model_type_key)
        register_pipeline(deploy_pipe)
        register_pipeline(model_type_pipe)

@@ -727,11 +676,7 @@ class TestPipelineRegistration:
        tmp_path,
    ):
        model_type_key = "registered_model_type_pipeline"
-        pipeline_cfg = PipelineConfig(
-            model_type=model_type_key,
-            stages=SINGLE_STAGE_PIPE_CFG,
-        )
-        register_pipeline(pipeline_cfg)
+        register_pipeline(PipelineConfig(model_type=model_type_key))

        deploy_path = tmp_path / "deploy.yaml"
        deploy_path.write_text("pipeline: missing_pipeline\n", encoding="utf-8")
@@ -766,7 +711,6 @@ class TestPipelineRegistration:
        pipe_cfg = PipelineConfig(
            model_type=pipeline_key,
            hf_architectures=("ArchitectureFallbackForTest",),
-            stages=SINGLE_STAGE_PIPE_CFG,
        )
        register_pipeline(pipe_cfg)

@@ -797,18 +741,15 @@ class TestPipelineRegistration:
            model_type="rejected_by_predicate",
            hf_architectures=(shared_arch,),
            hf_config_predicate=rejecting_predicate,
-            stages=SINGLE_STAGE_PIPE_CFG,
        )
        raises_cfg = PipelineConfig(
            model_type="raises_in_predicate",
            hf_architectures=(shared_arch,),
            hf_config_predicate=raising_predicate,
-            stages=SINGLE_STAGE_PIPE_CFG,
        )
        accept_cfg = PipelineConfig(
            model_type="accepted_by_predicate",
            hf_architectures=(shared_arch,),
-            stages=SINGLE_STAGE_PIPE_CFG,
        )
        register_pipeline(reject_cfg)
        register_pipeline(raises_cfg)
@@ -842,7 +783,6 @@ class TestPipelineRegistration:
            model_type="predicate_without_arch_match",
            hf_architectures=("DifferentArchitectureForTest",),
            hf_config_predicate=predicate,
-            stages=SINGLE_STAGE_PIPE_CFG,
        )
        register_pipeline(pipe_cfg)

@@ -866,7 +806,6 @@ class TestPipelineRegistration:
        resolved_cfg = PipelineConfig(
            model_type="callable_resolved_pipeline",
            hf_architectures=("CallableArchitectureForTest",),
-            stages=SINGLE_STAGE_PIPE_CFG,
        )
        seen_hf_configs = []

@@ -940,10 +879,7 @@ class TestPipelineRegistration:
    def test_pipeline_registration(self, clean_pipeline_registry):
        """Ensure that we can register and create a custom pipeline config."""
        new_model_type = "new_model_type"
-        pipe_cfg = PipelineConfig(
-            model_type=new_model_type,
-            stages=SINGLE_STAGE_PIPE_CFG,
-        )
+        pipe_cfg = PipelineConfig(model_type=new_model_type)

        # Register the new PipelineConfig
        assert new_model_type not in OMNI_PIPELINES
@@ -986,10 +922,7 @@ class TestPipelineRegistration:
        def custom_resolver(
            hf_config: FakeConfig,
        ) -> PipelineConfig:
-            return PipelineConfig(
-                model_type=resolved_type,
-                stages=SINGLE_STAGE_PIPE_CFG,
-            )
+            return PipelineConfig(model_type=resolved_type)

        # Register the new PipelineConfig
        assert new_model_type not in OMNI_PIPELINES
@@ -1046,7 +979,7 @@ class TestPipelineRegistration:
        pipe_cfg = PipelineConfig(
            model_type=pipeline_key,
            model_arch="DeployOnlyArch",
-            stages=SINGLE_STAGE_PIPE_CFG,
+            stages=(StagePipelineConfig(stage_id=0, model_stage="diffusion"),),
        )
        register_pipeline(pipe_cfg)

@@ -1079,11 +1012,7 @@ class TestPipelineRegistration:

    def test_structured_path_loads_explicit_deploy_config_once(self, clean_pipeline_registry, tmp_path):
        pipeline_key = "single_load_pipeline"
-        pipeline_cfg = PipelineConfig(
-            model_type=pipeline_key,
-            stages=SINGLE_STAGE_PIPE_CFG,
-        )
-        register_pipeline(pipeline_cfg)
+        register_pipeline(PipelineConfig(model_type=pipeline_key))
        deploy_path = tmp_path / "single_load.yaml"
        deploy_path.write_text(f"pipeline: {pipeline_key}\n", encoding="utf-8")

@@ -1113,7 +1042,6 @@ class TestPipelineRegistration:
        pipeline = PipelineConfig(
            model_type="pipeline_with_default",
            default_deploy_config_name=default_name,
-            stages=SINGLE_STAGE_PIPE_CFG,
        )

        with patch(
@@ -1133,16 +1061,8 @@ class TestPipelineRegistration:
            OmniServingCapability.COMPLETIONS,
            "pipeline_a blocks completions",
        )
-        pipe_a = PipelineConfig(
-            model_type="detect_type",
-            endpoint_restrictions=(restriction,),
-            stages=SINGLE_STAGE_PIPE_CFG,
-        )
-        pipe_b = PipelineConfig(
-            model_type="override_type",
-            endpoint_restrictions=(),
-            stages=SINGLE_STAGE_PIPE_CFG,
-        )
+        pipe_a = PipelineConfig(model_type="detect_type", endpoint_restrictions=(restriction,))
+        pipe_b = PipelineConfig(model_type="override_type", endpoint_restrictions=())
        register_pipeline(pipe_a)
        register_pipeline(pipe_b)

@@ -1224,7 +1144,7 @@ class TestDeployConfigLoading:
        stages = merge_pipeline_deploy(pipeline, deploy)

        assert deploy.session_mode == "duplex"
-        assert deploy.active_stream_window == max_sessions
+        assert deploy.active_stream_window == 1
        assert deploy.duplex_session.max_sessions == max_sessions
        assert [stage.session_mode for stage in stages] == ["duplex", "duplex", "duplex"]
        assert [stage.to_omegaconf().session_mode for stage in stages] == ["duplex", "duplex", "duplex"]
@@ -1404,13 +1324,6 @@ stages:
            ("step_audio2_ci.yaml", "step_audio_2_asr", 1, "text", "step_audio_2_asr"),
            ("hunyuan_video_15.yaml", "hunyuan_video_15", 1, "video", None),
            ("wan2_2_ti2v.yaml", "wan2_2_ti2v", 1, "video", None),
-            (
-                "minimax_h3_disaggregated.yaml",
-                "minimax_h3_disaggregated",
-                2,
-                "video",
-                "minimax_h3_disaggregated",
-            ),
        ],
    )
    def test_load_new_registry_backed_deploy_configs(
@@ -1431,91 +1344,9 @@ stages:
        assert len(stages) == stage_count
        assert stages[-1].final_output is True
        assert stages[-1].final_output_type == final_output_type
-        if pipeline_name == "minimax_h3_disaggregated":
-            assert stages[0].yaml_engine_args["model_path_resolver"].endswith(".resolve_minimax_h3_model_root")
-            assert (
-                stages[1].yaml_engine_args["model_path_resolver"].endswith(".resolve_minimax_h3_diffusion_model_path")
-            )
-            assert "model_path_resolver" not in stages[0].yaml_extras
-            assert "model_path_resolver" not in stages[1].yaml_extras
        if deploy.trust_remote_code is not None:
            assert {s.yaml_engine_args.get("trust_remote_code") for s in stages} == {deploy.trust_remote_code}

-    def test_minimax_h3_resolvers_reach_production_engine_args(self, tmp_path: Path, monkeypatch):
-        from vllm_omni.engine import stage_init_utils
-
-        model_root = tmp_path / "MiniMax-H3"
-        (model_root / "FL2VA" / "text_encoder").mkdir(parents=True)
-        (model_root / "Ref2VA" / "text_encoder").mkdir(parents=True)
-        monkeypatch.setattr(stage_init_utils, "resolve_worker_cls", lambda _engine_args: None)
-
-        deploy = load_deploy_config(get_deploy_config_path("minimax_h3_disaggregated.yaml"))
-        assert deploy.stages[1].engine_extras["model_loaded"] == {"text_encoder": False}
-        stages = merge_pipeline_deploy(OMNI_PIPELINES["minimax_h3_disaggregated"], deploy)
-        resolved = [stage_init_utils.build_engine_args_dict(stage.to_omegaconf(), str(model_root)) for stage in stages]
-
-        assert resolved[0]["model"] == str(model_root / "FL2VA" / "text_encoder")
-        assert resolved[1]["model"] == str(model_root / "FL2VA")
-        assert all("model_path_resolver" not in engine_args for engine_args in resolved)
-
-    def test_minimax_h3_text_encoder_tp_reaches_only_stage_zero(self):
-        pipeline = OMNI_PIPELINES["minimax_h3_disaggregated"]
-        stages, _ = StageConfigFactory._create_legacy_from_registry(
-            pipeline,
-            {"text_encoder_tp_size": 4},
-        )
-
-        assert stages[0].runtime_overrides["tensor_parallel_size"] == 4
-        assert "text_encoder_tp_size" not in stages[0].runtime_overrides
-        assert "text_encoder_tp_size" not in stages[1].runtime_overrides
-        assert stages[1].yaml_engine_args["parallel_config"]["tensor_parallel_size"] == 1
-
-    def test_minimax_h3_disaggregated_defaults_match_validated_topology(self):
-        pipeline = OMNI_PIPELINES["minimax_h3_disaggregated"]
-        deploy = load_deploy_config(Path(get_deploy_config_path("minimax_h3_disaggregated.yaml")))
-        stages = merge_pipeline_deploy(pipeline, deploy)
-
-        assert stages[0].yaml_engine_args["max_num_seqs"] == 1
-        assert stages[0].yaml_engine_args["model_path_resolver"].endswith(".resolve_minimax_h3_model_root")
-        assert stages[1].yaml_engine_args["model_path_resolver"].endswith(".resolve_minimax_h3_diffusion_model_path")
-        parallel = stages[1].yaml_engine_args["parallel_config"]
-        assert parallel["tensor_parallel_size"] == 1
-        assert parallel["ulysses_degree"] == 4
-        assert parallel["vae_patch_parallel_size"] == 4
-        assert stages[1].yaml_engine_args["inline_diffusion"] is True
-        assert stages[1].yaml_extras["default_sampling_params"]["num_inference_steps"] == 50
-
-        turbo = load_deploy_config(Path(get_deploy_config_path("minimax_h3_disaggregated_turbo.yaml")))
-        turbo_stages = merge_pipeline_deploy(pipeline, turbo)
-        turbo_sampling = turbo_stages[1].yaml_extras["default_sampling_params"]
-        assert turbo_sampling["num_inference_steps"] == 5
-        assert turbo_sampling["extra_args"] == {"flow_shift": 6.0, "audio_flow_shift": 3.0}
-
-    def test_minimax_h3_text_encoder_tp_alias_targets_stage_zero(self):
-        pipeline = OMNI_PIPELINES["minimax_h3_disaggregated"]
-
-        assert normalize_pipeline_cli_overrides(pipeline, {}) == {}
-        assert normalize_pipeline_cli_overrides(pipeline, {"text_encoder_tp_size": 4}) == {
-            "stage_0_tensor_parallel_size": 4
-        }
-
-    def test_minimax_h3_stage_zero_tp_override_wins_alias_conflict(self):
-        pipeline = OMNI_PIPELINES["minimax_h3_disaggregated"]
-
-        with pytest.warns(UserWarning, match="stage_0_tensor_parallel_size=2 takes precedence"):
-            normalized = normalize_pipeline_cli_overrides(
-                pipeline,
-                {"text_encoder_tp_size": 4, "stage_0_tensor_parallel_size": 2},
-            )
-
-        assert normalized == {"stage_0_tensor_parallel_size": 2}
-
-    def test_minimax_h3_rejects_text_encoder_tp_on_diffusion_stage(self):
-        pipeline = OMNI_PIPELINES["minimax_h3_disaggregated"]
-
-        with pytest.raises(ValueError, match="stage_1_text_encoder_tp_size cannot be set"):
-            normalize_pipeline_cli_overrides(pipeline, {"stage_1_text_encoder_tp_size": 4})
-
    @pytest.mark.parametrize(
        ("config_json", "model_index", "expected_pipeline"),
        [
@@ -1551,29 +1382,6 @@ stages:
        assert pipeline is not None
        assert pipeline.model_type == expected_pipeline

-    def test_minimax_h3_disaggregation_is_not_auto_discovered(self):
-        def get_model_file(filename, _model, revision=None):
-            del revision
-            if filename == "config.json":
-                return None
-            if filename == "model_index.json":
-                return {"_class_name": "MiniMaxH3Pipeline"}
-            return None
-
-        with (
-            patch.object(StageConfigFactory, "get_hf_config", return_value=None),
-            patch(
-                "vllm_omni.config.config_factory.get_hf_file_to_dict",
-                side_effect=get_model_file,
-            ),
-        ):
-            pipeline = StageConfigFactory.get_pipeline_config(
-                model="/models/unrelated-checkpoint-name",
-                trust_remote_code=False,
-            )
-
-        assert pipeline is None
-
    @pytest.mark.parametrize("deploy_name", ["step_audio_2.yaml", "step_audio_2_async_chunk.yaml"])
    def test_step_audio2_deploy_configs_fit_two_gpus(self, deploy_name):
        deploy = load_deploy_config(Path(get_deploy_config_path(deploy_name)))
@@ -1682,7 +1490,6 @@ stages:
                    model_stage="ar",
                    execution_type=StageExecutionType.LLM_AR,
                    requires_multimodal_data=True,
-                    final_output=True,
                ),
            ),
        )
@@ -1703,7 +1510,6 @@ stages:
                    model_stage="ar",
                    execution_type=StageExecutionType.LLM_AR,
                    scheduler_cls=scheduler_cls,
-                    final_output=True,
                ),
            ),
        )
@@ -1949,6 +1755,7 @@ class TestQwen3OmniPipeline:
        assert isinstance(p, PipelineConfig)
        assert p.model_arch == "Qwen3OmniMoeForConditionalGeneration"
        assert len(p.stages) == 3
+        assert p.validate() == []

    def test_thinker(self):
        p = resolve_pipeline_config(
@@ -2002,6 +1809,7 @@ class TestQwen2_5OmniPipeline:
        assert isinstance(p, PipelineConfig)
        assert p.model_arch == "Qwen2_5OmniForConditionalGeneration"
        assert len(p.stages) == 3
+        assert p.validate() == []

    def test_thinker(self):
        p = resolve_pipeline_config("qwen2_5_omni")
@@ -2050,6 +1858,7 @@ class TestQwen3TTSPipeline:
        assert p is not None
        assert p.model_arch == "Qwen3TTSTalkerForConditionalGeneration"
        assert len(p.stages) == 2
+        assert p.validate() == []

    def test_talker_stage(self):
        p = resolve_pipeline_config("qwen3_tts")
@@ -2129,6 +1938,7 @@ class TestMingFlashOmniPipeline:
        assert isinstance(p, PipelineConfig)
        assert p.model_arch == "MingFlashOmniForConditionalGeneration"
        assert len(p.stages) == 2
+        assert p.validate() == []

    def test_thinker_stage(self):
        p = resolve_pipeline_config("ming_flash_omni")
@@ -2185,6 +1995,7 @@ class TestMingFlashOmniPipeline:
        assert isinstance(p, PipelineConfig)
        assert p.model_arch == "MingFlashOmniTalkerForConditionalGeneration"
        assert len(p.stages) == 1
+        assert p.validate() == []

    def test_tts_stage(self):
        p = resolve_pipeline_config("ming_flash_omni_tts")
@@ -2241,6 +2052,7 @@ class TestMingFlashOmniPipeline:
        assert isinstance(p, PipelineConfig)
        assert p.model_arch == "MingFlashOmniForConditionalGeneration"
        assert len(p.stages) == 1
+        assert p.validate() == []

    def test_thinker_only_stage(self):
        p = resolve_pipeline_config("ming_flash_omni_thinker_only")
@@ -2279,6 +2091,7 @@ class TestMingFlashOmniPipeline:
        assert p is not None
        assert p.model_arch == "MingFlashOmniForConditionalGeneration"
        assert len(p.stages) == 2
+        assert p.validate() == []

    def test_image_thinker_stage(self):
        s = resolve_pipeline_config("ming_flash_omni_image").get_stage(0)
@@ -2399,21 +2212,6 @@ class TestBaseConfigInheritance:
        # CI overrides max_tokens
        assert s0["max_tokens"] == 150

-    def test_qwen3_omni_colocate_async_bounds_only_rocm_kv_cache(self):
-        ci_path = Path(get_deploy_config_path("ci/qwen3_omni_moe_colocate_async.yaml"))
-        pipeline = resolve_pipeline_config("qwen3_omni_moe_thinker_only")
-        assert isinstance(pipeline, PipelineConfig)
-
-        cuda = _apply_platform_overrides(load_deploy_config(ci_path), platform="cuda")
-        cuda_stage = merge_pipeline_deploy(pipeline, cuda)[0]
-        assert cuda_stage.yaml_engine_args["gpu_memory_utilization"] == 0.9
-        assert "kv_cache_memory_bytes" not in cuda_stage.yaml_engine_args
-
-        rocm = _apply_platform_overrides(load_deploy_config(ci_path), platform="rocm")
-        rocm_stage = merge_pipeline_deploy(pipeline, rocm)[0]
-        assert "gpu_memory_utilization" not in rocm_stage.yaml_engine_args
-        assert rocm_stage.yaml_engine_args["kv_cache_memory_bytes"] == 2 * 1024**3
-
    def test_pure_inheritance_overlay(self, tmp_path):
        """An overlay with only ``base_config`` inherits everything."""
        base = Path(get_deploy_config_path("qwen3_omni_moe.yaml"))
@@ -2474,20 +2272,6 @@ class TestPlatformOverrides:
        assert rocm.stages[0].enforce_eager is None
        assert rocm.stages[1].enforce_eager is True

-    def test_higgs_audio_v3_rocm_uses_triton_attention(self):
-        deploy_path = Path(get_deploy_config_path("higgs_multimodal_qwen3.yaml"))
-
-        base = load_deploy_config(deploy_path)
-        assert base.stages[0].engine_extras["attention_backend"] == "FLASHINFER"
-
-        rocm = _apply_platform_overrides(base, platform="rocm")
-        assert rocm.stages[0].engine_extras["attention_backend"] == "TRITON_ATTN"
-
-        pipeline = resolve_pipeline_config("higgs_multimodal_qwen3")
-        assert isinstance(pipeline, PipelineConfig)
-        stages = merge_pipeline_deploy(pipeline, rocm)
-        assert stages[0].yaml_engine_args["attention_backend"] == "TRITON_ATTN"
-
    def test_qwen3_omni_cuda_uses_thinker_rotary_custom_op(self):
        deploy_path = Path(get_deploy_config_path("qwen3_omni_moe.yaml"))
        pipeline = resolve_pipeline_config(
@@ -2875,54 +2659,6 @@ stages:
        assert omega_config.engine_args.max_model_len == 8192
        assert omega_config.engine_args.enforce_eager is True

-    def test_cli_attention_shorthand_replaces_yaml_structured_default(self):
-        stage = StageConfig(
-            stage_id=0,
-            model_stage="dit",
-            stage_type=StageType.DIFFUSION,
-            yaml_engine_args={
-                "diffusion_attention_config": {
-                    "default": {"backend": "FLASH_ATTN"},
-                    "per_role": {"cross": {"backend": "TORCH_SDPA"}},
-                },
-            },
-            runtime_overrides={"diffusion_attention_backend": "SAGE_ATTN"},
-        )
-
-        engine_args = stage.to_omegaconf().engine_args
-
-        assert engine_args.diffusion_attention_backend == "SAGE_ATTN"
-        assert "default" not in engine_args.diffusion_attention_config
-        assert engine_args.diffusion_attention_config.per_role.cross.backend == "TORCH_SDPA"
-
-    def test_cli_attention_shorthand_keeps_yaml_per_role_only_config(self):
-        stage = StageConfig(
-            stage_id=0,
-            model_stage="dit",
-            stage_type=StageType.DIFFUSION,
-            yaml_engine_args={"diffusion_attention_config": {"per_role": {"cross": {"backend": "TORCH_SDPA"}}}},
-            runtime_overrides={"diffusion_attention_backend": "SAGE_ATTN"},
-        )
-
-        engine_args = stage.to_omegaconf().engine_args
-
-        assert engine_args.diffusion_attention_backend == "SAGE_ATTN"
-        assert engine_args.diffusion_attention_config == {"per_role": {"cross": {"backend": "TORCH_SDPA"}}}
-
-    def test_cli_attention_config_replaces_yaml_shorthand(self):
-        stage = StageConfig(
-            stage_id=0,
-            model_stage="dit",
-            stage_type=StageType.DIFFUSION,
-            yaml_engine_args={"diffusion_attention_backend": "TORCH_SDPA"},
-            runtime_overrides={"diffusion_attention_config": {"default": {"backend": "SAGE_ATTN"}}},
-        )
-
-        engine_args = stage.to_omegaconf().engine_args
-
-        assert "diffusion_attention_backend" not in engine_args
-        assert engine_args.diffusion_attention_config.default.backend == "SAGE_ATTN"
-

 class TestSentinelDefaultPrecedence:
    """Caller-typed (non-None) values win over YAML; None values fall through
@@ -167,7 +167,7 @@ def test_inventory_matches_reviewed_snapshot_counts():
        EnvironmentVariableCategory.PUBLIC_OMNI: 24,
        EnvironmentVariableCategory.INHERITED_VLLM: 20,
        EnvironmentVariableCategory.PLATFORM_EXTERNAL: 27,
-        EnvironmentVariableCategory.MODEL_SPECIFIC: 57,
+        EnvironmentVariableCategory.MODEL_SPECIFIC: 56,
        EnvironmentVariableCategory.BENCHMARK_TRANSITIONAL: 20,
        EnvironmentVariableCategory.INTERNAL: 2,
    }
@@ -178,7 +178,7 @@ def test_inventory_matches_reviewed_snapshot_counts():
        if item.category is EnvironmentVariableCategory.MODEL_SPECIFIC
    )
    assert {disposition: disposition_counts[disposition] for disposition in ModelEnvironmentVariableDisposition} == {
-        ModelEnvironmentVariableDisposition.PROMOTE: 34,
+        ModelEnvironmentVariableDisposition.PROMOTE: 33,
        ModelEnvironmentVariableDisposition.REQUEST_SCOPE: 6,
        ModelEnvironmentVariableDisposition.EXTERNAL: 0,
        ModelEnvironmentVariableDisposition.INTERNALIZE: 11,
@@ -45,7 +45,6 @@ from vllm_omni.config.stage_config import (
    PipelineConfig,
    StageDeployConfig,
    StageExecutionType,
-    StagePipelineConfig,
    load_deploy_config,
    merge_pipeline_deploy,
 )
@@ -93,23 +92,10 @@ def _from_pipeline_key(
    )


-def test_minimax_h3_text_encoder_tp_targets_only_structured_stage_zero() -> None:
-    config = _from_pipeline_key(
-        "minimax_h3_disaggregated",
-        cli_overrides={"text_encoder_tp_size": 4},
-    )
-
-    assert config.stage_by_id(0).parallel_config.tensor_parallel_size == 4
-    diffusion = config.stage_by_id(1)
-    assert diffusion.parallel_config.tensor_parallel_size == 1
-    assert diffusion.parallel_config.text_encoder_tp_size == 1
-
-
 def test_duplex_session_capacity_propagates_to_every_structured_stage() -> None:
    omni_config = _from_pipeline_key("personaplex")

    assert [stage.model_config.duplex_max_sessions for stage in omni_config.stage_configs] == [2, 2]
-    assert [stage.model_config.session_mode for stage in omni_config.stage_configs] == ["duplex", "duplex"]


 def test_non_duplex_deploy_keeps_model_session_capacity_at_one(tmp_path: Path) -> None:
@@ -119,7 +105,6 @@ def test_non_duplex_deploy_keeps_model_session_capacity_at_one(tmp_path: Path) -
    omni_config = _from_pipeline_key("personaplex", deploy_config_path=str(deploy_path))

    assert [stage.model_config.duplex_max_sessions for stage in omni_config.stage_configs] == [1, 1]
-    assert [stage.model_config.session_mode for stage in omni_config.stage_configs] == ["turn", "turn"]


@pytest.mark.parametrize("model_type", sorted(OMNI_PIPELINES))
@@ -147,7 +132,6 @@ def test_vllm_omni_config_from_pipeline_config_matches_merge_pipeline_deploy(mod

        engine_args = legacy_stage.yaml_engine_args
        assert omni_stage.model_config.duplex_max_sessions == engine_args.get("duplex_max_sessions", 1)
-        assert omni_stage.model_config.session_mode == engine_args.get("session_mode", "turn")
        assert omni_stage.model_config.enforce_eager == engine_args.get("enforce_eager", False)
        assert omni_stage.load_config.load_format == engine_args.get("load_format", "auto")
        assert omni_stage.load_config.tokenizer_mode == engine_args.get("tokenizer_mode", "auto")
@@ -595,7 +579,6 @@ def test_sub_config_fields_match_structured_scopes():
        "interleave_mm_strings",
        "media_io_kwargs",
        "active_stream_window",
-        "session_mode",
        "duplex_max_sessions",
        "enable_sleep_mode",
        "default_sampling_params",
@@ -654,11 +637,10 @@ def test_sub_config_fields_match_structured_scopes():
        "ring_degree",
        "allgather_degree",
        "ulysses_mode",
-        "ulysses_a2a_permute",
        "cfg_parallel_size",
        "vae_patch_parallel_size",
-        "vae_parallel_mode",
        "text_encoder_tp_size",
+        "vae_parallel_mode",
        "use_hsdp",
        "mask_sp_padding",
        "hsdp_shard_size",
@@ -860,23 +842,6 @@ def test_from_pipeline_config_derives_sequence_parallel_size_from_allgather_degr
    assert stage.parallel_config.world_size == 2


-def test_from_pipeline_config_preserves_deployed_ulysses_a2a_permute() -> None:
-    pipeline = _resolve_pipeline_or_skip("dreamzero")
-    deploy = DeployConfig(
-        stages=[
-            StageDeployConfig(
-                stage_id=0,
-                ulysses_a2a_permute=True,
-            )
-        ]
-    )
-
-    stage = VllmOmniConfig.from_pipeline_config(pipeline, user_deploy_config=deploy).stage_by_id(0)
-
-    assert isinstance(stage, VllmOmniDiffusionStageConfig)
-    assert stage.parallel_config.ulysses_a2a_permute is True
-
-
 def test_diffusion_parallel_config_accepts_four_way_guidance_parallelism():
    cfg = OmniStageDiffusionParallelConfig(cfg_parallel_size=4)

@@ -971,35 +936,6 @@ def test_structured_diffusion_config_rejects_non_boolean_compile_dynamic():
        omni_config_module._DiffusionConfigProjection(diffusion_compile_dynamic="false")


-def test_from_pipeline_config_routes_ltx2_conv_vae_extra(tmp_path):
-    deploy_path = tmp_path / "dreamzero_ltx2_extras.yaml"
-    deploy_path.write_text(
-        "\n".join(
-            [
-                "pipeline: dreamzero",
-                "async_chunk: false",
-                "stages:",
-                "  - stage_id: 0",
-                "    extras:",
-                "      ltx2_use_conv_vae: true",
-            ]
-        )
-    )
-
-    stage = _from_pipeline_key("dreamzero", deploy_config_path=str(deploy_path)).stage_by_id(0)
-
-    assert stage.diffusion_config.extras["ltx2_use_conv_vae"] is True
-
-
-def test_stage_override_routes_ltx2_conv_vae_extra():
-    stage = _from_pipeline_key(
-        "dreamzero",
-        cli_overrides={"stage_0_extras": {"ltx2_use_conv_vae": True}},
-    ).stage_by_id(0)
-
-    assert stage.diffusion_config.extras["ltx2_use_conv_vae"] is True
-
-
 def test_structured_diffusion_config_rejects_invalid_compile_granularity():
    with pytest.raises(ValidationError, match="diffusion_compile_granularity"):
        omni_config_module._DiffusionConfigProjection(diffusion_compile_granularity="block")
@@ -1040,10 +976,7 @@ def test_from_pipeline_config_uses_hf_config_for_callable_resolver():


 def test_from_pipeline_config_accepts_pre_resolved_pipeline():
-    resolved_pipeline = PipelineConfig(
-        model_type="callable_resolved_variant",
-        stages=(StagePipelineConfig(stage_id=0, model_stage="a", final_output=True),),
-    )
+    resolved_pipeline = PipelineConfig(model_type="callable_resolved_variant")

    omni_config = VllmOmniConfig.from_pipeline_config(resolved_pipeline)

@@ -1076,7 +1009,6 @@ def test_from_pipeline_config_default_deploy_name_ignores_cwd(monkeypatch, tmp_p
    pipeline = PipelineConfig(
        model_type="pipeline_with_default",
        default_deploy_config_name=default_name,
-        stages=(StagePipelineConfig(stage_id=0, model_stage="a", final_output=True),),
    )
    loaded_paths = []

@@ -1,22 +1,16 @@
 # SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
 """Tests for out of tree registration to OMNI_PIPELINES."""

 import pytest
 from transformers import PretrainedConfig

 from vllm_omni.config.pipeline_registry import OMNI_PIPELINES, register_pipeline
-from vllm_omni.config.stage_config import PipelineConfig, StagePipelineConfig, pipeline_cfg_resolver
+from vllm_omni.config.stage_config import PipelineConfig, pipeline_cfg_resolver

 pytestmark = [pytest.mark.core_model, pytest.mark.cpu]


-def build_fake_pipeline_config(model_type: str) -> PipelineConfig:
-    return PipelineConfig(
-        model_type=model_type, stages=(StagePipelineConfig(stage_id=0, model_stage="a", final_output=True),)
-    )
-
-
@pytest.fixture
 def custom_resolver():
    """Build a reusable custom resolver for PipelineConfigs."""
@@ -28,7 +22,7 @@ def custom_resolver():
    def custom_resolver(
        hf_config: CustomConfigType,
    ) -> PipelineConfig:
-        return build_fake_pipeline_config("resolved_type")
+        return PipelineConfig(model_type="resolved_type")

    return custom_resolver

@@ -36,7 +30,7 @@ def custom_resolver():
 def test_register_pipeline_config(clean_pipeline_registry):
    """Ensure that we can register a custom pipeline config to OMNI_PIPELINES."""
    new_model_type = "new_model_type"
-    pipe_cfg = build_fake_pipeline_config(new_model_type)
+    pipe_cfg = PipelineConfig(model_type=new_model_type)
    assert new_model_type not in OMNI_PIPELINES
    register_pipeline(pipe_cfg)
    assert new_model_type in OMNI_PIPELINES
@@ -47,7 +41,7 @@ def test_register_pipeline_config_with_model_type(clean_pipeline_registry):
    """Ensure that we can register a custom pipeline config with an explicit model_type to OMNI_PIPELINES."""
    new_model_type = "new_model_type"
    unused_model_type = "foo"
-    pipe_cfg = build_fake_pipeline_config(unused_model_type)
+    pipe_cfg = PipelineConfig(model_type=unused_model_type)
    assert new_model_type not in OMNI_PIPELINES
    assert unused_model_type not in OMNI_PIPELINES

@@ -72,10 +66,3 @@ def test_register_resolver_requires_model_type(custom_resolver, clean_pipeline_r
    """Ensure that registering a custom resolver to OMNI_PIPELINES requires an explicit model_type."""
    with pytest.raises(ValueError):
        register_pipeline(custom_resolver)
-
-
-def test_minimax_h3_disaggregation_is_explicit_opt_in():
-    assert "minimax_h3" not in OMNI_PIPELINES
-    pipeline = OMNI_PIPELINES["minimax_h3_disaggregated"]
-    assert isinstance(pipeline, PipelineConfig)
-    assert pipeline.model_type == "minimax_h3_disaggregated"
@@ -25,9 +25,9 @@ pytest_plugins = (
    "tests.helpers.fixtures.clean",
    "tests.helpers.fixtures.log",
    "tests.helpers.fixtures.media",
-    "tests.helpers.fixtures.pytest_collection",
-    "tests.helpers.fixtures.pytest_run_args",
+    "tests.helpers.fixtures.run_args",
    "tests.helpers.fixtures.runtime",
+    "tests.helpers.fixtures.xdist",
 )


@@ -1,89 +0,0 @@
-"""Tests for ``OmniARScheduler._drop_aborted_queued_requests``.
-
-``schedule()`` sweeps ``FINISHED_ABORTED`` requests out of the queues before
-handing control to upstream, because upstream ``Scheduler.schedule()`` raises
-``RuntimeError: Invalid request status: FINISHED_ABORTED`` for any request it
-admits in a finished state -- which kills the stage's engine core, not just
-the request.
-
-Regression for the sweep missing ``skipped_waiting``, the third queue
-upstream admits from. An aborted duplex session parked there was re-selected
-by ``_select_waiting_queue_for_scheduling`` on a later tick and crashed the
-stage.
-"""
-
-from __future__ import annotations
-
-import pytest
-from vllm.v1.core.sched.request_queue import SchedulingPolicy, create_request_queue
-from vllm.v1.request import RequestStatus
-
-import vllm_omni.core.sched.omni_ar_scheduler as ar_sched_mod
-from vllm_omni.core.sched.omni_ar_scheduler import OmniARScheduler
-
-pytestmark = [pytest.mark.core_model, pytest.mark.cpu]
-
-_POLICIES = [
-    pytest.param(SchedulingPolicy.FCFS, id="fcfs"),
-    pytest.param(SchedulingPolicy.PRIORITY, id="priority"),
-]
-
-
-class _StubRequest:
-    """Minimal ``Request`` stub with the surface the sweep exercises.
-
-    Carries the ``(priority, arrival_time)`` ordering ``PriorityRequestQueue``
-    relies on so the same stub works under both scheduling policies.
-    """
-
-    def __init__(self, request_id: str, status: RequestStatus, arrival_time: float = 0.0) -> None:
-        self.request_id = request_id
-        self.status = status
-        self.priority = 0
-        self.arrival_time = arrival_time
-
-    def __lt__(self, other: _StubRequest) -> bool:
-        return (self.priority, self.arrival_time) < (other.priority, other.arrival_time)
-
-
-def _make_scheduler(policy: SchedulingPolicy, *, waiting=(), skipped_waiting=(), running=()):
-    """Bare scheduler with just the surface the sweep reads."""
-    scheduler = OmniARScheduler.__new__(OmniARScheduler)
-    scheduler.waiting = create_request_queue(policy)
-    for req in waiting:
-        scheduler.waiting.add_request(req)
-    scheduler.skipped_waiting = create_request_queue(policy)
-    for req in skipped_waiting:
-        scheduler.skipped_waiting.add_request(req)
-    scheduler.running = list(running)
-    return scheduler
-
-
-@pytest.mark.parametrize("policy", _POLICIES)
-def test_schedule_sweeps_skipped_waiting_before_upstream_selection(
-    monkeypatch: pytest.MonkeyPatch,
-    policy: SchedulingPolicy,
-) -> None:
-    """The public schedule path must remove aborted skipped requests first."""
-    aborted = _StubRequest("req-aborted", RequestStatus.FINISHED_ABORTED)
-    scheduler = _make_scheduler(policy, skipped_waiting=[aborted])
-    scheduler.chunk_transfer_adapter = None
-    scheduler.input_coordinator = None
-    scheduler.num_waiting_for_streaming_input = 0
-    scheduler._process_pending_omni_inputs = lambda model_mode: None
-    scheduler._resync_streaming_input_counter = lambda: None
-    scheduler._should_defer_waiting_admission = lambda: False
-    scheduler._restore_omni_wait_queues = lambda: None
-    scheduler._postprocess_omni_schedule_output = lambda *args, **kwargs: None
-    scheduler.get_finished_requests_needing_kv_transfer = lambda: {}
-    scheduler._wrap_omni_scheduler_output = lambda output, **kwargs: output
-
-    sentinel = object()
-
-    def upstream_schedule(_self, _throttle_prefills: bool = False):
-        assert list(scheduler.skipped_waiting) == []
-        return sentinel
-
-    monkeypatch.setattr(ar_sched_mod.VLLMScheduler, "schedule", upstream_schedule)
-
-    assert OmniARScheduler.schedule(scheduler) is sentinel
@@ -142,16 +142,13 @@ def test_exact_drain_delivers_new_segment_frame() -> None:
    session.num_computed_tokens = 6
    session.num_output_placeholders = 1
    session.num_in_flight_tokens = 1
-    session.async_tokens_to_discard = 1

    _replace_streaming_session(session)
    assert session.num_stale_output_tokens == 1
-    assert session.async_tokens_to_discard == 1

    sched = _make_drain_sched(session)
    assert _run_step(sched, session, num_scheduled=1, token=42) is False  # late frame dropped
    assert session.num_stale_output_tokens == 0
-    assert session.async_tokens_to_discard == 0
    assert _run_step(sched, session, num_scheduled=1, token=43) is True  # new segment survives


@@ -18,19 +18,17 @@ from vllm.sampling_params import SamplingParams
 from vllm.v1.core.sched.output import SchedulerOutput
 from vllm.v1.outputs import ModelRunnerOutput
 from vllm.v1.request import Request, RequestStatus, StreamingUpdate
-from vllm_omni.core.sched.omni_ar_scheduler import OmniARAsyncScheduler, OmniARScheduler
+from vllm_omni.core.sched.omni_ar_scheduler import OmniARScheduler

 # isort: on

 pytestmark = [pytest.mark.core_model, pytest.mark.cpu]


-def _make_scheduler(*, stage_id: int = 0, session_mode: str = "turn") -> OmniARScheduler:
+def _make_scheduler(*, stage_id: int = 0) -> OmniARScheduler:
    sched = OmniARScheduler.__new__(OmniARScheduler)
    sched._new_prompt_len_snapshot = {}
-    sched.vllm_config = SimpleNamespace(
-        model_config=SimpleNamespace(stage_id=stage_id, session_mode=session_mode),
-    )
+    sched.vllm_config = SimpleNamespace(model_config=SimpleNamespace(stage_id=stage_id))
    sched.num_waiting_for_streaming_input = 0
    sched.log_stats = False
    sched.chunk_transfer_adapter = None
@@ -187,7 +185,6 @@ def test_resumable_segment_boundary_keeps_pre_transition_send_watermark() -> Non

    def replace_with_next_segment(request: Request) -> bool:
        request.num_computed_tokens = 0
-        request._omni_segment_generation = 1
        request.status = RequestStatus.WAITING
        return False

@@ -204,7 +201,6 @@ def test_resumable_segment_boundary_keeps_pre_transition_send_watermark() -> Non
        True,
        new_token_ids=[42],
        confirmed_num_computed_tokens=26,
-        segment_generation=0,
    )


@@ -290,140 +286,6 @@ def test_queued_streaming_update_on_async_stop_fences_in_flight_once() -> None:
    assert session.num_stale_output_tokens == session.num_in_flight_tokens == 1


-def test_stale_async_frame_is_dropped_before_output_processing() -> None:
-    session = _make_request()
-    session.status = RequestStatus.RUNNING
-    session.num_in_flight_tokens = 2
-    session.num_computed_tokens = session.num_tokens + 1
-    session.num_output_placeholders = 1
-    session.async_tokens_to_discard = 1
-    session.sampling_params = SimpleNamespace(num_logprobs=1)
-    num_computed_tokens = session.num_computed_tokens
-    num_output_placeholders = session.num_output_placeholders
-
-    sched = MagicMock()
-    sched.requests = {session.request_id: session}
-    sched.perf_metrics = None
-    sched.structured_output_manager.should_advance.return_value = False
-
-    def discard_stale_output(request: Request, token_ids: list[int]) -> tuple[list[int], bool]:
-        request.async_tokens_to_discard = 0
-        return token_ids, False
-
-    sched._update_request_with_output.side_effect = discard_stale_output
-    sched._process_kv_transfer_trigger.return_value = False
-    sched.chunk_transfer_adapter = MagicMock()
-    sched.running = [session]
-    sched.waiting_for_transfer_free = set()
-    sched.transfer_triggered_requests = set()
-    sched.active_kv_transfers = set()
-    sched.pending_stop_after_extraction = set()
-    sched.connector = None
-    sched.kv_cache_manager.take_events.return_value = None
-    sched.kv_cache_manager.estimate_cached_tokens.return_value = 0
-    sched.finished_req_ids_dict = {}
-    sched.make_stats.return_value = None
-
-    scheduler_output = MagicMock(spec=SchedulerOutput)
-    scheduler_output.num_scheduled_tokens = {session.request_id: 1}
-    scheduler_output.scheduled_spec_decode_tokens = {}
-    scheduler_output.num_invalid_spec_tokens = 0
-
-    model_runner_output = MagicMock(spec=ModelRunnerOutput)
-    model_runner_output.sampled_token_ids = [[42]]
-    model_runner_output.logprobs = None
-    model_runner_output.prompt_logprobs_dict = {}
-    model_runner_output.pooler_output = None
-    model_runner_output.multimodal_outputs = None
-    model_runner_output.inter_stage_outputs = [{"hidden": object()}]
-    model_runner_output.num_nans_in_logits = None
-    model_runner_output.kv_connector_output = None
-    model_runner_output.cudagraph_stats = None
-    model_runner_output.req_id_to_index = {session.request_id: 0}
-    model_runner_output.routed_experts = None
-
-    OmniARScheduler.update_from_output(sched, scheduler_output, model_runner_output)
-
-    assert session.async_tokens_to_discard == 0
-    assert session.status == RequestStatus.RUNNING
-    assert session.num_computed_tokens == num_computed_tokens
-    assert session.num_output_placeholders == num_output_placeholders
-    sched.chunk_transfer_adapter.save_async.assert_not_called()
-
-    session.sampling_params = SamplingParams(max_tokens=8)
-    next_payload = {"hidden": object()}
-    sched._update_request_with_output.side_effect = None
-    sched._update_request_with_output.return_value = ([43], False)
-    model_runner_output.sampled_token_ids = [[43]]
-    model_runner_output.inter_stage_outputs = [next_payload]
-
-    OmniARScheduler.update_from_output(sched, scheduler_output, model_runner_output)
-
-    sched.chunk_transfer_adapter.save_async.assert_called_once_with(
-        next_payload,
-        session,
-        False,
-        new_token_ids=[43],
-        confirmed_num_computed_tokens=None,
-    )
-
-
-def test_legacy_stale_async_marker_bypasses_real_placeholder_accounting() -> None:
-    request = _make_request()
-    request.status = RequestStatus.RUNNING
-    request.num_in_flight_tokens = 1
-    request.num_output_placeholders = 0
-    request.async_tokens_to_discard = 1
-    request.num_stale_output_tokens = 0
-
-    sched = OmniARAsyncScheduler.__new__(OmniARAsyncScheduler)
-    sched.requests = {request.request_id: request}
-    sched.perf_metrics = None
-    sched.chunk_transfer_adapter = None
-    sched.connector = None
-    sched.ec_connector = None
-    sched.waiting_for_transfer_free = set()
-    sched.transfer_triggered_requests = set()
-    sched.active_kv_transfers = set()
-    sched.pending_stop_after_extraction = set()
-    sched.finished_req_ids_dict = {}
-    sched._new_prompt_len_snapshot = {}
-    sched.kv_cache_manager = MagicMock()
-    sched.kv_cache_manager.take_events.return_value = None
-    sched._remove_stopped_requests_from_queues = MagicMock()
-    sched._handle_failed_kv_load_outputs = MagicMock(return_value=[])
-    sched._cleanup_kv_tracking = MagicMock()
-    sched._aggregate_kv_connector_stats = MagicMock(return_value=None)
-    sched._publish_kv_cache_events = MagicMock()
-    sched._attach_finished_request_sets = MagicMock()
-    sched._attach_scheduler_stats = MagicMock()
-    sched._capture_omni_connector_output = MagicMock()
-
-    scheduler_output = MagicMock(spec=SchedulerOutput)
-    scheduler_output.num_scheduled_tokens = {request.request_id: 1}
-    scheduler_output.scheduled_spec_decode_tokens = {}
-    scheduler_output.num_invalid_spec_tokens = 0
-
-    model_runner_output = MagicMock(spec=ModelRunnerOutput)
-    model_runner_output.sampled_token_ids = [[42]]
-    model_runner_output.logprobs = None
-    model_runner_output.prompt_logprobs_dict = {}
-    model_runner_output.pooler_output = None
-    model_runner_output.multimodal_outputs = None
-    model_runner_output.inter_stage_outputs = None
-    model_runner_output.num_nans_in_logits = None
-    model_runner_output.kv_connector_output = None
-    model_runner_output.cudagraph_stats = None
-    model_runner_output.req_id_to_index = {request.request_id: 0}
-    model_runner_output.routed_experts = None
-
-    OmniARScheduler.update_from_output(sched, scheduler_output, model_runner_output)
-
-    assert request.async_tokens_to_discard == 0
-    assert request.num_output_placeholders == 0
-    assert request._output_token_ids == []
-
-
 def test_stage0_streaming_update_discards_outstanding_async_placeholder_token() -> None:
    sched = _make_scheduler(stage_id=0)
    session = _make_request()
@@ -584,15 +446,11 @@ def test_chunk_segment_cleanup_keeps_requeued_resumable_receiver() -> None:
        return result

    sched = OmniARScheduler.__new__(OmniARScheduler)
-    sched.vllm_config = SimpleNamespace(model_config=SimpleNamespace(session_mode="duplex"))
    sched.running = []
    sched.waiting = queue()
    sched.skipped_waiting = queue(session)
    sched.num_waiting_for_streaming_input = 1
-    sched.chunk_transfer_adapter = SimpleNamespace(
-        receives_chunks=True,
-        segment_finished_requests={session.request_id},
-    )
+    sched.chunk_transfer_adapter = SimpleNamespace(segment_finished_requests={session.request_id})

    sched._resume_downstream_chunk_receiver(session)
    sched._remove_stopped_requests_from_queues(set(), {session})
@@ -602,37 +460,3 @@ def test_chunk_segment_cleanup_keeps_requeued_resumable_receiver() -> None:
    assert session not in sched.skipped_waiting.requests
    assert sched.num_waiting_for_streaming_input == 0
    assert session.request_id not in sched.chunk_transfer_adapter.segment_finished_requests
-
-
-@pytest.mark.parametrize(
-    ("receives_chunks", "session_mode"),
-    [
-        (False, "duplex"),
-        (True, "turn"),
-    ],
-)
-def test_chunk_segment_cleanup_keeps_explicit_update_stage_parked(
-    receives_chunks: bool,
-    session_mode: str,
-) -> None:
-    """Only duplex connector-driven receivers resume without an update."""
-    session = _make_request()
-    session.status = RequestStatus.WAITING_FOR_STREAMING_REQ
-
-    sched = OmniARScheduler.__new__(OmniARScheduler)
-    sched.vllm_config = SimpleNamespace(model_config=SimpleNamespace(session_mode=session_mode))
-    sched.num_waiting_for_streaming_input = 1
-    sched.chunk_transfer_adapter = SimpleNamespace(
-        receives_chunks=receives_chunks,
-        segment_finished_requests={session.request_id},
-    )
-    sched.skipped_waiting = MagicMock()
-    sched._enqueue_waiting_request = MagicMock()
-
-    sched._resume_downstream_chunk_receiver(session)
-
-    assert session.status == RequestStatus.WAITING_FOR_STREAMING_REQ
-    assert sched.num_waiting_for_streaming_input == 1
-    assert session.request_id not in sched.chunk_transfer_adapter.segment_finished_requests
-    sched.skipped_waiting.remove_requests.assert_not_called()
-    sched._enqueue_waiting_request.assert_not_called()
@@ -76,9 +76,6 @@ def _make_scheduler(
        else:
            hf_config.voxcpm2_runtime_config = runtime_config
    sched.vllm_config = SimpleNamespace(model_config=model_config)
-    # Upstream ``Scheduler.__init__`` always creates ``skipped_waiting``; the
-    # abort sweep at the top of ``schedule()`` reads it.
-    sched.skipped_waiting = _MockQueue()
    return sched


@@ -166,15 +166,3 @@ def test_fast_path_schedule_defer_disabled_leaves_fence() -> None:
    OmniGenerationScheduler.schedule(sched)

    assert sched.sched_step_seq == 0
-
-
-def test_generation_schedule_sweeps_and_resyncs_around_pending_inputs() -> None:
-    sched, request = _make_fast_path_sched(defer_block_free=False)
-    calls: list[str] = []
-    sched._drop_aborted_queued_requests.side_effect = lambda: calls.append("sweep")
-    sched._process_pending_omni_inputs.side_effect = lambda model_mode: calls.append(model_mode)
-    sched._resync_streaming_input_counter.side_effect = lambda: calls.append("resync")
-
-    OmniGenerationScheduler.schedule(sched)
-
-    assert calls == ["sweep", "generation", "sweep", "resync"]
@@ -1,5 +1,5 @@
 # SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
 """Unit coverage for the two stage-input timeout hooks.

 ``_process_pending_input_timeouts`` covers the full-payload transport;
@@ -253,38 +253,6 @@ def test_log_failed_chunk_sends_noop_without_adapter():
    scheduler._log_failed_chunk_sends()


-class _FakeReceiveFailureAdapter:
-    def __init__(self, failures):
-        self._failures = dict(failures)
-
-    def collect_failed_receive_request_ids(self):
-        failures, self._failures = self._failures, {}
-        return failures
-
-
-class _FakeReceiveFailureScheduler(OmniSchedulerMixin):
-    def __init__(self, requests, adapter):
-        self.requests = requests
-        self.chunk_transfer_adapter = adapter
-        self.finish_calls = []
-
-    def finish_requests(self, req_ids, status):
-        self.finish_calls.append((set(req_ids), status))
-
-
-def test_process_chunk_receive_failures_delegates_to_finish_requests():
-    request_id = "invalid-chunk"
-    adapter = _FakeReceiveFailureAdapter({request_id: "prompt exceeds limit"})
-    scheduler = _FakeReceiveFailureScheduler({request_id: SimpleNamespace(request_id=request_id)}, adapter)
-
-    scheduler._process_chunk_receive_failures()
-
-    assert len(scheduler.finish_calls) == 1
-    finished_ids, status = scheduler.finish_calls[0]
-    assert finished_ids == {request_id}
-    assert getattr(status, "name", str(status)).endswith("FINISHED_ERROR")
-
-
 # --- R1.4: VLLM_OMNI_INPUT_WAIT_TIMEOUT_S is parsed at import time, so each of
 # these reloads the module under a different environment. One knob now gates the
 # deadline on both transports, which is why a silent 0 matters more than it did
@@ -1,168 +0,0 @@
-"""Tests for the ``num_waiting_for_streaming_input`` resync placed in
-``OmniARScheduler.finish_requests`` / ``OmniGenerationScheduler.finish_requests``
-and in ``OmniARScheduler.schedule``.
-
-Upstream keeps that counter incrementally (``+1`` when a resumable request
-parks in a waiting queue as ``WAITING_FOR_STREAMING_REQ``, ``-1`` when the
-next update arrives or the request is finished) and
-``Scheduler.get_num_unfinished_requests`` *subtracts* it from the waiting
-queues. Omni rewrites ``request.status`` outside both hooks -- the
-chunk-transfer adapter's park/restore and
-``_realign_request_status_to_queues`` -- so a counted request can leave
-``WAITING_FOR_STREAMING_REQ`` without the matching decrement and inflate the
-counter for the lifetime of the process.
-
-An inflated counter does not degrade throughput, it hangs the stage:
-``EngineCore.has_work()`` reads false while a live request sits in
-``waiting``, so the engine blocks in ``input_queue.get()`` and never calls
-``schedule()`` again. Regression for the Qwen3-Omni duplex symptom where the
-server answered the first conversation after a boot and was silent from the
-second on while still reporting healthy.
-"""
-
-from __future__ import annotations
-
-import pytest
-from vllm.v1.core.sched.request_queue import SchedulingPolicy, create_request_queue
-from vllm.v1.request import RequestStatus
-
-import vllm_omni.core.sched.omni_ar_scheduler as ar_sched_mod
-import vllm_omni.core.sched.omni_generation_scheduler as gen_sched_mod
-from vllm_omni.core.sched.omni_ar_scheduler import OmniARScheduler
-from vllm_omni.core.sched.omni_generation_scheduler import OmniGenerationScheduler
-
-pytestmark = [pytest.mark.core_model, pytest.mark.cpu]
-
-
-class _StubRequest:
-    def __init__(self, request_id: str, status: RequestStatus) -> None:
-        self.request_id = request_id
-        self.status = status
-
-    def is_finished(self) -> bool:
-        return RequestStatus.is_finished(self.status)
-
-
-_SCHEDULER_PARAMS = [
-    pytest.param(OmniARScheduler, ar_sched_mod, id="ar"),
-    pytest.param(OmniGenerationScheduler, gen_sched_mod, id="generation"),
-]
-
-
-def _make_scheduler(scheduler_cls, *, requests, running, waiting, counter, skipped=None):
-    scheduler = scheduler_cls.__new__(scheduler_cls)
-    scheduler.chunk_transfer_adapter = None
-    scheduler.input_coordinator = None
-    scheduler.requests = requests
-    scheduler.running = running
-    scheduler.waiting = waiting
-    scheduler.skipped_waiting = skipped if skipped is not None else []
-    scheduler.num_waiting_for_streaming_input = counter
-    return scheduler
-
-
-@pytest.mark.parametrize(("scheduler_cls", "scheduler_mod"), _SCHEDULER_PARAMS)
-def test_finish_requests_clears_counter_left_by_a_stomped_status(
-    monkeypatch: pytest.MonkeyPatch,
-    scheduler_cls,
-    scheduler_mod,
-) -> None:
-    """The abort case that wedged the talker.
-
-    The request was counted while parked for streaming input, then had its
-    status rewritten to ``RUNNING`` by omni's queue surgery, so upstream's
-    ``finish_requests`` took the running branch and never decremented. With
-    no parked requests left, the counter must land at 0.
-    """
-    aborted = _StubRequest("req-aborted", RequestStatus.RUNNING)
-    scheduler = _make_scheduler(
-        scheduler_cls,
-        requests={},  # upstream already dropped it
-        running=[],
-        waiting=[],
-        counter=1,
-    )
-
-    monkeypatch.setattr(
-        scheduler_mod.VLLMScheduler,
-        "finish_requests",
-        lambda self, request_ids, finished_status: [aborted],
-    )
-
-    scheduler_cls.finish_requests(scheduler, [aborted.request_id], RequestStatus.FINISHED_ABORTED)
-
-    assert scheduler.num_waiting_for_streaming_input == 0
-
-
-def test_resync_counts_parked_requests_in_skipped_waiting() -> None:
-    """``get_num_unfinished_requests`` sums ``waiting`` and ``skipped_waiting``,
-    so the resync has to look at both or it under-counts and reintroduces the
-    spin the counter exists to prevent."""
-    parked = _StubRequest("req-skipped", RequestStatus.WAITING_FOR_STREAMING_REQ)
-    scheduler = _make_scheduler(
-        OmniARScheduler,
-        requests={parked.request_id: parked},
-        running=[],
-        waiting=[],
-        counter=0,
-        skipped=[parked],
-    )
-
-    scheduler._resync_streaming_input_counter()
-
-    assert scheduler.num_waiting_for_streaming_input == 1
-
-
-def test_ar_schedule_resyncs_before_delegating_upstream(monkeypatch: pytest.MonkeyPatch) -> None:
-    """Pins placement in ``schedule``: the chunk-transfer adapter rewrites
-    status during ``process_pending_chunks``, so the resync has to run after
-    that and before upstream schedules, while ``self.waiting`` is still the
-    real queue."""
-    parked = _StubRequest("req-resumed", RequestStatus.RUNNING)
-    scheduler = OmniARScheduler.__new__(OmniARScheduler)
-    scheduler.chunk_transfer_adapter = None
-    scheduler.input_coordinator = None
-    scheduler.requests = {parked.request_id: parked}
-    scheduler.running = [parked]
-    scheduler.waiting = []
-    scheduler.skipped_waiting = []
-    scheduler.num_waiting_for_streaming_input = 1  # leaked by the status rewrite
-    scheduler.policy = "fcfs"
-
-    seen: dict[str, int] = {}
-
-    monkeypatch.setattr(
-        ar_sched_mod.OmniSchedulerMixin, "_consume_pending_connector_output", lambda self, model_mode: None
-    )
-    monkeypatch.setattr(ar_sched_mod.OmniSchedulerMixin, "_process_pending_input_timeouts", lambda self: None)
-    monkeypatch.setattr(OmniARScheduler, "_should_defer_waiting_admission", lambda self: False)
-
-    def fake_schedule(self, throttle_prefills=False):
-        seen["counter"] = self.num_waiting_for_streaming_input
-        raise _StopScheduleError
-
-    class _StopScheduleError(Exception):
-        pass
-
-    monkeypatch.setattr(ar_sched_mod.VLLMScheduler, "schedule", fake_schedule)
-
-    with pytest.raises(_StopScheduleError):
-        OmniARScheduler.schedule(scheduler)
-
-    assert seen["counter"] == 0
-
-
-def test_generation_scheduler_sweep_removes_aborted_requests_from_all_queues() -> None:
-    aborted = _StubRequest("req-generation-aborted", RequestStatus.FINISHED_ABORTED)
-    scheduler = OmniGenerationScheduler.__new__(OmniGenerationScheduler)
-    scheduler.waiting = create_request_queue(SchedulingPolicy.FCFS)
-    scheduler.waiting.add_request(aborted)
-    scheduler.skipped_waiting = create_request_queue(SchedulingPolicy.FCFS)
-    scheduler.skipped_waiting.add_request(aborted)
-    scheduler.running = [aborted]
-
-    scheduler._drop_aborted_queued_requests()
-
-    assert list(scheduler.waiting) == []
-    assert list(scheduler.skipped_waiting) == []
-    assert scheduler.running == []
@@ -1,8 +1,4 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
-
 import json
-import math
 import os
 import threading
 from collections.abc import Callable
@@ -163,18 +159,6 @@ def benchmark_params(request):

 def assert_result(result, params, num_prompt) -> None:
    assert result["completed"] == num_prompt, "Request failures exist"
-    baseline = params.get("baseline")
-    hardware = result.get("Hardware")
-    hardware_baseline = baseline.get(hardware) if isinstance(baseline, dict) and isinstance(hardware, str) else None
-    if isinstance(hardware_baseline, dict) and "mean_tpot_ms" in hardware_baseline:
-        num_tpot_samples = result.get("num_tpot_samples")
-        mean_tpot_ms = result.get("mean_tpot_ms")
-        assert isinstance(num_tpot_samples, int) and not isinstance(num_tpot_samples, bool) and num_tpot_samples > 0, (
-            "TPOT baseline is configured, but no measurable TPOT samples were produced"
-        )
-        assert (
-            isinstance(mean_tpot_ms, int | float) and not isinstance(mean_tpot_ms, bool) and math.isfinite(mean_tpot_ms)
-        ), "TPOT baseline is configured, but mean_tpot_ms is not finite"
    expected_audio_turns = params.get("expected_duplex_audio_turns_per_session")
    if expected_audio_turns is not None:
        session_metrics = result.get("duplex_session_metrics")
@@ -1,106 +0,0 @@
-[
-  {
-    "test_name": "test_hunyuanvideo15_i2v_single_device",
-    "mark": [
-      {
-        "hardware_marks": {
-          "res": {
-            "cuda": "H100"
-          },
-          "num_cards": 1
-        }
-      },
-      "full_model",
-      "diffusion"
-    ],
-    "description": "Single-device baseline",
-    "server_type": "vllm-omni",
-    "server_params": {
-      "model": "hunyuanvideo-community/HunyuanVideo-1.5-Diffusers-480p_i2v",
-      "serve_args": {
-        "flow-shift": 5.0,
-        "enable-diffusion-pipeline-profiler": true
-      }
-    },
-    "benchmark_params": [
-      {
-        "name": "832x480_frames33_steps4",
-        "dataset": "random",
-        "task": "i2v",
-        "num-prompts": 10,
-        "max-concurrency": 1,
-        "num-input-images": 1,
-        "seed": 42,
-        "enable-negative-prompt": true,
-        "random-request-config": [
-          {
-            "width": 832,
-            "height": 480,
-            "num_inference_steps": 4,
-            "num_frames": 33,
-            "fps": 24,
-            "weight": 1
-          }
-        ],
-        "extra-body": {
-          "guidance_scale": 6.0,
-          "flow_shift": 5.0
-        }
-      }
-    ]
-  },
-  {
-    "test_name": "test_hunyuanvideo15_i2v_cachedit_tp2_vae_patch2",
-    "mark": [
-      {
-        "hardware_marks": {
-          "res": {
-            "cuda": "H100"
-          },
-          "num_cards": 2
-        }
-      },
-      "full_model",
-      "diffusion"
-    ],
-    "description": "CacheDiT + TP=2 + VAE patch parallel=2 + VAE tiling",
-    "server_type": "vllm-omni",
-    "server_params": {
-      "model": "hunyuanvideo-community/HunyuanVideo-1.5-Diffusers-480p_i2v",
-      "serve_args": {
-        "flow-shift": 5.0,
-        "cache-backend": "cache_dit",
-        "tensor-parallel-size": 2,
-        "vae-patch-parallel-size": 2,
-        "vae-use-tiling": true,
-        "enable-diffusion-pipeline-profiler": true
-      }
-    },
-    "benchmark_params": [
-      {
-        "name": "832x480_frames33_steps4",
-        "dataset": "random",
-        "task": "i2v",
-        "num-prompts": 10,
-        "max-concurrency": 1,
-        "num-input-images": 1,
-        "seed": 42,
-        "enable-negative-prompt": true,
-        "random-request-config": [
-          {
-            "width": 832,
-            "height": 480,
-            "num_inference_steps": 4,
-            "num_frames": 33,
-            "fps": 24,
-            "weight": 1
-          }
-        ],
-        "extra-body": {
-          "guidance_scale": 6.0,
-          "flow_shift": 5.0
-        }
-      }
-    ]
-  }
-]
@@ -1,104 +0,0 @@
-[
-  {
-    "test_name": "test_hunyuanvideo15_t2v_single_device",
-    "mark": [
-      {
-        "hardware_marks": {
-          "res": {
-            "cuda": "H100"
-          },
-          "num_cards": 1
-        }
-      },
-      "full_model",
-      "diffusion"
-    ],
-    "description": "Single-device baseline",
-    "server_type": "vllm-omni",
-    "server_params": {
-      "model": "hunyuanvideo-community/HunyuanVideo-1.5-Diffusers-480p_t2v",
-      "serve_args": {
-        "flow-shift": 5.0,
-        "enable-diffusion-pipeline-profiler": true
-      }
-    },
-    "benchmark_params": [
-      {
-        "name": "832x480_frames33_steps4",
-        "dataset": "random",
-        "task": "t2v",
-        "num-prompts": 10,
-        "max-concurrency": 1,
-        "seed": 42,
-        "enable-negative-prompt": true,
-        "random-request-config": [
-          {
-            "width": 832,
-            "height": 480,
-            "num_inference_steps": 4,
-            "num_frames": 33,
-            "fps": 24,
-            "weight": 1
-          }
-        ],
-        "extra-body": {
-          "guidance_scale": 6.0,
-          "flow_shift": 5.0
-        }
-      }
-    ]
-  },
-  {
-    "test_name": "test_hunyuanvideo15_t2v_cachedit_tp2_vae_patch2",
-    "mark": [
-      {
-        "hardware_marks": {
-          "res": {
-            "cuda": "H100"
-          },
-          "num_cards": 2
-        }
-      },
-      "full_model",
-      "diffusion"
-    ],
-    "description": "CacheDiT + TP=2 + VAE patch parallel=2 + VAE tiling",
-    "server_type": "vllm-omni",
-    "server_params": {
-      "model": "hunyuanvideo-community/HunyuanVideo-1.5-Diffusers-480p_t2v",
-      "serve_args": {
-        "flow-shift": 5.0,
-        "cache-backend": "cache_dit",
-        "tensor-parallel-size": 2,
-        "vae-patch-parallel-size": 2,
-        "vae-use-tiling": true,
-        "enable-diffusion-pipeline-profiler": true
-      }
-    },
-    "benchmark_params": [
-      {
-        "name": "832x480_frames33_steps4",
-        "dataset": "random",
-        "task": "t2v",
-        "num-prompts": 10,
-        "max-concurrency": 1,
-        "seed": 42,
-        "enable-negative-prompt": true,
-        "random-request-config": [
-          {
-            "width": 832,
-            "height": 480,
-            "num_inference_steps": 4,
-            "num_frames": 33,
-            "fps": 24,
-            "weight": 1
-          }
-        ],
-        "extra-body": {
-          "guidance_scale": 6.0,
-          "flow_shift": 5.0
-        }
-      }
-    ]
-  }
-]
@@ -43,7 +43,7 @@
        "extra_body": {
          "save_duplex_request_metrics": true
        },
-        "percentile-metrics": "ttft,tpot,itl,e2el,audio_ttfp,audio_rtf,audio_duration",
+        "percentile-metrics": "ttft,e2el,audio_ttfp,audio_rtf,audio_duration",
        "baseline": {
          "H100": {
            "request_throughput": [