[CI][MiniCPM-o] Add NPU Video-MME accuracy nightly as a standalone job (#6596)

Signed-off-by: amy-why-3459 <wuhaiyan17@huawei.com>

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

@@ -6,6 +6,10 @@
 
 set -euo pipefail
 
+# minicpm-challenge branch: skip entire AMD CI (no pipeline upload => no image build / tests).
+echo "--- Skipping AMD CI on this branch (including image build)"
+exit 0
+
 source .buildkite/common/scripts/resolve_skip_ci.sh
 
 if [[ -z "${RUN_ALL:-}" ]]; then

.buildkite/common/scripts/skip_ci.py

@@ -282,7 +282,7 @@ def _git(*args: str) -> subprocess.CompletedProcess[str]:
 
 
 def _resolve_diff_range() -> str | None:
-    """Return a git diff range for PR or main builds, or None when unavailable."""
+    """Return a git diff range for PR or push-branch builds, or None when unavailable."""
     is_pr = os.environ.get("BUILDKITE_PULL_REQUEST", "false") != "false" and os.environ.get(
         "BUILDKITE_PULL_REQUEST", ""
     )
@@ -302,13 +302,15 @@ def _resolve_diff_range() -> str | None:
                 return None
         return f"{base_ref}...{commit}"
 
-    if os.environ.get("BUILDKITE_BRANCH", "") == "main":
+    # main: ready/nightly/weekly push CI; minicpm-challenge: merge (L3) push CI
+    branch = os.environ.get("BUILDKITE_BRANCH", "")
+    if branch in ("main", "minicpm-challenge"):
         if _git("rev-parse", "--verify", f"{commit}^").returncode != 0:
-            _log("main commit has no parent; using safe defaults")
+            _log(f"{branch} commit has no parent; using safe defaults")
             return None
         return f"{commit}^..{commit}"
 
-    _log("not PR/main build; using safe defaults")
+    _log("not PR/main/minicpm-challenge build; using safe defaults")
     return None
 
 

.buildkite/common/scripts/upload_pipeline.py

@@ -59,11 +59,12 @@ E2E_GROUP_MARKER = "E2E Test"
 CI_MIRROR_HARDWARES_PATH = ROOT / ".buildkite/common/ci_mirror_hardwares.yml"
 
 CUDA_NIGHTLY_ONLY = (
-    '(build.pull_request.labels includes "nightly-test") || (build.branch == "main" && build.env("NIGHTLY") == "1")'
+    '(build.pull_request.labels includes "nightly-test") || '
+    '(build.branch == "minicpm-challenge" && build.env("NIGHTLY") == "1")'
 )
 NPU_NIGHTLY_ONLY = (
-    '(build.branch == "main" && build.env("NIGHTLY") == "1") || '
-    '(build.branch != "main" && ('
+    '(build.branch == "minicpm-challenge" && build.env("NIGHTLY") == "1") || '
+    '(build.branch != "minicpm-challenge" && ('
     'build.pull_request.labels includes "nightly-test" || '
     'build.pull_request.labels includes "omni-test" || '
     'build.pull_request.labels includes "tts-test" || '
@@ -103,7 +104,8 @@ def _format_bootstrap_if(expr: str) -> str:
 
 def _compute_bootstrap_if_exprs(*, decision, platform: str) -> dict[str, str]:
     disabled = "false"
-    nightly_main = 'build.branch == "main" && build.env("NIGHTLY") == "1"'
+    # L4 nightly scheduled push CI runs on minicpm-challenge (not main).
+    nightly_main = 'build.branch == "minicpm-challenge" && build.env("NIGHTLY") == "1"'
     nightly_only = NPU_NIGHTLY_ONLY if platform == "npu" else CUDA_NIGHTLY_ONLY
 
     if platform == "npu":
@@ -118,11 +120,12 @@ def _compute_bootstrap_if_exprs(*, decision, platform: str) -> dict[str, str]:
         merge_base = disabled
     else:
         ready_pr = 'build.branch != "main" && build.pull_request.labels includes "ready"'
-        merge_main = 'build.branch == "main" && build.env("NIGHTLY") != "1" && build.env("WEEKLY") != "1"'
-        merge_pr = 'build.branch != "main" && build.pull_request.labels includes "merge-test"'
+        # L3 merge push CI runs on minicpm-challenge (not main).
+        merge_main = 'build.branch == "minicpm-challenge" && build.env("NIGHTLY") != "1" && build.env("WEEKLY") != "1"'
+        merge_pr = 'build.branch != "minicpm-challenge" && build.pull_request.labels includes "merge-test"'
         nightly_label_if = (
-            '(build.branch == "main" && build.env("NIGHTLY") == "1") || '
-            '(build.branch != "main" && ('
+            '(build.branch == "minicpm-challenge" && build.env("NIGHTLY") == "1") || '
+            '(build.branch != "minicpm-challenge" && ('
             'build.pull_request.labels includes "nightly-test" || '
             'build.pull_request.labels includes "omni-test" || '
             'build.pull_request.labels includes "tts-test" || '
@@ -140,8 +143,8 @@ def _compute_bootstrap_if_exprs(*, decision, platform: str) -> dict[str, str]:
     weekly_main = 'build.branch == "main" && build.env("WEEKLY") == "1"'
 
     if decision.skip_all:
-        # Docs / skip-mark only: no PR-label escape hatch. Main scheduled
-        # NIGHTLY=1 still runs L4 + L2/L3 (--e2e); WEEKLY=1 still runs L5.
+        # Docs / skip-mark only: no PR-label escape hatch. Scheduled
+        # NIGHTLY=1 (minicpm-challenge) still runs L4 + L2/L3 (--e2e); WEEKLY=1 (main) still runs L5.
         image_expr = f"({nightly_main}) || ({weekly_main})" if platform == "cuda" else nightly_main
         ready_expr = nightly_main
         merge_expr = nightly_main if platform == "cuda" else disabled

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

@@ -8,35 +8,35 @@ steps:
     agents:
       queue: "cpu_queue_premerge"
 
-  # L2 Test — PR ready label (diff-aware) or main+NIGHTLY=1 (--e2e, full E2E group)
+  # L2 Test — PR ready label (diff-aware) or minicpm-challenge+NIGHTLY=1 (--e2e, full E2E group)
   - label: "Upload Ready Pipeline"
     depends_on: image-build
     key: upload-ready-pipeline
     commands:
       - |
         EXTRA=""
-        if [ "${NIGHTLY:-}" = "1" ] && [ "${BUILDKITE_BRANCH}" = "main" ]; then
+        if [ "${NIGHTLY:-}" = "1" ] && [ "${BUILDKITE_BRANCH}" = "minicpm-challenge" ]; then
           EXTRA="--e2e"
         fi
         python3 .buildkite/common/scripts/upload_pipeline.py --upload $$EXTRA .buildkite/cuda/test-ready.yml
     agents:
       queue: "cpu_queue_premerge"
 
-  # L3 Test — PR merge-test label (diff-aware) or main+NIGHTLY=1 (--e2e, full E2E group)
+  # L3 Test — PR merge-test label (diff-aware) or minicpm-challenge+NIGHTLY=1 (--e2e, full E2E group)
   - label: "Upload Merge Pipeline"
     depends_on: image-build
     key: upload-merge-pipeline
     commands:
       - |
         EXTRA=""
-        if [ "${NIGHTLY:-}" = "1" ] && [ "${BUILDKITE_BRANCH}" = "main" ]; then
+        if [ "${NIGHTLY:-}" = "1" ] && [ "${BUILDKITE_BRANCH}" = "minicpm-challenge" ]; then
           EXTRA="--e2e"
         fi
         python3 .buildkite/common/scripts/upload_pipeline.py --upload $$EXTRA .buildkite/cuda/test-merge.yml
     agents:
       queue: "cpu_queue_premerge"
 
-  # L4 Test — main+NIGHTLY=1 (scheduled), or PR with specific label (e.g. add label then Rebuild)
+  # L4 Test — minicpm-challenge+NIGHTLY=1 (scheduled), or PR with specific label (e.g. add label then Rebuild)
   - label: "Upload Nightly Pipeline"
     depends_on: image-build
     key: upload-nightly-pipeline

.buildkite/cuda/test-merge.yml

@@ -18,261 +18,227 @@ steps:
           - "timeout 20m pytest -sv -m 'core_model and cpu' --ignore=tests/diffusion --ignore=tests/model_executor --cov=vllm_omni --cov-report=term-missing:skip-covered --cov-report=xml"
         mirror_hardwares: l4_1
 
-  - group: ":card_index_dividers: Diffusion Test"
-    depends_on: upload-merge-pipeline
-    steps:
-      - label: "Diffusion · LoRA Test"
-        timeout_in_minutes: 30
-        commands:
-          - pytest -sv tests/diffusion/lora/ -m 'advanced_model and cuda' --run-level "advanced_model"
-        mirror_hardwares: l4_1
+  # - group: ":card_index_dividers: Diffusion Test"
+  #   depends_on: upload-merge-pipeline
+  #   steps:
+  #     - label: "Diffusion · LoRA Test"
+  #       timeout_in_minutes: 30
+  #       commands:
+  #         - pytest -sv tests/diffusion/lora/ -m 'advanced_model and cuda' --run-level "advanced_model"
+  #       mirror_hardwares: l4_1
 
-      - label: "Diffusion · Offloader Test"
-        timeout_in_minutes: 40
-        commands:
-          - timeout 40m pytest -s -v tests/diffusion/offloader -m "advanced_model and cuda"
-        mirror_hardwares: l4_1
+  #     - label: "Diffusion · Offloader Test"
+  #       timeout_in_minutes: 40
+  #       commands:
+  #         - timeout 40m pytest -s -v tests/diffusion/offloader -m "advanced_model and cuda"
+  #       mirror_hardwares: l4_1
 
-  - label: "Distributed Test"
-    depends_on: upload-merge-pipeline
-    timeout_in_minutes: 60
-    commands:
-      - |
-        timeout 55m bash -c "
-          set -e
-          export VLLM_IMAGE_FETCH_TIMEOUT=60
-          pytest -s -v tests/distributed -m 'advanced_model and cuda and H100' --run-level 'advanced_model'
-        "
-    mirror_hardwares: h100_1
+  # - label: "Distributed Test"
+  #   depends_on: upload-merge-pipeline
+  #   timeout_in_minutes: 60
+  #   commands:
+  #     - |
+  #       timeout 55m bash -c "
+  #         set -e
+  #         export VLLM_IMAGE_FETCH_TIMEOUT=60
+  #         pytest -s -v tests/distributed -m 'advanced_model and cuda and H100' --run-level 'advanced_model'
+  #       "
+  #   mirror_hardwares: h100_1
 
-  - group: ":card_index_dividers: Entrypoints Test"
-    depends_on: upload-merge-pipeline
-    steps:
-      - label: "Entrypoint Test with H100"
-        # TEMPORARY 60 -> 90: engine startup/sleep-wake is ~13-15 min slower per
-        # run on the vLLM 0.26 image (model loading ~2x, init ~2x; see builds
-        # 2915 vs 2916), which pushed this suite past 60 min. Revert once the
-        # 0.26 startup regression is understood and fixed.
-        timeout_in_minutes: 90
-        commands:
-          - pytest -s -v tests/entrypoints/ -m "advanced_model and cuda and H100" --run-level "advanced_model"
-        mirror_hardwares: h100_2
+  # - group: ":card_index_dividers: Entrypoints Test"
+  #   depends_on: upload-merge-pipeline
+  #   steps:
+  #     - label: "Entrypoint Test with H100"
+  #       timeout_in_minutes: 60
+  #       commands:
+  #         - pytest -s -v tests/entrypoints/ -m "advanced_model and cuda and H100" --run-level "advanced_model"
+  #       mirror_hardwares: h100_2
 
-      - label: "Entrypoint Test with L4"
-        timeout_in_minutes: 60
-        commands:
-          - pytest -s -v tests/entrypoints/ -m "advanced_model and cuda and L4" --run-level "advanced_model"
-        mirror_hardwares: l4_1
+  #     - label: "Entrypoint Test with L4"
+  #       timeout_in_minutes: 30
+  #       commands:
+  #         - pytest -s -v tests/entrypoints/ -m "advanced_model and cuda and L4" --run-level "advanced_model"
+  #       mirror_hardwares: l4_1
 
-  - group: ":card_index_dividers: E2E Test"
-    depends_on: upload-merge-pipeline
-    steps:
-      - label: "TTS · Qwen3-TTS CustomVoice Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_qwen3_tts_customvoice.py
-          - tests/e2e/offline_inference/test_qwen3_tts_customvoice.py
-          - vllm_omni/model_executor/models/qwen3_tts/
-          - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
-          - vllm_omni/deploy/qwen3_tts.yaml
-        commands:
-          - |
-            timeout 40m bash -c "
-              export VLLM_LOGGING_LEVEL=DEBUG
-              export VLLM_ALLOW_LONG_MAX_MODEL_LEN=1
-              pytest -s -v tests/e2e/online_serving/test_qwen3_tts_customvoice.py tests/e2e/offline_inference/test_qwen3_tts_customvoice.py -m 'advanced_model and cuda' --run-level 'advanced_model'
-            "
-        mirror_hardwares: l4_1
+  # - group: ":card_index_dividers: E2E Test"
+  #   depends_on: upload-merge-pipeline
+  #   steps:
+  #     - label: "TTS · Qwen3-TTS CustomVoice Test"
+  #       source_file_dependencies:
+  #         - tests/e2e/online_serving/test_qwen3_tts_customvoice.py
+  #         - tests/e2e/offline_inference/test_qwen3_tts_customvoice.py
+  #         - vllm_omni/model_executor/models/qwen3_tts/
+  #         - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
+  #         - vllm_omni/deploy/qwen3_tts.yaml
+  #       commands:
+  #         - |
+  #           timeout 40m bash -c "
+  #             export VLLM_LOGGING_LEVEL=DEBUG
+  #             export VLLM_ALLOW_LONG_MAX_MODEL_LEN=1
+  #             pytest -s -v tests/e2e/online_serving/test_qwen3_tts_customvoice.py tests/e2e/offline_inference/test_qwen3_tts_customvoice.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+  #           "
+  #       mirror_hardwares: l4_1
 
-      - label: "TTS · Qwen3-TTS Base Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_qwen3_tts_base.py
-          - tests/e2e/offline_inference/test_qwen3_tts_base.py
-          - vllm_omni/model_executor/models/qwen3_tts/
-          - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
-          - vllm_omni/deploy/qwen3_tts.yaml
-        commands:
-          - |
-            timeout 40m bash -c "
-              export VLLM_LOGGING_LEVEL=DEBUG
-              export VLLM_ALLOW_LONG_MAX_MODEL_LEN=1
-              pytest -s -v tests/e2e/online_serving/test_qwen3_tts_base.py tests/e2e/offline_inference/test_qwen3_tts_base.py -m 'advanced_model and cuda' --run-level 'advanced_model'
-            "
-        mirror_hardwares: l4_1
+  #     - label: "TTS · Qwen3-TTS Base Test"
+  #       source_file_dependencies:
+  #         - tests/e2e/online_serving/test_qwen3_tts_base.py
+  #         - tests/e2e/offline_inference/test_qwen3_tts_base.py
+  #         - vllm_omni/model_executor/models/qwen3_tts/
+  #         - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
+  #         - vllm_omni/deploy/qwen3_tts.yaml
+  #       commands:
+  #         - |
+  #           timeout 40m bash -c "
+  #             export VLLM_LOGGING_LEVEL=DEBUG
+  #             export VLLM_ALLOW_LONG_MAX_MODEL_LEN=1
+  #             pytest -s -v tests/e2e/online_serving/test_qwen3_tts_base.py tests/e2e/offline_inference/test_qwen3_tts_base.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+  #           "
+  #       mirror_hardwares: l4_1
 
-      - label: "TTS · Higgs-Audio-V3 Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_higgs_audio_v3.py
-          - vllm_omni/model_executor/models/higgs_audio_v3/
-          - vllm_omni/model_executor/stage_input_processors/higgs_audio_v3.py
-          - vllm_omni/deploy/higgs_multimodal_qwen3.yaml
-        commands:
-          - |
-            timeout 20m bash -c "
-              export VLLM_WORKER_MULTIPROC_METHOD=spawn
-              export VLLM_USE_DEEP_GEMM=0
-              export VLLM_MOE_USE_DEEP_GEMM=0
-              pytest -s -v tests/e2e/online_serving/test_higgs_audio_v3.py -m 'advanced_model' --run-level 'advanced_model'
-            "
-        mirror_hardwares: l4_1
+  #     - label: "TTS · Higgs-Audio-V3 Test"
+  #       source_file_dependencies:
+  #         - tests/e2e/online_serving/test_higgs_audio_v3.py
+  #         - vllm_omni/model_executor/models/higgs_audio_v3/
+  #         - vllm_omni/model_executor/stage_input_processors/higgs_audio_v3.py
+  #         - vllm_omni/deploy/higgs_multimodal_qwen3.yaml
+  #       commands:
+  #         - |
+  #           timeout 20m bash -c "
+  #             export VLLM_WORKER_MULTIPROC_METHOD=spawn
+  #             export VLLM_USE_DEEP_GEMM=0
+  #             export VLLM_MOE_USE_DEEP_GEMM=0
+  #             pytest -s -v tests/e2e/online_serving/test_higgs_audio_v3.py -m 'advanced_model' --run-level 'advanced_model'
+  #           "
+  #       mirror_hardwares: l4_1
 
-      - label: "Omni · Qwen3-Omni Test"
-        source_file_dependencies:
-          - tests/e2e/offline_inference/test_qwen3_omni.py
-          - tests/e2e/online_serving/test_qwen3_omni.py
-          - vllm_omni/model_executor/models/qwen3_omni/
-          - vllm_omni/model_executor/stage_input_processors/qwen3_omni.py
-          - vllm_omni/deploy/qwen3_omni_moe.yaml
-        timeout_in_minutes: 50
-        commands:
-          - pytest -s -v tests/e2e/offline_inference/test_qwen3_omni.py tests/e2e/online_serving/test_qwen3_omni.py -m 'advanced_model and cuda' --run-level 'advanced_model'
-        mirror_hardwares: h100_2
+  #     - label: "Omni · Qwen3-Omni Test"
+  #       source_file_dependencies:
+  #         - tests/e2e/offline_inference/test_qwen3_omni.py
+  #         - tests/e2e/online_serving/test_qwen3_omni.py
+  #         - vllm_omni/model_executor/models/qwen3_omni/
+  #         - vllm_omni/model_executor/stage_input_processors/qwen3_omni.py
+  #         - vllm_omni/deploy/qwen3_omni_moe.yaml
+  #       timeout_in_minutes: 50
+  #       commands:
+  #         - pytest -s -v tests/e2e/offline_inference/test_qwen3_omni.py tests/e2e/online_serving/test_qwen3_omni.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+  #       mirror_hardwares: h100_2
 
       - label: "Omni · MiniCPM-o 4.5 Test"
-        source_file_dependencies:
-          - tests/e2e/offline_inference/test_minicpmo_4_5.py
-          - tests/e2e/online_serving/test_minicpmo_4_5.py
-          - vllm_omni/deploy/minicpmo_4_5.yaml
         timeout_in_minutes: 50
         commands:
           - pytest -s -v tests/e2e/offline_inference/test_minicpmo_4_5.py tests/e2e/online_serving/test_minicpmo_4_5.py -m 'advanced_model and cuda' --run-level 'advanced_model'
-        agents:
-          queue: "mithril-h100-pool"
-        plugins:
-          - kubernetes:
-              podSpec:
-                containers:
-                  - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                    resources:
-                      limits:
-                        nvidia.com/gpu: 2
-                    volumeMounts:
-                      - name: devshm
-                        mountPath: /dev/shm
-                      - name: hf-cache
-                        mountPath: /root/.cache/huggingface
-                    env:
-                      - name: HF_HOME
-                        value: /root/.cache/huggingface
-                      - name: HF_TOKEN
-                        valueFrom:
-                          secretKeyRef:
-                            name: hf-token-secret
-                            key: token
-                nodeSelector:
-                  node.kubernetes.io/instance-type: gpu-h100-sxm
-                volumes:
-                  - name: devshm
-                    emptyDir:
-                      medium: Memory
-                  - name: hf-cache
-                    hostPath:
-                      path: /mnt/hf-cache
-                      type: DirectoryOrCreate
-
-      - label: "TTS · VoxCPM2 Test"
-        source_file_dependencies:
-          - tests/e2e/offline_inference/test_voxcpm2_tts.py
-          - tests/e2e/online_serving/test_voxcpm2_tts.py
-          - vllm_omni/model_executor/models/voxcpm2/
-          - vllm_omni/deploy/voxcpm2.yaml
-        timeout_in_minutes: 30
-        commands:
-          - |
-            timeout 30m bash -c "
-              export VLLM_LOGGING_LEVEL=DEBUG
-              export VLLM_WORKER_MULTIPROC_METHOD=spawn
-              pytest -s -v tests/e2e/offline_inference/test_voxcpm2_tts.py tests/e2e/online_serving/test_voxcpm2_tts.py -m 'advanced_model and cuda' --run-level 'advanced_model'
-            "
-        mirror_hardwares: l4_1
-
-      - label: "TTS · SoulX-Singer Test"
-        source_file_dependencies:
-          - tests/e2e/offline_inference/test_soulxsinger.py
-          - vllm_omni/diffusion/models/soulx_singer/
-          - vllm_omni/deploy/soulxsinger_svc.yaml
-          - vllm_omni/deploy/soulxsinger_svs.yaml
-        timeout_in_minutes: 30
-        commands:
-          - |
-            timeout 30m bash -c "
-              export VLLM_LOGGING_LEVEL=DEBUG
-              export VLLM_WORKER_MULTIPROC_METHOD=spawn
-              pytest -s -v tests/e2e/offline_inference/test_soulxsinger.py -k 'svc or precomputed' -m 'advanced_model and cuda' --run-level 'advanced_model'
-            "
-        mirror_hardwares: l4_1
-
-      - label: "Diffusion · Qwen Image Edit Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_qwen_image_edit.py
-          - vllm_omni/diffusion/models/qwen_image/
-        timeout_in_minutes: 20
-        commands:
-          - pytest -s -v tests/e2e/online_serving/test_qwen_image_edit.py -m 'advanced_model and cuda' --run-level 'advanced_model'
-        mirror_hardwares: h100_1
-      - label: "Diffusion · Bagel Test"
-        source_file_dependencies:
-          - tests/e2e/offline_inference/test_bagel.py
-          - tests/e2e/online_serving/test_bagel.py
-          - vllm_omni/model_executor/models/bagel/
-          - vllm_omni/diffusion/models/bagel/
-          - vllm_omni/model_executor/stage_input_processors/bagel.py
-          - vllm_omni/deploy/bagel.yaml
-        timeout_in_minutes: 20
-        commands:
-          - pytest -s -v tests/e2e/offline_inference/test_bagel.py tests/e2e/online_serving/test_bagel.py -m 'advanced_model and cuda' --run-level 'advanced_model'
         mirror_hardwares: h100_1
 
-      - label: "Diffusion · Qwen Image Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_qwen_image.py
-          - vllm_omni/diffusion/models/qwen_image/
+      - label: "Omni · MiniCPM-o 4.5 Duplex Test"
+        timeout_in_minutes: 50
         commands:
-          - |
-            timeout 40m bash -c '
-              export VLLM_IMAGE_FETCH_TIMEOUT=60
-              pytest -s -v tests/e2e/online_serving/test_qwen_image.py -m 'advanced_model and cuda' --run-level 'advanced_model'
-            '
+          - pytest -s -v tests/e2e/online_serving/test_minicpmo_4_5_duplex.py -m 'advanced_model and cuda' --run-level 'advanced_model'
         mirror_hardwares: h100_1
 
-      - label: "Diffusion · Qwen Image Layered Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_qwen_image_layered.py
-          - vllm_omni/diffusion/models/qwen_image/
-        commands:
-          - |
-            timeout 40m bash -c "
-              export VLLM_IMAGE_FETCH_TIMEOUT=60
-              pytest -s -v tests/e2e/online_serving/test_qwen_image_layered.py -m 'advanced_model and cuda' --run-level 'advanced_model'
-            "
-        mirror_hardwares: h100_1
+      # - label: "TTS · VoxCPM2 Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/offline_inference/test_voxcpm2_tts.py
+      #     - tests/e2e/online_serving/test_voxcpm2_tts.py
+      #     - vllm_omni/model_executor/models/voxcpm2/
+      #     - vllm_omni/deploy/voxcpm2.yaml
+      #   timeout_in_minutes: 30
+      #   commands:
+      #     - |
+      #       timeout 30m bash -c "
+      #         export VLLM_LOGGING_LEVEL=DEBUG
+      #         export VLLM_WORKER_MULTIPROC_METHOD=spawn
+      #         pytest -s -v tests/e2e/offline_inference/test_voxcpm2_tts.py tests/e2e/online_serving/test_voxcpm2_tts.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+      #       "
+      #   mirror_hardwares: l4_1
 
-      - label: "Diffusion · Wan22 Test"
-        source_file_dependencies:
-          - tests/e2e/offline_inference/test_wan22_t2v.py
-          - tests/e2e/online_serving/test_wan22_t2v.py
-          - vllm_omni/diffusion/models/wan2_2/
-        commands:
-          - |
-            timeout 40m bash -c "
-              export VLLM_IMAGE_FETCH_TIMEOUT=60
-              pytest -s -v tests/e2e/offline_inference/test_wan22_t2v.py tests/e2e/online_serving/test_wan22_t2v.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
-            "
-        mirror_hardwares: h100_1
+      # - label: "TTS · SoulX-Singer Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/offline_inference/test_soulxsinger.py
+      #     - vllm_omni/diffusion/models/soulx_singer/
+      #     - vllm_omni/deploy/soulxsinger_svc.yaml
+      #     - vllm_omni/deploy/soulxsinger_svs.yaml
+      #   timeout_in_minutes: 30
+      #   commands:
+      #     - |
+      #       timeout 30m bash -c "
+      #         export VLLM_LOGGING_LEVEL=DEBUG
+      #         export VLLM_WORKER_MULTIPROC_METHOD=spawn
+      #         pytest -s -v tests/e2e/offline_inference/test_soulxsinger.py -k 'svc or precomputed' -m 'advanced_model and cuda' --run-level 'advanced_model'
+      #       "
+      #   mirror_hardwares: l4_1
 
-      - label: "Diffusion · Z Image Test"
-        source_file_dependencies:
-          - tests/e2e/offline_inference/test_z_image.py
-          - tests/e2e/online_serving/test_z_image.py
-          - vllm_omni/diffusion/models/z_image/
-        timeout_in_minutes: 30
-        commands:
-          - pytest -s -v tests/e2e/offline_inference/test_z_image.py tests/e2e/online_serving/test_z_image.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
-        mirror_hardwares: l4_1
-      - label: "Diffusion · FLUX.2 Klein Test"
-        source_file_dependencies:
-          - tests/e2e/offline_inference/test_flux2_klein.py
-          - vllm_omni/diffusion/models/flux2_klein/
-        timeout_in_minutes: 30
-        commands:
-          - pytest -s -v tests/e2e/offline_inference/test_flux2_klein.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
-        mirror_hardwares: l4_1
+      # - label: "Diffusion · Qwen Image Edit Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/online_serving/test_qwen_image_edit.py
+      #     - vllm_omni/diffusion/models/qwen_image/
+      #   timeout_in_minutes: 20
+      #   commands:
+      #     - pytest -s -v tests/e2e/online_serving/test_qwen_image_edit.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+      #   mirror_hardwares: h100_1
+      # - label: "Diffusion · Bagel Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/offline_inference/test_bagel.py
+      #     - tests/e2e/online_serving/test_bagel.py
+      #     - vllm_omni/model_executor/models/bagel/
+      #     - vllm_omni/diffusion/models/bagel/
+      #     - vllm_omni/model_executor/stage_input_processors/bagel.py
+      #     - vllm_omni/deploy/bagel.yaml
+      #   timeout_in_minutes: 20
+      #   commands:
+      #     - pytest -s -v tests/e2e/offline_inference/test_bagel.py tests/e2e/online_serving/test_bagel.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+      #   mirror_hardwares: h100_1
+
+      # - label: "Diffusion · Qwen Image Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/online_serving/test_qwen_image.py
+      #     - vllm_omni/diffusion/models/qwen_image/
+      #   commands:
+      #     - |
+      #       timeout 40m bash -c '
+      #         export VLLM_IMAGE_FETCH_TIMEOUT=60
+      #         pytest -s -v tests/e2e/online_serving/test_qwen_image.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+      #       '
+      #   mirror_hardwares: h100_1
+
+      # - label: "Diffusion · Qwen Image Layered Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/online_serving/test_qwen_image_layered.py
+      #     - vllm_omni/diffusion/models/qwen_image/
+      #   commands:
+      #     - |
+      #       timeout 40m bash -c "
+      #         export VLLM_IMAGE_FETCH_TIMEOUT=60
+      #         pytest -s -v tests/e2e/online_serving/test_qwen_image_layered.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+      #       "
+      #   mirror_hardwares: h100_1
+
+      # - label: "Diffusion · Wan22 Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/offline_inference/test_wan22_t2v.py
+      #     - tests/e2e/online_serving/test_wan22_t2v.py
+      #     - vllm_omni/diffusion/models/wan2_2/
+      #   commands:
+      #     - |
+      #       timeout 40m bash -c "
+      #         export VLLM_IMAGE_FETCH_TIMEOUT=60
+      #         pytest -s -v tests/e2e/offline_inference/test_wan22_t2v.py tests/e2e/online_serving/test_wan22_t2v.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
+      #       "
+      #   mirror_hardwares: h100_1
+
+      # - label: "Diffusion · Z Image Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/offline_inference/test_z_image.py
+      #     - tests/e2e/online_serving/test_z_image.py
+      #     - vllm_omni/diffusion/models/z_image/
+      #   timeout_in_minutes: 30
+      #   commands:
+      #     - pytest -s -v tests/e2e/offline_inference/test_z_image.py tests/e2e/online_serving/test_z_image.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
+      #   mirror_hardwares: l4_1
+      # - label: "Diffusion · FLUX.2 Klein Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/offline_inference/test_flux2_klein.py
+      #     - vllm_omni/diffusion/models/flux2_klein/
+      #   timeout_in_minutes: 30
+      #   commands:
+      #     - pytest -s -v tests/e2e/offline_inference/test_flux2_klein.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
+      #   mirror_hardwares: l4_1

.buildkite/cuda/test-ready.yml

@@ -18,339 +18,302 @@ steps:
           - "timeout 20m pytest -sv -m 'core_model and cpu' --ignore=tests/diffusion --ignore=tests/model_executor --cov=vllm_omni --cov-report=term-missing:skip-covered --cov-report=xml"
         mirror_hardwares: l4_1
 
-  - label: "Custom Pipeline Test"
-    depends_on: upload-ready-pipeline
-    commands:
-      - timeout 20m pytest -s -v tests/e2e/offline_inference/custom_pipeline/ -m "core_model"
-    mirror_hardwares: l4_1
+  # - label: "Custom Pipeline Test"
+  #   depends_on: upload-ready-pipeline
+  #   commands:
+  #     - timeout 20m pytest -s -v tests/e2e/offline_inference/custom_pipeline/ -m "core_model"
+  #   mirror_hardwares: l4_1
 
-  - group: ":card_index_dividers: Diffusion Test"
-    depends_on: upload-ready-pipeline
-    steps:
+  # - group: ":card_index_dividers: Diffusion Test"
+  #   depends_on: upload-ready-pipeline
+  #   steps:
 
-      - label: "Diffusion · Offloader Test"
-        commands:
-          - timeout 40m pytest -s -v tests/diffusion/offloader -m "core_model and cuda" --run-level "core_model"
-        mirror_hardwares: l4_1
+  #     - label: "Diffusion · Offloader Test"
+  #       commands:
+  #         - timeout 40m pytest -s -v tests/diffusion/offloader -m "core_model and cuda" --run-level "core_model"
+  #       mirror_hardwares: l4_1
 
-      - label: "Diffusion · Cache Test"
-        commands:
-          - timeout 15m pytest -sv tests/diffusion/cache -m 'core_model and cuda' --run-level "core_model"
-        mirror_hardwares: l4_1
+  #     - label: "Diffusion · Cache Test"
+  #       commands:
+  #         - timeout 15m pytest -sv tests/diffusion/cache -m 'core_model and cuda' --run-level "core_model"
+  #       mirror_hardwares: l4_1
 
-      - label: "Diffusion · Batch Test"
-        commands:
-          - timeout 15m pytest -sv tests/diffusion/batching -m 'core_model and cuda' --run-level "core_model"
-        mirror_hardwares: l4_1
+  #     - label: "Diffusion · Batch Test"
+  #       commands:
+  #         - timeout 15m pytest -sv tests/diffusion/batching -m 'core_model and cuda' --run-level "core_model"
+  #       mirror_hardwares: l4_1
 
-      - label: "Diffusion · Distributed Test"
-        source_file_dependencies:
-          - tests/diffusion/distributed/
-          - vllm_omni/diffusion/distributed/
-          - vllm_omni/diffusion/attention/parallel/
-          - vllm_omni/diffusion/data.py
-          - vllm_omni/diffusion/models/
-        commands:
-          - timeout 20m pytest -sv tests/diffusion/distributed -m 'core_model and cuda' --run-level "core_model"
-        mirror_hardwares: l4_4
+  #     - label: "Diffusion · Distributed Test"
+  #       source_file_dependencies:
+  #         - tests/diffusion/distributed/
+  #         - vllm_omni/diffusion/distributed/
+  #         - vllm_omni/diffusion/attention/parallel/
+  #         - vllm_omni/diffusion/data.py
+  #         - vllm_omni/diffusion/models/
+  #       commands:
+  #         - timeout 20m pytest -sv tests/diffusion/distributed -m 'core_model and cuda' --run-level "core_model"
+  #       mirror_hardwares: l4_4
 
-      - label: "Diffusion · Model Test"
-        commands:
-          - timeout 15m pytest -sv tests/diffusion/models/ -m 'core_model and cuda' --run-level "core_model"
-        mirror_hardwares: l4_1
+  #     - label: "Diffusion · Model Test"
+  #       commands:
+  #         - timeout 15m pytest -sv tests/diffusion/models/ -m 'core_model and cuda' --run-level "core_model"
+  #       mirror_hardwares: l4_1
 
-      - label: "Diffusion · Other Test"
-        commands:
-          - timeout 15m pytest -sv tests/diffusion/*.py -m 'core_model and cuda' --run-level "core_model"
-        mirror_hardwares: l4_1
+  #     - label: "Diffusion · Other Test"
+  #       commands:
+  #         - timeout 15m pytest -sv tests/diffusion/*.py -m 'core_model and cuda' --run-level "core_model"
+  #       mirror_hardwares: l4_1
 
-  - group: ":card_index_dividers: Tiny Model Tests [Single GPU]"
-    depends_on: upload-ready-pipeline
-    steps:
+  # - group: ":card_index_dividers: Tiny Model Tests [Single GPU]"
+  #   depends_on: upload-ready-pipeline
+  #   steps:
 
-      - label: "Tiny Model Tests · Diffusion (base)"
-        commands:
-          - timeout 15m pytest -sv tests/model_tests/diffusion -m 'core_model and cuda' --run-level "core_model"
-        mirror_hardwares: l4_1
+  #     - label: "Tiny Model Tests · Diffusion (base)"
+  #       commands:
+  #         - timeout 15m pytest -sv tests/model_tests/diffusion -m 'core_model and cuda' --run-level "core_model"
+  #       mirror_hardwares: l4_1
 
-  - label: "Engine&Model Executor Test"
-    depends_on: upload-ready-pipeline
-    commands:
-      - |
-        timeout 15m bash -c "
-                  pytest -sv tests/engine/ tests/model_executor/ -m 'core_model and cuda' --run-level 'core_model'
-        "
-    mirror_hardwares: l4_1
+  # - label: "Engine&Model Executor Test"
+  #   depends_on: upload-ready-pipeline
+  #   commands:
+  #     - |
+  #       timeout 15m bash -c "
+  #                 pytest -sv tests/engine/ tests/model_executor/ -m 'core_model and cuda' --run-level 'core_model'
+  #       "
+  #   mirror_hardwares: l4_1
 
-  - label: "Distributed Test"
-    depends_on: upload-ready-pipeline
-    commands:
-      - |
-        timeout 20m bash -c "
-          pytest -sv tests/distributed/ -m 'core_model and cuda and L4' --run-level 'core_model'
-        "
-    mirror_hardwares: l4_1
+  # - label: "Distributed Test"
+  #   depends_on: upload-ready-pipeline
+  #   commands:
+  #     - |
+  #       timeout 20m bash -c "
+  #         pytest -sv tests/distributed/ -m 'core_model and cuda and L4' --run-level 'core_model'
+  #       "
+  #   mirror_hardwares: l4_1
 
-  - label: "Entrypoints Test"
-    depends_on: upload-ready-pipeline
-    commands:
-      - |
-        timeout 30m bash -c "
-          export VLLM_WORKER_MULTIPROC_METHOD=spawn
-          pytest -sv tests/entrypoints/ -m 'core_model and cuda' --run-level 'core_model'
-        "
-    mirror_hardwares: l4_1
+  # - label: "Entrypoints Test"
+  #   depends_on: upload-ready-pipeline
+  #   commands:
+  #     - |
+  #       timeout 30m bash -c "
+  #         export VLLM_WORKER_MULTIPROC_METHOD=spawn
+  #         pytest -sv tests/entrypoints/ -m 'core_model and cuda' --run-level 'core_model'
+  #       "
+  #   mirror_hardwares: l4_1
 
   - group: ":card_index_dividers: E2E Test"
     depends_on: upload-ready-pipeline
     steps:
 
-      - label: "RLHF · VeRL-Omni E2E Test"
-        source_file_dependencies:
-          - tests/e2e/offline_inference/rlhf_test/test_verl_omni_e2e.py
-          - tests/e2e/offline_inference/custom_pipeline/
-          - vllm_omni/diffusion/
-          - vllm_omni/engine/
-          - vllm_omni/entrypoints/
-          - vllm_omni/inputs/
-          - vllm_omni/outputs/
-        commands:
-          - |
-            timeout 20m bash -c "
-              pytest -s -v tests/e2e/offline_inference/rlhf_test/test_verl_omni_e2e.py
-            "
-        mirror_hardwares: l4_1
+      # - label: "RLHF · VeRL-Omni E2E Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/offline_inference/rlhf_test/test_verl_omni_e2e.py
+      #     - tests/e2e/offline_inference/custom_pipeline/
+      #     - vllm_omni/diffusion/
+      #     - vllm_omni/engine/
+      #     - vllm_omni/entrypoints/
+      #     - vllm_omni/inputs/
+      #     - vllm_omni/outputs/
+      #   commands:
+      #     - |
+      #       timeout 20m bash -c "
+      #         pytest -s -v tests/e2e/offline_inference/rlhf_test/test_verl_omni_e2e.py
+      #       "
+      #   mirror_hardwares: l4_1
 
-      - label: "Omni · Qwen3-Omni Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_qwen3_omni.py
-          - vllm_omni/model_executor/models/qwen3_omni/
-          - vllm_omni/model_executor/stage_input_processors/qwen3_omni.py
-          - vllm_omni/deploy/qwen3_omni_moe.yaml
-        commands:
-          - |
-            timeout 20m bash -c "
-              pytest -s -v tests/e2e/online_serving/test_qwen3_omni.py -m 'core_model' --run-level 'core_model'
-            "
-        mirror_hardwares: h100_2
+      # - label: "Omni · Qwen3-Omni Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/online_serving/test_qwen3_omni.py
+      #     - vllm_omni/model_executor/models/qwen3_omni/
+      #     - vllm_omni/model_executor/stage_input_processors/qwen3_omni.py
+      #     - vllm_omni/deploy/qwen3_omni_moe.yaml
+      #   commands:
+      #     - |
+      #       timeout 20m bash -c "
+      #         pytest -s -v tests/e2e/online_serving/test_qwen3_omni.py -m 'core_model' --run-level 'core_model'
+      #       "
+      #   mirror_hardwares: h100_2
 
       - label: "Omni · MiniCPM-o 4.5 Offline Test"
-        source_file_dependencies:
-          - tests/e2e/offline_inference/test_minicpmo_4_5.py
-          - vllm_omni/deploy/minicpmo_4_5.yaml
         commands:
           - |
             timeout 20m bash -c "
               pytest -s -v tests/e2e/offline_inference/test_minicpmo_4_5.py -m 'core_model' --run-level 'core_model'
             "
-        agents:
-          queue: "mithril-h100-pool"
-        plugins:
-          - kubernetes:
-              podSpec:
-                containers:
-                  - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                    resources:
-                      limits:
-                        nvidia.com/gpu: 2
-                    volumeMounts:
-                      - name: devshm
-                        mountPath: /dev/shm
-                      - name: hf-cache
-                        mountPath: /root/.cache/huggingface
-                    env:
-                      - name: HF_HOME
-                        value: /root/.cache/huggingface
-                nodeSelector:
-                  node.kubernetes.io/instance-type: gpu-h100-sxm
-                volumes:
-                  - name: devshm
-                    emptyDir:
-                      medium: Memory
-                  - name: hf-cache
-                    hostPath:
-                      path: /mnt/hf-cache
-                      type: DirectoryOrCreate
+        mirror_hardwares: h100_1
 
       - label: "Omni · MiniCPM-o 4.5 Online Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_minicpmo_4_5.py
-          - vllm_omni/deploy/minicpmo_4_5.yaml
         commands:
           - |
             timeout 20m bash -c "
               pytest -s -v tests/e2e/online_serving/test_minicpmo_4_5.py -m 'core_model' --run-level 'core_model'
             "
-        agents:
-          queue: "mithril-h100-pool"
-        plugins:
-          - kubernetes:
-              podSpec:
-                containers:
-                  - image: 936637512419.dkr.ecr.us-west-2.amazonaws.com/vllm-ci-pull-through-cache/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-                    resources:
-                      limits:
-                        nvidia.com/gpu: 2
-                    volumeMounts:
-                      - name: devshm
-                        mountPath: /dev/shm
-                      - name: hf-cache
-                        mountPath: /root/.cache/huggingface
-                    env:
-                      - name: HF_HOME
-                        value: /root/.cache/huggingface
-                nodeSelector:
-                  node.kubernetes.io/instance-type: gpu-h100-sxm
-                volumes:
-                  - name: devshm
-                    emptyDir:
-                      medium: Memory
-                  - name: hf-cache
-                    hostPath:
-                      path: /mnt/hf-cache
-                      type: DirectoryOrCreate
+        mirror_hardwares: h100_1
 
-      - label: "TTS · VoxCPM2 Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_voxcpm2_tts.py
-          - vllm_omni/model_executor/models/voxcpm2/
-          - vllm_omni/deploy/voxcpm2.yaml
-        timeout_in_minutes: 30
-        commands:
-          - |
-            timeout 30m bash -c "
-              export VLLM_LOGGING_LEVEL=DEBUG
-              export VLLM_WORKER_MULTIPROC_METHOD=spawn
-              pytest -s -v tests/e2e/online_serving/test_voxcpm2_tts.py -m 'core_model' --run-level 'core_model'
-            "
-        mirror_hardwares: l4_1
-
-      - label: "TTS · Qwen3-TTS CustomVoice Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_qwen3_tts_customvoice.py
-          - vllm_omni/model_executor/models/qwen3_tts/
-          - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
-          - vllm_omni/deploy/qwen3_tts.yaml
+      - label: "Omni · MiniCPM-o 4.5 Duplex Test"
         commands:
           - |
             timeout 20m bash -c "
-              export VLLM_ALLOW_LONG_MAX_MODEL_LEN=1
-              pytest -s -v tests/e2e/online_serving/test_qwen3_tts_customvoice.py -m 'core_model' --run-level 'core_model'
-            "
-        mirror_hardwares: l4_1
-
-      - label: "TTS · Qwen3-TTS Base Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_qwen3_tts_base.py
-          - vllm_omni/model_executor/models/qwen3_tts/
-          - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
-          - vllm_omni/deploy/qwen3_tts.yaml
-        commands:
-          - |
-            timeout 20m bash -c "
-              export VLLM_WORKER_MULTIPROC_METHOD=spawn
-              export VLLM_ALLOW_LONG_MAX_MODEL_LEN=1
-              export VLLM_OMNI_USE_V2_RUNNER=1
-              pytest -s -v tests/e2e/online_serving/test_qwen3_tts_base.py -m 'core_model' --run-level 'core_model'
-            "
-        mirror_hardwares: l4_1
-
-      - label: "TTS · Higgs-Audio-V3 Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_higgs_audio_v3.py
-          - vllm_omni/model_executor/models/higgs_audio_v3/
-          - vllm_omni/model_executor/stage_input_processors/higgs_audio_v3.py
-          - vllm_omni/deploy/higgs_multimodal_qwen3.yaml
-        commands:
-          - |
-            timeout 20m bash -c "
-              export VLLM_WORKER_MULTIPROC_METHOD=spawn
-              export VLLM_USE_DEEP_GEMM=0
-              export VLLM_MOE_USE_DEEP_GEMM=0
-              pytest -s -v tests/e2e/online_serving/test_higgs_audio_v3.py -m 'core_model' --run-level 'core_model'
-            "
-        mirror_hardwares: l4_1
-
-      - label: "Diffusion · Qwen Image Edit Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_qwen_image_edit.py
-          - vllm_omni/diffusion/models/qwen_image/
-        commands:
-          - |
-            timeout 20m bash -c "
-              pytest -s -v tests/e2e/online_serving/test_qwen_image_edit.py -m 'core_model' --run-level 'core_model'
+              pytest -s -v tests/e2e/online_serving/test_minicpmo_4_5_duplex.py -m 'core_model and cuda' --run-level 'core_model'
             "
         mirror_hardwares: h100_1
 
-      - label: "Diffusion · Bagel Test"
+      - label: "Omni · MiniCPM-o 4.5 Perf Test"
+        timeout_in_minutes: 60
         source_file_dependencies:
-          - tests/e2e/online_serving/test_bagel.py
-          - vllm_omni/model_executor/models/bagel/
-          - vllm_omni/diffusion/models/bagel/
-          - vllm_omni/model_executor/stage_input_processors/bagel.py
-          - vllm_omni/deploy/bagel.yaml
+          - tests/dfx/perf/tests/test_minicpmo_4_5_ready.json
+          - vllm_omni/deploy/minicpmo_4_5.yaml
+          - vllm_omni/model_executor/models/minicpmo_4_5/
+          - vllm_omni/model_executor/models/step_audio2/
+          - vllm_omni/model_executor/models/cosyvoice3/code2wav_core/
+          - vllm_omni/model_executor/models/cosyvoice3/utils.py
+          - vllm_omni/model_executor/stage_input_processors/minicpmo_4_5_omni.py
         commands:
-          - |
-            timeout 40m bash -c "
-              export VLLM_IMAGE_FETCH_TIMEOUT=60
-              pytest -s -v tests/e2e/online_serving/test_bagel.py -m 'core_model' --run-level 'core_model'
-            "
+          - export BENCHMARK_DIR=tests/dfx/perf/results
+          - pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_minicpmo_4_5_ready.json -m 'core_model and cuda'
         mirror_hardwares: h100_1
 
-      - label: "Diffusion · Wan22 Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_wan22_t2v.py
-          - vllm_omni/diffusion/models/wan2_2/
-        commands:
-          - |
-            timeout 40m bash -c "
-              export VLLM_IMAGE_FETCH_TIMEOUT=60
-              pytest -s -v tests/e2e/online_serving/test_wan22_t2v.py -m 'core_model' --run-level 'core_model'
-            "
-        mirror_hardwares: h100_1
+      # - label: "TTS · VoxCPM2 Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/online_serving/test_voxcpm2_tts.py
+      #     - vllm_omni/model_executor/models/voxcpm2/
+      #     - vllm_omni/deploy/voxcpm2.yaml
+      #   timeout_in_minutes: 30
+      #   commands:
+      #     - |
+      #       timeout 30m bash -c "
+      #         export VLLM_LOGGING_LEVEL=DEBUG
+      #         export VLLM_WORKER_MULTIPROC_METHOD=spawn
+      #         pytest -s -v tests/e2e/online_serving/test_voxcpm2_tts.py -m 'core_model' --run-level 'core_model'
+      #       "
+      #   mirror_hardwares: l4_1
 
-      - label: "Diffusion · Cosmos3 Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_cosmos3.py
-          - vllm_omni/diffusion/models/cosmos3/
-        commands:
-          - |
-            timeout 40m bash -c "
-              export VLLM_IMAGE_FETCH_TIMEOUT=60
-              pytest -s -v tests/e2e/online_serving/test_cosmos3.py -m 'core_model' --run-level 'core_model'
-            "
-        mirror_hardwares: h100_1
+      # - label: "TTS · Qwen3-TTS CustomVoice Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/online_serving/test_qwen3_tts_customvoice.py
+      #     - vllm_omni/model_executor/models/qwen3_tts/
+      #     - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
+      #     - vllm_omni/deploy/qwen3_tts.yaml
+      #   commands:
+      #     - |
+      #       timeout 20m bash -c "
+      #         export VLLM_ALLOW_LONG_MAX_MODEL_LEN=1
+      #         pytest -s -v tests/e2e/online_serving/test_qwen3_tts_customvoice.py -m 'core_model' --run-level 'core_model'
+      #       "
+      #   mirror_hardwares: l4_1
 
-      - label: "Diffusion · Qwen Image Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_qwen_image.py
-          - vllm_omni/diffusion/models/qwen_image/
-        commands:
-          - |
-            timeout 40m bash -c "
-              export VLLM_IMAGE_FETCH_TIMEOUT=60
-              pytest -s -v tests/e2e/online_serving/test_qwen_image.py -m 'core_model' --run-level 'core_model'
-            "
-        mirror_hardwares: h100_1
+      # - label: "TTS · Qwen3-TTS Base Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/online_serving/test_qwen3_tts_base.py
+      #     - vllm_omni/model_executor/models/qwen3_tts/
+      #     - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
+      #     - vllm_omni/deploy/qwen3_tts.yaml
+      #   commands:
+      #     - |
+      #       timeout 20m bash -c "
+      #         export VLLM_WORKER_MULTIPROC_METHOD=spawn
+      #         export VLLM_ALLOW_LONG_MAX_MODEL_LEN=1
+      #         export VLLM_OMNI_USE_V2_RUNNER=1
+      #         pytest -s -v tests/e2e/online_serving/test_qwen3_tts_base.py -m 'core_model' --run-level 'core_model'
+      #       "
+      #   mirror_hardwares: l4_1
 
-      - label: "Diffusion · Qwen Image Layered Test"
-        source_file_dependencies:
-          - tests/e2e/online_serving/test_qwen_image_layered.py
-          - vllm_omni/diffusion/models/qwen_image/
-        commands:
-          - |
-            timeout 40m bash -c "
-              export VLLM_IMAGE_FETCH_TIMEOUT=60
-              pytest -s -v tests/e2e/online_serving/test_qwen_image_layered.py -m 'core_model' --run-level 'core_model'
-            "
-        mirror_hardwares: h100_1
+      # - label: "TTS · Higgs-Audio-V3 Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/online_serving/test_higgs_audio_v3.py
+      #     - vllm_omni/model_executor/models/higgs_audio_v3/
+      #     - vllm_omni/model_executor/stage_input_processors/higgs_audio_v3.py
+      #     - vllm_omni/deploy/higgs_multimodal_qwen3.yaml
+      #   commands:
+      #     - |
+      #       timeout 20m bash -c "
+      #         export VLLM_WORKER_MULTIPROC_METHOD=spawn
+      #         export VLLM_USE_DEEP_GEMM=0
+      #         export VLLM_MOE_USE_DEEP_GEMM=0
+      #         pytest -s -v tests/e2e/online_serving/test_higgs_audio_v3.py -m 'core_model' --run-level 'core_model'
+      #       "
+      #   mirror_hardwares: l4_1
 
-      - label: "Diffusion · GGUF Plugin Test"
-        soft_fail: true
-        source_file_dependencies:
-          - tests/plugin/test_gguf_diffusion.py
-          - vllm_omni/diffusion/model_loader
-        commands:
-          - pip install "vllm-gguf-plugin>=0.0.3"
-          - timeout 30m pytest -sv tests/plugin/test_gguf_diffusion.py
-        mirror_hardwares: l4_1
+      # - label: "Diffusion · Qwen Image Edit Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/online_serving/test_qwen_image_edit.py
+      #     - vllm_omni/diffusion/models/qwen_image/
+      #   commands:
+      #     - |
+      #       timeout 20m bash -c "
+      #         pytest -s -v tests/e2e/online_serving/test_qwen_image_edit.py -m 'core_model' --run-level 'core_model'
+      #       "
+      #   mirror_hardwares: h100_1
+
+      # - label: "Diffusion · Bagel Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/online_serving/test_bagel.py
+      #     - vllm_omni/model_executor/models/bagel/
+      #     - vllm_omni/diffusion/models/bagel/
+      #     - vllm_omni/model_executor/stage_input_processors/bagel.py
+      #     - vllm_omni/deploy/bagel.yaml
+      #   commands:
+      #     - |
+      #       timeout 40m bash -c "
+      #         export VLLM_IMAGE_FETCH_TIMEOUT=60
+      #         pytest -s -v tests/e2e/online_serving/test_bagel.py -m 'core_model' --run-level 'core_model'
+      #       "
+      #   mirror_hardwares: h100_1
+
+      # - label: "Diffusion · Wan22 Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/online_serving/test_wan22_t2v.py
+      #     - vllm_omni/diffusion/models/wan2_2/
+      #   commands:
+      #     - |
+      #       timeout 40m bash -c "
+      #         export VLLM_IMAGE_FETCH_TIMEOUT=60
+      #         pytest -s -v tests/e2e/online_serving/test_wan22_t2v.py -m 'core_model' --run-level 'core_model'
+      #       "
+      #   mirror_hardwares: h100_1
+
+      # - label: "Diffusion · Cosmos3 Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/online_serving/test_cosmos3.py
+      #     - vllm_omni/diffusion/models/cosmos3/
+      #   commands:
+      #     - |
+      #       timeout 40m bash -c "
+      #         export VLLM_IMAGE_FETCH_TIMEOUT=60
+      #         pytest -s -v tests/e2e/online_serving/test_cosmos3.py -m 'core_model' --run-level 'core_model'
+      #       "
+      #   mirror_hardwares: h100_1
+
+      # - label: "Diffusion · Qwen Image Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/online_serving/test_qwen_image.py
+      #     - vllm_omni/diffusion/models/qwen_image/
+      #   commands:
+      #     - |
+      #       timeout 40m bash -c "
+      #         export VLLM_IMAGE_FETCH_TIMEOUT=60
+      #         pytest -s -v tests/e2e/online_serving/test_qwen_image.py -m 'core_model' --run-level 'core_model'
+      #       "
+      #   mirror_hardwares: h100_1
+
+      # - label: "Diffusion · Qwen Image Layered Test"
+      #   source_file_dependencies:
+      #     - tests/e2e/online_serving/test_qwen_image_layered.py
+      #     - vllm_omni/diffusion/models/qwen_image/
+      #   commands:
+      #     - |
+      #       timeout 40m bash -c "
+      #         export VLLM_IMAGE_FETCH_TIMEOUT=60
+      #         pytest -s -v tests/e2e/online_serving/test_qwen_image_layered.py -m 'core_model' --run-level 'core_model'
+      #       "
+      #   mirror_hardwares: h100_1
+
+      # - label: "Diffusion · GGUF Plugin Test"
+      #   soft_fail: true
+      #   source_file_dependencies:
+      #     - tests/plugin/test_gguf_diffusion.py
+      #     - vllm_omni/diffusion/model_loader
+      #   commands:
+      #     - pip install "vllm-gguf-plugin>=0.0.3"
+      #     - timeout 30m pytest -sv tests/plugin/test_gguf_diffusion.py
+      #   mirror_hardwares: l4_1

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

@@ -4,6 +4,10 @@
 
 set -euo pipefail
 
+# minicpm-challenge branch: skip entire Intel/XPU CI (no pipeline upload => no image build / tests).
+echo "--- Skipping Intel/XPU CI on this branch (including image build)"
+exit 0
+
 source .buildkite/common/scripts/resolve_skip_ci.sh
 
 upload_pipeline() {

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

@@ -1,58 +1,117 @@
+env:
+  VLLM_WORKER_MULTIPROC_METHOD: spawn
+  VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
+  HF_HUB_DOWNLOAD_TIMEOUT: 300
+  HF_HUB_ETAG_TIMEOUT: 60
+
 steps:
+  # - group: ":card_index_dividers: Diffusion X2V Model Test"
+  #   key: nightly-diffusion-x2v-group
+  #   depends_on: upload-nightly-pipeline
+  #   if: >-
+  #     build.env("NIGHTLY") == "1" ||
+  #     build.pull_request.labels includes "nightly-test" ||
+  #     build.pull_request.labels includes "diffusion-x2v-test"
+  #   steps:
+  #     - label: ":full_moon: Diffusion X2V · Function Test"
+  #       mirror_hardwares: a2b3_npu_4
+  #       env:
+  #         VLLM_WORKER_MULTIPROC_METHOD: spawn
+  #         HF_TOKEN: "${HF_TOKEN}"
+  #       commands:
+  #         - pytest -s -v tests/e2e/online_serving/test_wan22_expansion.py -k npu_i2v --run-level "full_model"
+  # - group: ":card_index_dividers: TTS Model Test"
+  #   key: nightly-tts-test-group
+  #   depends_on: upload-nightly-pipeline
+  #   if: build.env("NIGHTLY") == "1" || build.pull_request.labels includes "nightly-test" || build.pull_request.labels includes "tts-test"
+  #   steps:
+  #     - label: ":full_moon: TTS · Perf Test"
+  #       mirror_hardwares: a3_npu_2
+  #       env:
+  #         VLLM_WORKER_MULTIPROC_METHOD: spawn
+  #         VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
+  #         HF_TOKEN: "${HF_TOKEN}"
+  #       commands:
+  #         - export BENCHMARK_DIR=tests/dfx/perf/results
+  #         - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
+  #         - |
+  #           set +e
+  #           pytest -s -v tests/dfx/perf/scripts/run_benchmark.py -m npu --test-config-file tests/dfx/perf/tests/test_tts.json
+  #           EXIT=$$?
+  #           buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
+  #           exit $$EXIT
+
   - group: ":card_index_dividers: Omni Model Test"
-    key: nightly-omni-test-group
+    key: nightly-omni-group
     depends_on: upload-nightly-pipeline
-    if: >-
-      build.env("NIGHTLY") == "1" ||
-      build.pull_request.labels includes "nightly-test" ||
-      build.pull_request.labels includes "omni-test"
+    if: build.env("NIGHTLY") == "1" || build.pull_request.labels includes "nightly-test" || build.pull_request.labels includes "omni-test"
     steps:
-      - label: ":full_moon: Omni · Perf Test · No Async Chunk"
-        key: nightly-omni-performance-no-async-chunk
+      - label: ":full_moon: Omni · MiniCPM-o 4.5 Test"
         timeout_in_minutes: 180
-        mirror_hardwares: a3_npu_4
-        env:
-          VLLM_WORKER_MULTIPROC_METHOD: spawn
-          VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
-          HF_TOKEN: "${HF_TOKEN}"
-        commands:
-          - export BENCHMARK_DIR=tests/dfx/perf/results
-          - |
-            set +e
-            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py -m npu --test-config-file tests/dfx/perf/tests/test_qwen3_omni_no_async_chunk.json
-            EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
-            exit $$EXIT
-      - label: ":full_moon: Omni · Perf Test · Async Chunk"
-        key: nightly-omni-performance-async-chunk
-        timeout_in_minutes: 180
-        mirror_hardwares: a3_npu_4
-        env:
-          VLLM_WORKER_MULTIPROC_METHOD: spawn
-          VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
-          HF_TOKEN: "${HF_TOKEN}"
-        commands:
-          - export BENCHMARK_DIR=tests/dfx/perf/results
-          - |
-            set +e
-            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py -m npu --test-config-file tests/dfx/perf/tests/test_qwen3_omni_async_chunk.json
-            EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
-            exit $$EXIT
-      - label: ":robot: Omni · MiniCPM-o 4.5 Test"
         mirror_hardwares: a3_npu_2
         env:
           VLLM_WORKER_MULTIPROC_METHOD: spawn
           VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
           HF_TOKEN: "${HF_TOKEN}"
         commands:
-          - pytest -s -v tests/e2e/offline_inference/test_minicpmo_4_5.py tests/e2e/online_serving/test_minicpmo_4_5.py tests/e2e/online_serving/test_minicpmo_4_5_expansion.py -m 'full_model' --run-level 'full_model'
-  - group: ":card_index_dividers: TTS Model Test"
-    key: nightly-tts-test-group
-    depends_on: upload-nightly-pipeline
-    if: build.env("NIGHTLY") == "1" || build.pull_request.labels includes "nightly-test" || build.pull_request.labels includes "tts-test"
-    steps:
-      - label: ":full_moon: TTS · Perf Test"
+          - pytest -s -v tests/e2e/offline_inference/test_minicpmo_4_5.py tests/e2e/online_serving/test_minicpmo_4_5.py tests/e2e/online_serving/test_minicpmo_4_5_expansion.py -m 'full_model and npu and A3 and omni' --run-level 'full_model'
+      - label: ":full_moon: Omni · MiniCPM-o 4.5 · Duplex Test"
+        timeout_in_minutes: 180
+        mirror_hardwares: a3_npu_2
+        env:
+          VLLM_WORKER_MULTIPROC_METHOD: spawn
+          VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
+          HF_TOKEN: "${HF_TOKEN}"
+        commands:
+          # NPU has no merge-level pipeline, so advanced_model duplex coverage
+          # rides on nightly alongside full_model.
+          #
+          # test_duplex_soft_interrupt is CUDA-only and carries no npu mark, so
+          # the marker filter below drops it: it needs real-time duplex
+          # throughput that the current NPU stack does not reach.
+          - >-
+            pytest -s -v
+            tests/e2e/online_serving/test_minicpmo_4_5_duplex.py
+            tests/e2e/online_serving/test_minicpmo_4_5_duplex_expansion.py
+            -m "(full_model or advanced_model) and npu and A3 and omni"
+            --run-level full_model
+      - label: ":full_moon: Omni · MiniCPM-o 4.5 · Accuracy Test"
+        timeout_in_minutes: 180
+        mirror_hardwares: a3_npu_2
+        env:
+          VLLM_WORKER_MULTIPROC_METHOD: spawn
+          VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
+          HF_TOKEN: "${HF_TOKEN}"
+        commands:
+          - export SEED_TTS_WER_EVAL=1
+          - export SEED_TTS_SIM_EVAL=1
+          - export SEED_TTS_EVAL_DEVICE=npu:1
+          - |
+            set +e
+            pytest -s -v tests/e2e/accuracy/minicpmo_4_5/test_minicpmo_4_5.py -k "not videomme" -m "full_model and npu and A3" --run-level full_model
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/e2e/accuracy/minicpmo_4_5/results/*.json"
+            exit $$EXIT
+      - label: ":full_moon: Omni · MiniCPM-o 4.5 · Video-MME Accuracy Test"
+        key: nightly-omni-accuracy-minicpmo-4-5-videomme-npu
+        timeout_in_minutes: 180
+        mirror_hardwares: a3_npu_2
+        env:
+          VLLM_WORKER_MULTIPROC_METHOD: spawn
+          VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
+          HF_TOKEN: "${HF_TOKEN}"
+        commands:
+          # Video-MME needs no repo override here: the NPU runners pre-stage the dataset under
+          # the lmms-eval mirror, which is now VIDEOMME_DEFAULT_HF_REPO, so snapshot_download
+          # resolves it straight out of the hub cache.
+          - |
+            set +e
+            pytest -s -v tests/e2e/accuracy/minicpmo_4_5/test_minicpmo_4_5.py -k videomme -m "full_model and npu and A3" --run-level full_model
+            EXIT=$$?
+            buildkite-agent artifact upload "tests/e2e/accuracy/minicpmo_4_5/results/*.json"
+            exit $$EXIT
+      - label: ":full_moon: Omni · MiniCPM-o 4.5 · Perf Test"
+        key: nightly-omni-performance-minicpmo-4-5-npu
         timeout_in_minutes: 180
         mirror_hardwares: a3_npu_2
         env:
@@ -61,63 +120,42 @@ steps:
           HF_TOKEN: "${HF_TOKEN}"
         commands:
           - export BENCHMARK_DIR=tests/dfx/perf/results
-          - export VLLM_ALLOW_LONG_MAX_MODEL_LEN="1"
           - |
             set +e
-            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py -m npu --test-config-file tests/dfx/perf/tests/test_tts.json
+            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py -m "npu and A3" --test-config-file tests/dfx/perf/tests/test_minicpmo_4_5.json
             EXIT=$$?
             buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
             exit $$EXIT
-  - group: ":card_index_dividers: Diffusion X2I(&A&T) Model Test"
-    key: nightly-diffusion-x2iat-group
-    depends_on: upload-nightly-pipeline
-    if: >-
-      build.env("NIGHTLY") == "1" ||
-      build.pull_request.labels includes "nightly-test" ||
-      build.pull_request.labels includes "diffusion-x2iat-test"
-    steps:
-      - label: ":full_moon: Diffusion X2I(&A&T) · Perf Test · Qwen-Image-Edit-2511"
-        key: nightly-diffusion-x2iat-performance-qwen-image-edit-2511
+      - label: ":full_moon: Omni · MiniCPM-o 4.5 · Perf Test (A2)"
+        key: nightly-omni-performance-minicpmo-4-5-npu-a2
         timeout_in_minutes: 180
-        mirror_hardwares: a3_npu_8
+        mirror_hardwares: a2b3_npu_1
         env:
           VLLM_WORKER_MULTIPROC_METHOD: spawn
           VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
           HF_TOKEN: "${HF_TOKEN}"
-          DIFFUSION_ATTENTION_BACKEND: TORCH_SDPA
+          HF_HUB_OFFLINE: "1"
         commands:
-          - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
-          - export CACHE_DIT_VERSION=1.3.0
+          - export BENCHMARK_DIR=tests/dfx/perf/results
           - |
             set +e
-            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py -m npu --test-config-file tests/dfx/perf/tests/test_qwen_image_edit_2511_vllm_omni.json
+            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py -m "npu and A2" --test-config-file tests/dfx/perf/tests/test_minicpmo_4_5.json
             EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/diffusion_result_*.json"
-            buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
+            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
             exit $$EXIT
-  - group: ":card_index_dividers: Diffusion X2V Model Test"
-    key: nightly-diffusion-x2v-group
-    depends_on: upload-nightly-pipeline
-    if: >-
-      build.env("NIGHTLY") == "1" ||
-      build.pull_request.labels includes "nightly-test" ||
-      build.pull_request.labels includes "diffusion-x2v-test"
-    steps:
-      - label: ":full_moon: Diffusion X2V · Perf Test"
-        key: nightly-diffusion-x2v-performance
+      - label: ":full_moon: Omni · MiniCPM-o 4.5 · Duplex Seed-TTS Perf Test"
+        key: nightly-omni-performance-minicpmo-4-5-duplex-seed-tts-npu
         timeout_in_minutes: 180
-        mirror_hardwares: a3_npu_4
+        mirror_hardwares: a3_npu_2
         env:
           VLLM_WORKER_MULTIPROC_METHOD: spawn
           VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
           HF_TOKEN: "${HF_TOKEN}"
-          DIFFUSION_ATTENTION_BACKEND: TORCH_SDPA
         commands:
-          - export DIFFUSION_BENCHMARK_DIR=tests/dfx/perf/results
+          - export BENCHMARK_DIR=tests/dfx/perf/results
           - |
             set +e
-            pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py -m npu --test-config-file tests/dfx/perf/tests/test_wan22_i2v_vllm_omni.json
+            pytest -s -v tests/dfx/perf/scripts/run_benchmark.py -m "npu and A3" --test-config-file tests/dfx/perf/tests/test_minicpmo_4_5_duplex_seed_tts.json
             EXIT=$$?
-            buildkite-agent artifact upload "tests/dfx/perf/results/diffusion_result_*.json"
-            buildkite-agent artifact upload "tests/dfx/perf/results/logs/*.log"
+            buildkite-agent artifact upload "tests/dfx/perf/results/*.json"
             exit $$EXIT

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

@@ -3,32 +3,41 @@ steps:
     key: ready-omni-group
     depends_on: upload-ready-pipeline
     steps:
-      - label: ":robot: Omni · MiniCPM-o 4.5 Test"
-        agents:
-          queue: "ascend-a3"
-          resource_class: "npu-2"
-        image: "swr.cn-southwest-2.myhuaweicloud.com/modelfoundry/vllm-omni-ci-npu:${BUILDKITE_COMMIT}-a3"
-        plugins:
-          - kubernetes:
-              podSpecPatch:
-                imagePullSecrets:
-                  - name: swr-secret
+      - label: "Omni · MiniCPM-o 4.5 Test"
+        mirror_hardwares: a3_npu_2
         env:
           VLLM_WORKER_MULTIPROC_METHOD: spawn
           VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
           HF_TOKEN: "${HF_TOKEN}"
         commands:
-          - pytest -s -v tests/e2e/offline_inference/test_minicpmo_4_5.py tests/e2e/online_serving/test_minicpmo_4_5.py -m 'core_model' --run-level 'core_model'
+          - pytest -s -v tests/e2e/offline_inference/test_minicpmo_4_5.py tests/e2e/online_serving/test_minicpmo_4_5.py -m 'core_model and npu and A3 and omni' --run-level 'core_model'
+      - label: "Omni · MiniCPM-o 4.5 Duplex Test"
+        mirror_hardwares: a3_npu_2
+        env:
+          VLLM_WORKER_MULTIPROC_METHOD: spawn
+          VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
+          HF_TOKEN: "${HF_TOKEN}"
+        commands:
+          - pytest -s -v tests/e2e/online_serving/test_minicpmo_4_5_duplex.py -m 'core_model and npu and A3 and omni' --run-level 'core_model'
+      - label: "Omni · MiniCPM-o 4.5 Perf Test"
+        timeout_in_minutes: 60
+        mirror_hardwares: a3_npu_2
+        env:
+          VLLM_WORKER_MULTIPROC_METHOD: spawn
+          VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
+        commands:
+          - export BENCHMARK_DIR=tests/dfx/perf/results
+          - pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_minicpmo_4_5_ready.json -m 'core_model and npu and A3 and omni'
 
-  - group: ":card_index_dividers: TTS Model Test"
-    key: ready-tts-group
-    depends_on: upload-ready-pipeline
-    steps:
-      - label: ":sound: TTS · Qwen3-TTS CustomVoice Test"
-        mirror_hardwares: a2b3_npu_1
-        env:
-          VLLM_WORKER_MULTIPROC_METHOD: spawn
-          VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
-          HF_TOKEN: "${HF_TOKEN}"
-        commands:
-          - pytest -s -v tests/e2e/online_serving/test_qwen3_tts_customvoice.py -m 'core_model and npu' --run-level 'core_model'
+  # - group: ":card_index_dividers: TTS Model Test"
+  #   key: ready-tts-group
+  #   depends_on: upload-ready-pipeline
+  #   steps:
+  #     - label: ":sound: TTS · Qwen3-TTS CustomVoice Test"
+  #       mirror_hardwares: a2b3_npu_1
+  #       env:
+  #         VLLM_WORKER_MULTIPROC_METHOD: spawn
+  #         VLLM_ALLOW_LONG_MAX_MODEL_LEN: "1"
+  #         HF_TOKEN: "${HF_TOKEN}"
+  #       commands:
+  #         - pytest -s -v tests/e2e/online_serving/test_qwen3_tts_customvoice.py -m 'core_model' --run-level 'core_model'

docker/Dockerfile.ci

@@ -1,7 +1,7 @@
-# Target vllm version: v0.26.0 (releases/v0.26.0 branch).
+# Target vllm version: v0.25.0 (releases/v0.25.0 branch).
 # Keep VLLM_BASE_TAG aligned with the target vLLM release on every rebase.
 ARG VLLM_BASE_IMAGE=vllm/vllm-openai
-ARG VLLM_BASE_TAG=v0.26.0
+ARG VLLM_BASE_TAG=v0.25.0
 
 # We use registry-based layer caching for the dependencies,
 # where the tag is determined based on the hash of the contents

docker/Dockerfile.cuda

@@ -1,4 +1,4 @@
-ARG BASE_IMAGE=vllm/vllm-openai:v0.26.0
+ARG BASE_IMAGE=vllm/vllm-openai:v0.25.0
 FROM ${BASE_IMAGE}
 
 ARG COMMON_WORKDIR=/app

docker/Dockerfile.npu.ci

@@ -32,18 +32,20 @@ RUN source /usr/local/Ascend/ascend-toolkit/set_env.sh && \
       "pytest-cov>=4.0.0" \
       "pytest-mock>=3.10.0" \
       "openpyxl>=3.0.0" \
+      "opencc==1.4.1" \
       "psutil>=7.2.0" \
       "soundfile>=0.13.1" \
       "pyttsx3>=2.99" \
       "opencv-python>=4.12.0.88" \
       "imageio[ffmpeg]>=0.6.0" \
-      "jiwer>=3.0.0" \
-      "zhon>=2.0.0" \
-      "zhconv>=1.4.2" \
-      "funasr>=1.0.0" \
       "hyperpyyaml>=1.2.0" \
       "onnx>=1.16.0" \
       "librosa>=0.10.0" \
+      "jiwer>=4.0.0" \
+      "zhon>=2.1.1" \
+      "zhconv>=1.4.3" \
+      "funasr>=1.3.0" \
+      "datasets>=3.0.0" \
       --index-url http://cache-service.nginx-pypi-cache.svc.cluster.local/pypi/simple \
       --trusted-host cache-service.nginx-pypi-cache.svc.cluster.local \
       --default-timeout=60 && \

docker/Dockerfile.npu.ci.a3

@@ -32,18 +32,20 @@ RUN source /usr/local/Ascend/ascend-toolkit/set_env.sh && \
       "pytest-cov>=4.0.0" \
       "pytest-mock>=3.10.0" \
       "openpyxl>=3.0.0" \
+      "opencc==1.4.1" \
       "psutil>=7.2.0" \
       "soundfile>=0.13.1" \
       "pyttsx3>=2.99" \
       "opencv-python>=4.12.0.88" \
       "imageio[ffmpeg]>=0.6.0" \
-      "jiwer>=3.0.0" \
-      "zhon>=2.0.0" \
-      "zhconv>=1.4.2" \
-      "funasr>=1.0.0" \
       "hyperpyyaml>=1.2.0" \
       "onnx>=1.16.0" \
       "librosa>=0.10.0" \
+      "jiwer>=4.0.0" \
+      "zhon>=2.1.1" \
+      "zhconv>=1.4.3" \
+      "funasr>=1.3.0" \
+      "datasets>=3.0.0" \
       --index-url http://cache-service.nginx-pypi-cache.svc.cluster.local/pypi/simple \
       --trusted-host cache-service.nginx-pypi-cache.svc.cluster.local \
       --default-timeout=60 && \

docker/Dockerfile.rocm

@@ -1,4 +1,4 @@
-ARG BASE_IMAGE=vllm/vllm-openai-rocm:v0.26.0
+ARG BASE_IMAGE=vllm/vllm-openai-rocm:v0.25.0
 FROM ${BASE_IMAGE} AS base
 
 # Declare a variable to know if we want to use the nightly build or the stable build.
@@ -10,7 +10,7 @@ FROM ${BASE_IMAGE} AS base
 #    we should swap over to use stable release ASAP.
 #    We should avoid relying on custom commits.
 ARG USE_NIGHTLY_BUILD=0
-ARG VLLM_VERSION_OR_COMMIT_HASH=v0.26.0
+ARG VLLM_VERSION_OR_COMMIT_HASH=89138b21cc246ae944c741d5c399c148e2b770ab
 ARG ARG_PYTORCH_ROCM_ARCH
 ENV PYTORCH_ROCM_ARCH=${ARG_PYTORCH_ROCM_ARCH:-${PYTORCH_ROCM_ARCH}}
 

docker/Dockerfile.xpu

@@ -88,7 +88,7 @@ ENV UV_EXTRA_INDEX_URL=${PIP_EXTRA_INDEX_URL}
 ENV UV_INDEX_STRATEGY="unsafe-best-match"
 ENV UV_LINK_MODE="copy"
 
-ARG VLLM_VERSION=v0.26.0
+ARG VLLM_VERSION=v0.25.0
 RUN git clone -b ${VLLM_VERSION} https://github.com/vllm-project/vllm
 WORKDIR /workspace/vllm
 

docs/contributing/ci/test_examples/l4_performance_tests.inc.md

@@ -44,9 +44,11 @@ Pass **`--test-config-file`** to run one JSON file as-is, or omit it for the bul
             "ignore_eos": true,
             "percentile-metrics": "ttft,tpot,itl,e2el,audio_rtf,audio_ttfp,audio_duration",
             "baseline": {
-                "mean_ttft_ms": [500, 800],
-                "mean_audio_ttfp_ms": [2000, 3500],
-                "mean_audio_rtf": [0.25, 0.35]
+                "H100": {
+                    "mean_ttft_ms": [500, 800],
+                    "mean_audio_ttfp_ms": [2000, 3500],
+                    "mean_audio_rtf": [0.25, 0.35]
+                }
             }
         }
     ]
@@ -142,10 +144,6 @@ pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py \
   --test-config-file tests/dfx/perf/tests/test_bagel_vllm_omni.json
 pytest -s -v tests/dfx/perf/scripts/run_benchmark.py \
   --test-config-file tests/dfx/perf/tests/test_qwen3_omni_async_chunk.json
-
-# Optional baseline assertion (default off)
-pytest -sv tests/dfx/perf/scripts/run_diffusion_benchmark.py --assert-baseline \
-  --test-config-file tests/dfx/perf/tests/test_qwen_image_vllm_omni.json
 ```
 
 See also [Markers for Tests](#markers-for-tests) for registered hardware markers (`H100`, `L4`, `cuda`, `distributed_cuda`, …).
@@ -202,4 +200,4 @@ You can add any benchmark running parameters you need here. For all optional par
 | num_prompts     | int / array | Yes      | 10,[10, 20, 30] | Number of requests. Supports single values or arrays. If a single value is used, it will be automatically expanded to match the number of qps or max_concurrency, e.g., [10,10,10]. If an array is used, its length must match the number of qps or max_concurrency. |
 | request_rate    | float / array | No  | 0.5, [0.5, 1, inf] | Queries per second. Supports single values or arrays. If a single value is used, it will be automatically expanded to match the number of num_prompts, e.g., [1,1,1]. If an array is used, its length must match the number of num_prompts.                          |
 | max_concurrency | int / array | No       | 1, [1, 2, 3]    | Maximum concurrent in-flight requests. Same array / expansion rules as `request_rate` (mutually exclusive with QPS mode).                                                                                                                                                                                                             |
-| baseline        | object      | No       | see above       | Optional per-metric thresholds; keys must match benchmark output fields. Scalar, list (per sweep step), or object (keyed by concurrency or QPS string).  
+| baseline        | object      | No       | see below       | Optional thresholds. **Hardware-nested only**: top-level keys must be `[hardware-resource]` pytest markers from `pyproject.toml` (`get_hardware_mark_list()`). Under each hardware key, metric names are free-form; values are scalars or sweep-aligned lists. Example: `{"H100": {"throughput_qps": […], "custom_stage_ms": 1.0}, "A3": {…}}`. Runtime filename aliases (`_RUNTIME_DEVICE_ALIASES`) are separate and do not expand this allowlist. |

docs/getting_started/installation/gpu.md

@@ -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, and 0.26.0. 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, and 0.25.0. 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).
 
 === "NVIDIA CUDA"
 

docs/getting_started/installation/gpu/cuda.inc.md

@@ -20,7 +20,7 @@ Therefore, it is recommended to install vLLM and vLLM-Omni with a **fresh new**
 
 vLLM-Omni is built based on vLLM. Please install it with command below.
 ```bash
-uv pip install vllm==0.26.0 --torch-backend=auto
+uv pip install vllm==0.25.0 --torch-backend=auto
 ```
 
 #### Installation of vLLM-Omni
@@ -39,13 +39,13 @@ uv pip install 'vllm-omni[demo]'
 # --8<-- [start:build-wheel-from-source]
 
 #### Installation of vLLM
-If you do not need to modify source code of vLLM, you can directly install the stable 0.26.0 release version of the library
+If you do not need to modify source code of vLLM, you can directly install the stable 0.25.0 release version of the library
 
 ```bash
-uv pip install vllm==0.26.0 --torch-backend=auto
+uv pip install vllm==0.25.0 --torch-backend=auto
 ```
 
-The 0.26.0 release of vLLM ships CUDA 13.0-compatible binaries by default. If you need a different CUDA variant or want to reuse an existing PyTorch installation, build vLLM from source instead.
+The 0.25.0 release of vLLM ships CUDA 13.0-compatible binaries by default. If you need a different CUDA variant or want to reuse an existing PyTorch installation, build vLLM from source instead.
 
 #### Installation of vLLM-Omni
 Since vllm-omni is rapidly evolving, it's recommended to install it from source
@@ -66,12 +66,12 @@ If you want to check, modify or debug with source code of vLLM, install the libr
 ```bash
 git clone https://github.com/vllm-project/vllm.git
 cd vllm
-git checkout v0.26.0
+git checkout v0.25.0
 ```
 Set up environment variables to get pre-built wheels. If there are internet problems, just download the whl file manually. And set `VLLM_PRECOMPILED_WHEEL_LOCATION` as your local absolute path of whl file.
 ```bash
-#For CUDA 13.0 (the default for v0.26.0; the wheel filename has no `+cu130` suffix)
-export VLLM_PRECOMPILED_WHEEL_LOCATION=https://github.com/vllm-project/vllm/releases/download/v0.26.0/vllm-0.26.0-cp38-abi3-manylinux_2_28_x86_64.whl
+#For CUDA 13.0 (the default for v0.25.0; the wheel filename has no `+cu130` suffix)
+export VLLM_PRECOMPILED_WHEEL_LOCATION=https://github.com/vllm-project/vllm/releases/download/v0.25.0/vllm-0.25.0-cp38-abi3-manylinux_2_28_x86_64.whl
 ```
 Install vllm with command below (If you have no existing PyTorch).
 ```bash
@@ -102,7 +102,7 @@ docker run --runtime nvidia --gpus 2 \
     --env "HF_TOKEN=$HF_TOKEN" \
     -p 8091:8091 \
     --ipc=host \
-    vllm/vllm-omni:v0.26.0 \
+    vllm/vllm-omni:v0.25.0 \
     vllm serve Qwen/Qwen3-Omni-30B-A3B-Instruct --omni --port 8091
 ```
 

docs/getting_started/installation/gpu/rocm.inc.md

@@ -13,7 +13,7 @@ vLLM-Omni current recommends the steps in under setup through Docker Images.
 
 vLLM-Omni is built based on vLLM. Please install it with command below.
 ```bash
-uv pip install vllm==0.26.0+rocm723 --extra-index-url https://wheels.vllm.ai/rocm/0.26.0/rocm723
+uv pip install vllm==0.25.0+rocm723 --extra-index-url https://wheels.vllm.ai/rocm/0.25.0/rocm723
 ```
 
 #### Installation of vLLM-Omni
@@ -34,13 +34,13 @@ uv pip install onnxruntime-rocm
 # --8<-- [start:build-wheel-from-source]
 
 #### Installation of vLLM
-If you do not need to modify source code of vLLM, you can directly install the stable 0.26.0 release version of the library
+If you do not need to modify source code of vLLM, you can directly install the stable 0.25.0 release version of the library
 
 ```bash
-uv pip install vllm==0.26.0+rocm723 --extra-index-url https://wheels.vllm.ai/rocm/0.26.0/rocm723
+uv pip install vllm==0.25.0+rocm723 --extra-index-url https://wheels.vllm.ai/rocm/0.25.0/rocm723
 ```
 
-The pre-built 0.26.0 vLLM wheel targets ROCm 7.2.3. If you need a different ROCm stack or want to reuse an existing PyTorch installation, build vLLM from source instead.
+The pre-built 0.25.0 vLLM wheel targets ROCm 7.2.3. If you need a different ROCm stack or want to reuse an existing PyTorch installation, build vLLM from source instead.
 
 #### Installation of vLLM-Omni
 Since vllm-omni is rapidly evolving, it's recommended to install it from source
@@ -58,7 +58,7 @@ If you want to check, modify or debug with source code of vLLM, install the libr
 ```bash
 git clone https://github.com/vllm-project/vllm.git
 cd vllm
-git checkout v0.26.0
+git checkout v0.25.0
 python3 -m pip install -r requirements/rocm.txt
 python3 setup.py develop
 ```
@@ -130,7 +130,7 @@ docker run --rm \
   -v ~/.cache/huggingface:/root/.cache/huggingface \
   --env "HF_TOKEN=$HF_TOKEN" \
   -p 8091:8091 \
-  vllm/vllm-omni-rocm:v0.26.0 \
+  vllm/vllm-omni-rocm:v0.25.0 \
   --model Qwen/Qwen3-Omni-30B-A3B-Instruct --omni --port 8091
 ```
 
@@ -149,7 +149,7 @@ docker run --rm -it \
   -v ~/.cache/huggingface:/root/.cache/huggingface \
   --env "HF_TOKEN=$HF_TOKEN" \
   --entrypoint bash \
-  vllm/vllm-omni-rocm:v0.26.0
+  vllm/vllm-omni-rocm:v0.25.0
 ```
 
 # --8<-- [end:pre-built-images]

docs/getting_started/quickstart.md

@@ -19,10 +19,10 @@ uv venv --python 3.12 --seed
 source .venv/bin/activate
 
 # On CUDA
-uv pip install vllm==0.26.0 --torch-backend=auto
+uv pip install vllm==0.25.0 --torch-backend=auto
 
 # On ROCm
-uv pip install vllm==0.26.0+rocm723 --extra-index-url https://wheels.vllm.ai/rocm/0.26.0/rocm723
+uv pip install vllm==0.25.0+rocm723 --extra-index-url https://wheels.vllm.ai/rocm/0.25.0/rocm723
 
 git clone https://github.com/vllm-project/vllm-omni.git
 cd vllm-omni
@@ -34,7 +34,7 @@ For additional installation methods — please see the [installation guide](inst
 !!! note
     It is important to install the same major & minor version of vLLM and vLLM Omni, otherwise things may not work as expected. If the versions are misaligned, you will see a warning when you import vLLM Omni.
 
-    If you are seeing strange behavior with the `vllm` command not handling the `--omni` flag correctly, you most likely have a version mismatch with vLLM < `0.26.0` and vLLM Omni `0.26.0`, as vLLM Omni no longer hijacks the vLLM entrypoint. Updating vLLM should resolve this issue.
+    If you are seeing strange behavior with the `vllm` command not handling the `--omni` flag correctly, you most likely have a version mismatch with vLLM < `0.25.0` and vLLM Omni `0.25.0`, as vLLM Omni no longer hijacks the vLLM entrypoint. Updating vLLM should resolve this issue.
 
 ## Offline Inference
 

docs/models/supported_models.md

@@ -41,8 +41,12 @@ th {
 | `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` | ✅︎ |   |   |   |
-| `LTX2Pipeline` | LTX-2 / LTX-2.3 one-stage T2V and I2V | `Lightricks/LTX-2`, `diffusers/LTX-2.3-Diffusers` | ✅︎ | ✅︎ | | |
-| `LTX2DistilledPipeline` | LTX-2 distilled two-stage T2V and I2V | `rootonchair/LTX-2-19b-distilled` | ✅︎ | ✅︎ | | |
+| `LTX2Pipeline` | LTX-2-T2V | `Lightricks/LTX-2` | ✅︎ | ✅︎ | | |
+| `LTX2ImageToVideoPipeline` | LTX-2-I2V | `Lightricks/LTX-2` | ✅︎ | ✅︎ | | |
+| `LTX2TwoStagesPipeline` | LTX-2-T2V | `rootonchair/LTX-2-19b-distilled` | ✅︎ | ✅︎ | | |
+| `LTX2ImageToVideoTwoStagesPipeline` | LTX-2-I2V | `rootonchair/LTX-2-19b-distilled` | ✅︎ | ✅︎ | | |
+| `LTX23Pipeline` | LTX-2.3-T2V | `diffusers/LTX-2.3-Diffusers` | ✅︎ | ✅︎ | | |
+| `LTX23ImageToVideoPipeline` | LTX-2.3-I2V | `diffusers/LTX-2.3-Diffusers` | ✅︎ | ✅︎ | | |
 | `LingBotVideoPipeline` | LingBot-Video dense and MoE T2V | `robbyant/lingbot-video-dense-1.3b`, `robbyant/lingbot-video-moe-30b-a3b` | ✅︎ | | | |
 | `DreamZeroPipeline` | DreamZero-DROID | `GEAR-Dreams/DreamZero-DROID` | ✅︎ | ✅︎ | ✅︎ | ✅︎ |
 | `HeliosPipeline`, `HeliosPyramidPipeline` | Helios | `BestWishYsh/Helios-Base`, `BestWishYsh/Helios-Mid`, `BestWishYsh/Helios-Distilled` | ✅︎ | ✅︎ | ✅︎ | |

docs/serving/videos_api.md

@@ -73,7 +73,7 @@ curl -L "http://localhost:8091/v1/videos/${video_id}/content" -o output.mp4
 | `audio_reference` | string | null | JSON-encoded audio reference for speech-to-video: `{"audio_url": "..."}` — supports HTTP(s) URLs or base64 data URLs |
 | `width` | integer | model default | Output video width |
 | `height` | integer | model default | Output video height |
-| `num_frames` | integer | 1 | Number of generated frames |
+| `num_frames` | integer | model default | Number of generated frames |
 | `fps` | integer | model default | Output frames per second |
 | `num_inference_steps` | integer | model default | Number of diffusion steps |
 | `guidance_scale` | number | null | CFG guidance scale for the low-noise stage |

docs/user_guide/examples/offline_inference/image_to_video.md

@@ -4,7 +4,7 @@ Source <https://github.com/vllm-project/vllm-omni/tree/main/examples/offline_inf
 
 
 This example demonstrates how to generate videos from images using Wan2.2,
-LTX-2, and HunyuanVideo-1.5 Image-to-Video models with vLLM-Omni's
+LTX-2/LTX-2.3, and HunyuanVideo-1.5 Image-to-Video models with vLLM-Omni's
 offline inference API.
 
 ## Local CLI Usage
@@ -53,22 +53,34 @@ python image_to_video.py \
   --output i2v_output.mp4
 ```
 
-### LTX-2
+### LTX-2.3
 
 ```bash
 python image_to_video.py \
-  --model Lightricks/LTX-2 \
+  --model diffusers/LTX-2.3-Diffusers \
+  --model-class-name LTX23ImageToVideoPipeline \
   --image cherry_blossom.jpg \
   --prompt "Cherry blossoms swaying gently in the breeze with synchronized ambient sound" \
-  --output ltx2_i2v_output.mp4
+  --negative-prompt "worst quality, inconsistent motion, blurry, jittery, distorted" \
+  --height 384 \
+  --width 512 \
+  --num-frames 25 \
+  --guidance-scale 4.0 \
+  --num-inference-steps 20 \
+  --frame-rate 24 \
+  --fps 24 \
+  --output ltx23_i2v_output.mp4
 ```
 
-See the [LTX-2 recipe](../../../../recipes/LTX/LTX-2.md) for all checkpoints,
-pipeline selection, T2V, defaults, and advanced options.
+Use the Diffusers-format checkpoint `diffusers/LTX-2.3-Diffusers`; the
+upstream `Lightricks/LTX-2.3` raw safetensors repo is not directly loadable by
+this pipeline. Pass `--model-class-name LTX23ImageToVideoPipeline` to select
+the LTX-2.3 image-to-video pipeline.
 
 Key arguments:
 
-- `--model`: Model ID (I2V-A14B for MoE, TI2V-5B for unified T2V+I2V, or LTX-2).
+- `--model`: Model ID (I2V-A14B for MoE, TI2V-5B for unified T2V+I2V, or
+  LTX-2/LTX-2.3).
 - `--image`: Path to input image (required).
 - `--prompt`: Text description of desired motion/animation.
 - `--height/--width`: Output resolution (auto-calculated from image if not set).

docs/user_guide/examples/offline_inference/text_to_video.md

@@ -12,7 +12,7 @@ For backend selection and SageAttention usage, see the [Diffusion Attention Back
 | Model | Default Resolution | Default Frames | Default Steps | Guidance | VRAM (BF16) |
 |---|---|---|---|---|---|
 | `Wan-AI/Wan2.2-T2V-A14B-Diffusers` | 720x1280 | 81 | 40 | 4.0 | ~60 GiB |
-| `Lightricks/LTX-2` | 512x768 | 121 | 40 | video 3.0 / audio 7.0 | Model-dependent |
+| `diffusers/LTX-2.3-Diffusers` | 384x512 | 25 | 20 | 4.0 | 96GB-class GPU |
 | `hunyuanvideo-community/HunyuanVideo-1.5-Diffusers-480p_t2v` | 480x832 | 121 | 50 | 6.0 | 1×A100 80GB |
 | `hunyuanvideo-community/HunyuanVideo-1.5-Diffusers-720p_t2v` | 720x1280 | 121 | 50 | 6.0 | FP8 + VAE tiling required |
 
@@ -35,17 +35,45 @@ python text_to_video.py \
   --output t2v_out.mp4
 ```
 
-### LTX-2
+LTX2 example:
 
 ```bash
 python text_to_video.py \
-  --model Lightricks/LTX-2 \
-  --prompt "Cherry blossoms swaying gently in the breeze with synchronized ambient sound" \
-  --output ltx2_output.mp4
+  --model "Lightricks/LTX-2" \
+  --prompt "A cinematic close-up of ocean waves at golden hour." \
+  --negative-prompt "worst quality, inconsistent motion, blurry, jittery, distorted" \
+  --height 512 \
+  --width 768 \
+  --num-frames 121 \
+  --num-inference-steps 40 \
+  --guidance-scale 4.0 \
+  --frame-rate 24 \
+  --output ltx2_out.mp4
 ```
 
-See the [LTX-2 recipe](../../../../recipes/LTX/LTX-2.md) for all checkpoints,
-pipeline selection, I2V, defaults, and advanced options.
+### LTX-2.3
+
+```bash
+python text_to_video.py \
+  --model diffusers/LTX-2.3-Diffusers \
+  --model-class-name LTX23Pipeline \
+  --prompt "Cherry blossoms swaying gently in the breeze with synchronized ambient sound" \
+  --negative-prompt "worst quality, inconsistent motion, blurry, jittery, distorted" \
+  --height 384 \
+  --width 512 \
+  --num-frames 25 \
+  --num-inference-steps 20 \
+  --guidance-scale 4.0 \
+  --frame-rate 24 \
+  --fps 24 \
+  --audio-sample-rate 48000 \
+  --output ltx23_t2v_output.mp4
+```
+
+Use the Diffusers-format checkpoint `diffusers/LTX-2.3-Diffusers`; the
+upstream `Lightricks/LTX-2.3` raw safetensors repo is not directly loadable by
+this pipeline. Pass `--model-class-name LTX23Pipeline` to select the LTX-2.3
+text-to-video pipeline explicitly.
 
 ### HunyuanVideo-1.5 (480p)
 
@@ -107,7 +135,8 @@ python text_to_video.py \
 ### Common
 
 - `--model`: Diffusers model ID or local path.
-- `--model-class-name`: Optional explicit pipeline override.
+- `--model-class-name`: Optional explicit pipeline class. Use `LTX23Pipeline`
+  for LTX-2.3 text-to-video.
 - `--prompt`: text description (string).
 - `--height/--width`: output resolution. Default depends on model.
 - `--num-frames`: number of frames. Default depends on model.
@@ -122,7 +151,8 @@ python text_to_video.py \
 - `--enable-cpu-offload`: enable CPU offloading for diffusion models.
 - `--enable-layerwise-offload`: enable layerwise offloading on DiT modules.
 - `--frame-rate`: generation FPS for pipelines that require it (e.g., LTX2).
-- `--audio-sample-rate`: fallback audio sample rate when the pipeline returns audio.
+- `--audio-sample-rate`: audio sample rate for embedded audio (when the
+  pipeline returns audio; LTX-2.3 outputs 48kHz audio).
 - `--quantization`: quantization method (`fp8` for FP8, `gguf` for GGUF).
 - `--flow-shift`: scheduler flow_shift parameter.
 

docs/user_guide/examples/online_serving/text_to_video.md

@@ -13,6 +13,7 @@ This example demonstrates how to deploy text-to-video models for online video ge
 | Wan2.1 T2V (14B) | `Wan-AI/Wan2.1-T2V-14B-Diffusers` |
 | Wan2.2 T2V | `Wan-AI/Wan2.2-T2V-A14B-Diffusers` |
 | LTX-2 | `Lightricks/LTX-2` |
+| LTX-2.3 | `diffusers/LTX-2.3-Diffusers` |
 
 ## Wan2.2 T2V
 
@@ -191,7 +192,7 @@ curl -X POST http://localhost:8091/v1/videos \
 | `negative_prompt`     | str    | None    | Negative prompt                                  |
 | `width`               | int    | None    | Video width in pixels                            |
 | `height`              | int    | None    | Video height in pixels                           |
-| `num_frames`          | int    | 1       | Number of frames to generate; set explicitly for video generation |
+| `num_frames`          | int    | None    | Number of frames to generate                     |
 | `fps`                 | int    | None    | Frames per second for output video               |
 | `num_inference_steps` | int    | None    | Number of denoising steps                        |
 | `guidance_scale`      | float  | None    | CFG guidance scale (low-noise stage)             |
@@ -289,12 +290,101 @@ done
 
 ## LTX-2
 
+### Start Server
+
+#### Basic Start
+
 ```bash
-vllm serve Lightricks/LTX-2 --omni --port 8098
+vllm serve Lightricks/LTX-2 --omni --port 8098 \
+    --enforce-eager --boundary-ratio 1.0
 ```
 
-See the [LTX-2 recipe](../../../../recipes/LTX/LTX-2.md) for all checkpoints,
-pipeline selection, requests, defaults, and advanced options.
+For multi-GPU memory reduction, you can enable HSDP:
+
+```bash
+vllm serve Lightricks/LTX-2 --omni --port 8098 \
+    --enforce-eager --boundary-ratio 1.0 \
+    --use-hsdp --hsdp-shard-size 2
+```
+
+#### Start with Optimization Presets
+
+Use the LTX-2 startup script with built-in optimization presets:
+
+```bash
+# Baseline (1 GPU, eager)
+bash run_server_ltx2.sh baseline
+
+# 4-GPU Ulysses sequence parallelism (lossless)
+bash run_server_ltx2.sh ulysses4
+
+# Cache-DiT lossy acceleration (1 GPU, ~1.4× speedup)
+bash run_server_ltx2.sh cache-dit
+
+# Best combo: 4-GPU Ulysses SP + Cache-DiT (~2.2× speedup)
+bash run_server_ltx2.sh best-combo
+```
+
+#### Optimization Benchmarks
+
+Benchmarked on H800, online serving (480×768, 41 frames, 20 steps, `seed=42`).
+"Inference" is the server-reported inference time; excludes HTTP/poll overhead.
+
+| Preset | Server Command | Inference (s) | Speedup | Type |
+|--------|---------------|---------------|---------|------|
+| `baseline` | `--enforce-eager` | 10.3 | 1.00× | — |
+| `compile` | *(default, no --enforce-eager)* | ~10.3 (warm) | ~1.00× | Lossless |
+| `ulysses4` | `--enforce-eager --usp 4` | ~10.3 | ~1.00× | Lossless |
+| `cache-dit` | `--enforce-eager --cache-backend cache_dit` | 7.4 avg | ~1.4× | Lossy |
+| `best-combo` | `--enforce-eager --usp 4 --cache-backend cache_dit` | 4.7 avg | **~2.2×** | Lossless + Lossy |
+
+**Observations**:
+- **torch.compile**: On H800, warm-request inference time matches the eager baseline (~10.3s).
+  The first request pays ~6s compilation overhead. Benefit depends on model architecture and GPU.
+- **Ulysses SP (4 GPU)**: No measurable speedup alone for 41-frame generation at this resolution.
+  Communication overhead outweighs gains at this sequence length.
+- **Cache-DiT**: Inference varies per request (6–10s) due to dynamic caching decisions.
+  Average is ~7.4s (~1.4× speedup) with slight quality tradeoff.
+- **Best combo**: 4-GPU Ulysses SP + Cache-DiT synergize well — Cache-DiT reduces per-step
+  computation, making the communication overhead of Ulysses SP worthwhile. Average ~4.7s
+  (~2.2× speedup).
+- **FP8 quantization**: Reduces VRAM but does not speed up LTX-2 on H800 (compute-bound).
+
+**Deployment Recommendations**:
+- For **production with quality priority**: use `baseline` with `--enforce-eager`
+- For **maximum throughput** (4 GPUs, quality tradeoff): use `best-combo` (~2.2× speedup)
+- For **single-GPU throughput**: use `cache-dit` (~1.4× speedup)
+- `--enforce-eager` is recommended to avoid torch.compile warmup latency on first request
+
+### Send Requests (curl)
+
+```bash
+# Using the provided script
+bash run_curl_ltx2.sh
+
+# Or directly
+curl -sS -X POST http://localhost:8098/v1/videos \
+  -H "Accept: application/json" \
+  -F "prompt=A serene lakeside sunrise with mist over the water." \
+  -F "width=768" \
+  -F "height=480" \
+  -F "num_frames=41" \
+  -F "fps=24" \
+  -F "num_inference_steps=20" \
+  -F "guidance_scale=3.0" \
+  -F "seed=42"
+```
+
+## LTX-2.3
+
+LTX-2.3 uses a Diffusers-format checkpoint and a dedicated pipeline class:
+
+```bash
+vllm serve diffusers/LTX-2.3-Diffusers --omni --port 8098 \
+    --model-class-name LTX23Pipeline
+```
+
+Send requests through the same `/v1/videos` API shown in the LTX-2 example.
 
 ## Example materials
 

examples/offline_inference/audex/README.md

@@ -1,114 +0,0 @@
-# Audex (Nemotron-Labs-Audex-2B) offline inference
-
-One checkpoint, four tasks — each script builds its own `Omni` engine with
-the right deploy yaml, so they run as plain `python <script>` with no extra
-flags. Pass the HF repo ROOT as `--model` (default
-`nvidia/Nemotron-Labs-Audex-2B`); per-stage subfolders and download subsets
-resolve automatically.
-
-| script | pipeline (`vllm_omni/deploy/<name>.yaml`) | audio in | text out | speech out | general audio out |
-|---|---|---|---|---|---|
-| `text_to_speech.py` | `audex_tts` | ❌ | ❌ | ✅ | ❌ |
-| `text_to_audio.py` | `audex_tta` | ❌ | ❌ | ❌ | ✅ |
-| `audio_qa.py` | `audex_thinker_only` | ✅ | ✅ | ❌ | ❌ |
-| `speech_to_speech.py` | `audex_s2s` | ✅ | ✅ | ✅ | ❌ |
-
-Shared plumbing (corpus/WAV IO, tokenizer loading, the CFG cond/uncond
-pair contract) lives in `common.py`; each task script keeps only its own
-prompt recipe and flow. Note `text_to_audio.py` here drives the Audex
-two-stage AR pipeline — for diffusion text-to-audio models (Stable Audio
-Open, AudioX) use the generic
-`examples/offline_inference/text_to_audio/text_to_audio.py` instead.
-
-## text_to_speech.py — text → speech
-
-Thinker generates `<speechcodec_N>` tokens; the streaming causal decoder
-turns them into 16 kHz WAVs (one file per prompt, slugified filename).
-
-```bash
-# three built-in demo sentences -> results/audex_wavs/
-python examples/offline_inference/audex/text_to_speech.py
-
-# your own texts, with classifier-free guidance (official quality setting)
-python examples/offline_inference/audex/text_to_speech.py \
-    --texts "Hello there." "Nice to meet you." --cfg-scale 1.5 \
-    --output-dir results/my_tts
-```
-
-Key flags: `--texts` / `--texts-file` (TSV `utt_id<TAB>text`),
-`--cfg-scale` (default 1.0 = off; 1.5 recommended — guided requests run
-one at a time, each paired with a length-matched null prompt),
-`--temperature` (overrides the deploy yaml's stage-0 default).
-
-## text_to_audio.py — caption → general audio (sound effects)
-
-Same thinker over the interleaved 4-codebook `<audiocodec_N>` RVQ block,
-decoded by the external XCodec1 checkpoint. CFG is effectively mandatory
-(default scale 3.0). Clips are capped at `--codec-cap 4000` codec tokens
-(10 s).
-
-```bash
-# three built-in demo captions -> results/audex_tta_wavs/
-python examples/offline_inference/audex/text_to_audio.py
-
-python examples/offline_inference/audex/text_to_audio.py \
-    --captions "Thunder rolling across a valley." --output-dir results/my_tta
-```
-
-XCodec1 resolves from `--xcodec1-path`, the `XCODEC1_PATH` env var, or the
-default `hf-audio/xcodec-hubert-general-balanced` repo (downloaded on
-first use).
-
-## audio_qa.py — audio (+ instruction) → text
-
-Single-stage audio understanding on the full checkpoint (NV-Whisper
-encoder + projector + LM). The default question is ASR
-("Transcribe the input speech."); without `--audio-files`, vLLM's public
-`mary_had_lamb` asset is transcribed.
-
-```bash
-# transcribe the built-in demo asset
-python examples/offline_inference/audex/audio_qa.py
-
-# your own clips / free-form QA
-python examples/offline_inference/audex/audio_qa.py \
-    --audio-files a.wav b.wav --question "What language is being spoken?"
-```
-
-## speech_to_speech.py — spoken question → spoken answer
-
-The official three-pass cascade over ONE `audex_s2s` deployment: ASR
-(audio → transcript, text-final) → chat (transcript → answer, text-final)
-→ TTS (answer → speech through the streaming decoder). Only the TTS pass
-touches the codec path. Without `--audio-file`, the `mary_had_lamb` asset
-is used as the spoken input.
-
-```bash
-# full cascade on the built-in demo asset -> results/audex_s2s_answer.wav
-python examples/offline_inference/audex/speech_to_speech.py
-
-python examples/offline_inference/audex/speech_to_speech.py \
-    --audio-file question.wav --output results/answer.wav --cfg-scale 1.5
-```
-
-`--cfg-scale` (default 1.5) applies to the TTS pass only; 1.0 disables it.
-
-Online-serving counterparts (server + HTTP clients) live in
-`examples/online_serving/audex/`.
-
-## 30B-A3B (nvidia/Nemotron-Labs-Audex-30B-A3B)
-
-The same four scripts serve the 30B MoE checkpoint (hybrid Mamba + MoE
-NemotronH thinker; identical token space and decoder). The 30B REQUIRES an
-explicit 30B deploy yaml — the model root's default resolution lands on the
-2B-tuned configs:
-
-    python examples/offline_inference/audex/text_to_speech.py \
-        --model nvidia/Nemotron-Labs-Audex-30B-A3B \
-        --deploy-config vllm_omni/deploy/audex_tts_30b.yaml
-
-Per-mode yamls: `audex_{tts,tta,thinker_only,s2s}_30b.yaml`. Defaults fit a
-single H100 80 GB (thinker + decoder share the card; prefix caching is off —
-hybrid Mamba); if long sequences OOM, set `tensor_parallel_size: 2` on
-stage 0. First run downloads ~60 GB. Verified on a single H100 80 GB: ~61 GiB
-weights + healthy KV headroom; TP2 was not needed.

examples/offline_inference/audex/audio_qa.py

@@ -1,93 +0,0 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
-"""Offline Audex (Nemotron-Labs-Audex-2B) audio understanding example.
-
-Speech (or general audio) + text instruction → text, through the
-single-stage ``audex_thinker_only`` pipeline: NV-Whisper
-encoder + relu2 projector + 2B dense LM on ``checkpoint_folder_full``.
-
-The prompt format mirrors the official audioqa script: a ChatML user turn
-holding the ``<so_embedding>`` placeholder (expanded by the processor to
-750 embedding positions per 30 s clip) plus the instruction, with a closed
-``<think></think>`` priming.
-
-Examples (without --audio-files, vLLM's public ``mary_had_lamb`` asset is
-transcribed):
-
-    # Transcribe WAVs (ASR):
-    python examples/offline_inference/audex/audio_qa.py \\
-        --audio-files a.wav b.wav
-
-    # Free-form audio QA:
-    python examples/offline_inference/audex/audio_qa.py \\
-        --audio-files a.wav --question "What language is being spoken?"
-"""
-
-from __future__ import annotations
-
-import argparse
-from pathlib import Path
-
-from common import default_deploy_config, load_audio_file, request_index
-from vllm.assets.audio import AudioAsset
-
-from vllm_omni import Omni
-
-ASR_QUESTION = "Transcribe the input speech."
-
-
-def build_prompt(question: str) -> str:
-    return f"<|im_start|>user\n<so_embedding>\n{question}<|im_end|>\n<|im_start|>assistant\n<think></think>"
-
-
-def parse_args():
-    parser = argparse.ArgumentParser(description="Offline Audex audio understanding")
-    parser.add_argument("--model", type=str, default="nvidia/Nemotron-Labs-Audex-2B")
-    parser.add_argument(
-        "--audio-files",
-        type=str,
-        nargs="+",
-        default=None,
-        help="Input clips. Defaults to vLLM's mary_had_lamb audio asset.",
-    )
-    parser.add_argument("--question", type=str, default=ASR_QUESTION)
-    parser.add_argument(
-        "--deploy-config",
-        type=str,
-        # The model root's default deploy yaml is the TTS pipeline; audio
-        # understanding needs the single-stage thinker-only pipeline.
-        default=default_deploy_config("audex_thinker_only.yaml"),
-        help="Deploy yaml (defaults to the audex_thinker_only pipeline).",
-    )
-    return parser.parse_args()
-
-
-def main():
-    args = parse_args()
-    engine = Omni(model=args.model, deploy_config=args.deploy_config, trust_remote_code=True)
-
-    prompt_text = build_prompt(args.question)
-    if args.audio_files:
-        clips = [load_audio_file(path) for path in args.audio_files]
-    else:
-        clips = [AudioAsset("mary_had_lamb").audio_and_sample_rate]
-    prompts = []
-    for audio, sr in clips:
-        prompts.append(
-            {
-                "prompt": prompt_text,
-                "multi_modal_data": {"audio": (audio, sr)},
-            }
-        )
-
-    outputs = engine.generate(prompts)
-
-    ordered = sorted(outputs, key=request_index)
-    names = [Path(p).name for p in args.audio_files] if args.audio_files else ["mary_had_lamb"]
-    for name, req_output in zip(names, ordered):
-        text = req_output.outputs[0].text if req_output.outputs else ""
-        print(f"{name}: {text.strip()}")
-
-
-if __name__ == "__main__":
-    main()

examples/offline_inference/audex/common.py

@@ -1,149 +0,0 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
-"""Shared helpers for the offline Audex examples in this directory.
-
-Import style follows the repo's example convention (same-directory module,
-see ``text_to_speech/qwen3_tts/tts_common.py``): scripts run as files, so
-``from common import ...`` resolves via the script's own directory.
-"""
-
-from __future__ import annotations
-
-import copy
-import re
-from collections.abc import Callable
-from pathlib import Path
-from typing import Any
-
-import numpy as np
-import soundfile as sf
-import torch
-
-from vllm_omni import Omni
-from vllm_omni.model_executor.models.audex.prompt import build_null_prompt
-
-SAMPLE_RATE = 16_000
-_DEPLOY_DIR = Path(__file__).resolve().parents[3] / "vllm_omni" / "deploy"
-
-# Guided decoding sharpens the distribution, so the unguided temperature
-# (0.1) adds excess sampling noise under CFG. Measured on the en-24 gate
-# corpus at cfg 1.5: temp 0.1 -> CER 7.31%, temp 0.05 -> CER 6.87% (vs the
-# unguided baseline 7.24%).
-GUIDED_TEMPERATURE = 0.05
-
-
-def default_deploy_config(name: str) -> str:
-    """Absolute path of a deploy yaml bundled with vllm_omni."""
-    return str(_DEPLOY_DIR / name)
-
-
-def slugify(text: str, fallback: str = "prompt") -> str:
-    slug = re.sub(r"\s+", "_", text.strip().lower())
-    slug = re.sub(r"[^a-z0-9_]+", "", slug)
-    return slug[:48] or fallback
-
-
-def load_corpus(
-    corpus_file: str | None,
-    texts: list[str] | None,
-    default_texts: tuple[str, ...],
-) -> list[tuple[str, str]]:
-    """(utt_id, text) pairs from a TSV corpus file, CLI texts, or defaults."""
-    if corpus_file:
-        corpus = []
-        for line in Path(corpus_file).read_text().splitlines():
-            if not line.strip():
-                continue
-            utt, text = line.split("\t", 1)
-            corpus.append((utt, text.strip()))
-        return corpus
-    resolved = texts if texts else list(default_texts)
-    return [(slugify(t), t) for t in resolved]
-
-
-def load_audio_file(path: str) -> tuple[np.ndarray, int]:
-    """Mono float32 samples + native sample rate (processors resample)."""
-    audio, sr = sf.read(path, dtype="float32")
-    if audio.ndim == 2:
-        audio = audio.mean(axis=1)
-    return audio, sr
-
-
-def extract_pcm(multimodal_output: dict) -> torch.Tensor:
-    """Flatten one request's audio payload to a 1-D float32 CPU tensor."""
-    audio = multimodal_output.get("model_outputs")
-    if audio is None:
-        audio = multimodal_output.get("audio")
-    if audio is None:
-        raise ValueError(f"no audio key in multimodal_output: {list(multimodal_output.keys())}")
-    if isinstance(audio, list):
-        valid = [torch.as_tensor(a).float().cpu().reshape(-1) for a in audio if a is not None]
-        if not valid:
-            raise ValueError("audio list is empty")
-        return torch.cat(valid, dim=0) if len(valid) > 1 else valid[0]
-    return torch.as_tensor(audio).float().cpu().reshape(-1)
-
-
-def write_wav(path: Path, pcm: torch.Tensor, sample_rate: int = SAMPLE_RATE) -> None:
-    path.parent.mkdir(parents=True, exist_ok=True)
-    arr = (np.clip(pcm.numpy(), -1.0, 1.0) * 32767.0).astype(np.int16)
-    sf.write(str(path), arr, sample_rate, format="WAV", subtype="PCM_16")
-
-
-def load_audex_tokenizer(model: str, profile: str = "tts", folder: str = "checkpoint_folder_audiogen"):
-    """Thinker tokenizer for CFG/TTA prompt building (resolves the snapshot)."""
-    from transformers import AutoTokenizer
-
-    from vllm_omni.model_executor.models.audex.checkpoint import ensure_audex_snapshot
-
-    root = ensure_audex_snapshot(model, profile=profile)
-    return AutoTokenizer.from_pretrained(str(Path(root) / folder), trust_remote_code=True)
-
-
-def request_index(req_output) -> int:
-    """Submission index from the numeric request-id prefix.
-
-    ``Omni.generate`` may return outputs out of submission order, and a
-    lexicographic sort would misplace "10_..." before "2_...".
-    """
-    match = re.search(r"(\d+)", str(req_output.request_id))
-    return int(match.group(1)) if match else 0
-
-
-def guided_stage_params(
-    engine: Omni,
-    cond_prompt: str,
-    tokenizer,
-    cfg_scale: float,
-    pair_id: str,
-    *,
-    temperature: float | None = None,
-    extra_args: dict[str, Any] | None = None,
-    null_prompt_builder: Callable[[str, Any], str] = build_null_prompt,
-) -> list:
-    """Stage sampling params carrying the CFG pair contract for ONE request.
-
-    Clones the engine's resolved defaults (never mutates the shared list),
-    applies ``extra_args`` unconditionally (e.g. the TTA RVQ contract), and
-    attaches the cond/uncond pair contract when ``cfg_scale > 1.0``. Guided
-    requests must go one at a time: each needs its own pair id and a
-    length-matched null prompt built from its own text.
-    """
-    params = copy.deepcopy(engine.resolve_sampling_params_list(None))
-    stage0 = params[0]
-    if temperature is not None:
-        stage0.temperature = temperature
-    if stage0.extra_args is None:
-        stage0.extra_args = {}
-    if extra_args:
-        stage0.extra_args.update(extra_args)
-    if cfg_scale > 1.0:
-        stage0.extra_args.update(
-            {
-                "cfg_scale": float(cfg_scale),
-                "cfg_role": "cond",
-                "cfg_pair_id": pair_id,
-                "cfg_null_prompt": null_prompt_builder(cond_prompt, tokenizer),
-            }
-        )
-    return params

examples/offline_inference/audex/speech_to_speech.py

@@ -1,277 +0,0 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
-"""Offline Audex (Nemotron-Labs-Audex-2B) cascaded speech-to-speech example.
-
-The official three-pass cascade over ONE ``audex_s2s``
-deployment:
-
-  1. ASR pass  (audio + "Transcribe the input speech." → transcript, text)
-  2. chat pass (transcript → answer, text)
-  3. TTS pass  (answer → speech, audio via the streaming causal decoder)
-
-Text passes carry ``modalities: ["text"]`` and finish at stage 0; only the
-TTS pass (``modalities: ["audio"]``) streams codec frames into code2wav.
-
-Example (without --audio-file, vLLM's public ``mary_had_lamb`` asset is
-used as the spoken input):
-
-    python examples/offline_inference/audex/speech_to_speech.py \\
-        --audio-file question.wav --output results/answer.wav
-"""
-
-from __future__ import annotations
-
-import argparse
-import copy
-import re
-from pathlib import Path
-
-import torch
-from common import (
-    SAMPLE_RATE,
-    default_deploy_config,
-    guided_stage_params,
-    load_audex_tokenizer,
-    load_audio_file,
-    write_wav,
-)
-from vllm.assets.audio import AudioAsset
-
-from vllm_omni import Omni
-from vllm_omni.model_executor.models.audex.prompt import build_cond_prompt
-
-ASR_QUESTION = "Transcribe the input speech."
-SYSTEM_PROMPT = "You are a helpful and harmless assistant.\n\nYou are not allowed to use any tools."
-# The official chat-pass recipe below (SYSTEM_PROMPT + TEXT_INSTRUCTION
-# prefix, open <think> priming, temp 1.0/top_p 0.95, seed retry) applies to
-# the 30B-A3B ONLY (see is_30b()/_chat_prompt below and
-# tests/e2e/offline_inference/test_audex_s2s.py, which mirrors this same
-# split). It is load-bearing on the 30B: without the formatting-instruction
-# prefix a bare transcript turn is treated as a TTS request and the model
-# emits literal <speechcodec_N> tokens instead of a text answer (the system
-# prompt alone does not prevent this). The 2B must NOT use this recipe: it
-# is long-verified stable on a bare prompt + engine-default sampling, while
-# the official temp-1.0 seed-0 draw hallucinates off-language text (yet
-# closes </think> cleanly, so the seed retry never fires) and fails the
-# round-trip WER gate — measured, reproduced byte-identically.
-TEXT_INSTRUCTION = (
-    "SYSTEM & FORMATTING INSTRUCTION: You are Audex, created by NVIDIA based on the Nemotron-Cascade-2 architecture. "
-    "You may output your reasoning, followed by your final response. "
-    "You may format your reasoning block however you like. "
-    "However, your final response must strictly follow these rules: "
-    "* Write in plain, unformatted prose like a book or newspaper article. "
-    "* Do not use markdown, bullet points, lists, or headers. "
-    "* Standard numbers, abbreviations, and symbols are acceptable. "
-    "* [CRITICAL] You must press enter after every single sentence, placing each sentence on its own separate line."
-)
-
-
-def is_30b(model: str) -> bool:
-    # Users pass arbitrary repo IDs or local snapshot paths, so this is a
-    # heuristic (not an exact match against the official 30B-A3B repo ID)
-    # — good enough for an example script; keep in sync with the same
-    # branch in examples/online_serving/audex/client.py and
-    # tests/e2e/offline_inference/test_audex_s2s.py.
-    return "30b" in model.lower()
-
-
-def parse_args():
-    parser = argparse.ArgumentParser(description="Offline Audex cascaded S2S")
-    parser.add_argument("--model", type=str, default="nvidia/Nemotron-Labs-Audex-2B")
-    parser.add_argument(
-        "--audio-file",
-        type=str,
-        default=None,
-        help="Spoken question (WAV). Defaults to vLLM's mary_had_lamb audio asset.",
-    )
-    parser.add_argument("--output", type=str, default="results/audex_s2s_answer.wav")
-    parser.add_argument(
-        "--deploy-config",
-        type=str,
-        # The model root's default deploy yaml is the TTS-only pipeline; the
-        # cascade needs the audio-capable 2-stage full pipeline.
-        default=default_deploy_config("audex_s2s.yaml"),
-        help="Deploy yaml (defaults to the audex_s2s pipeline).",
-    )
-    parser.add_argument(
-        "--cfg-scale",
-        type=float,
-        default=1.5,
-        help="CFG strength for the TTS pass (1.0 disables; official setting 1.5).",
-    )
-    return parser.parse_args()
-
-
-def _chat_prompt(user_text: str) -> str:
-    # Official chat-pass recipe: system prompt + formatting instruction
-    # prefixed to the user text, thinking ENABLED via an open <think> block
-    # (the official demo's web UI ships "Enable reasoning" checked, i.e.
-    # build_chat_prompt(..., enable_thinking=True); chat_template.jinja then
-    # renders "<|im_start|>assistant\n<think>\n"). The instruction invites
-    # the model to reason before answering — priming the block open keeps
-    # that reasoning inside <think>...</think> so _extract_final_response
-    # below can drop it; with a closed block the 30B leaks the reasoning as
-    # plain text in front of the answer.
-    return (
-        f"<|im_start|>system\n{SYSTEM_PROMPT}<|im_end|>\n"
-        f"<|im_start|>user\n{TEXT_INSTRUCTION}\n\n{user_text}<|im_end|>\n"
-        "<|im_start|>assistant\n<think>\n"
-    )
-
-
-def _asr_prompt() -> str:
-    return f"<|im_start|>user\n<so_embedding>\n{ASR_QUESTION}<|im_end|>\n<|im_start|>assistant\n<think></think>"
-
-
-def _extract_final_response(text: str) -> str:
-    # Mirror the official demo's extract_spoken_answer
-    # (cascaded_s2s_web_server.py): keep only what follows the LAST closing
-    # think tag, falling back to the full text when the model skipped
-    # reasoning or never closed the block. Benign no-op for the 2B, whose
-    # answers carry no think block.
-    if "</think>" in text:
-        answer = text.rsplit("</think>", 1)[-1].strip()
-        if answer:
-            return answer
-    return text.strip()
-
-
-def _clean(text: str) -> str:
-    # The full checkpoint prefixes transcripts with a language tag sentence
-    # and quotes the content; extract the quoted payload when present. Span
-    # first-to-last quote (not split on every apostrophe) so contractions
-    # inside the transcript (e.g. "don't") survive intact.
-    text = text.strip()
-    match = re.search(r"'(.*)'", text, flags=re.S)
-    if match and match.group(1).strip():
-        return match.group(1).strip()
-    return text
-
-
-def _text_of(outputs) -> str:
-    (req_output,) = outputs
-    return (req_output.outputs[0].text or "").strip()
-
-
-CHAT_MAX_TOKENS = 2048  # official --text-max-tokens default
-# The official demo samples this pass unseeded; some trajectories deliberate
-# past the token budget without ever closing the think block, so retry a few
-# deterministic seeds until one closes (see the chat pass below).
-CHAT_SEEDS = (0, 1, 2)
-
-
-def _chat_params(engine: Omni, seed: int):
-    # Official demo text-pass sampling (cascaded_s2s_web_server.py defaults:
-    # --text-temperature 1.0, --text-top-p 0.95, --text-max-tokens 2048).
-    # The yaml stage-0 defaults (temperature 0.1, top_k 80) are TTS-thinker
-    # settings; under near-greedy decoding the 30B loops inside its
-    # reasoning block until the token budget runs out and never emits
-    # </think>, so the extracted "answer" would be truncated reasoning.
-    params = copy.deepcopy(engine.resolve_sampling_params_list(None))
-    stage0 = params[0]
-    stage0.temperature = 1.0
-    stage0.top_p = 0.95
-    stage0.top_k = -1
-    stage0.max_tokens = CHAT_MAX_TOKENS
-    stage0.seed = seed
-    return params
-
-
-def _chat_once(engine: Omni, prompt: str, seed: int) -> tuple[str, bool]:
-    (req_output,) = engine.generate([{"prompt": prompt, "modalities": ["text"]}], _chat_params(engine, seed))
-    out = req_output.outputs[0]
-    text = (out.text or "").strip()
-    truncated = getattr(out, "finish_reason", None) == "length" or len(out.token_ids or []) >= CHAT_MAX_TOKENS
-    return text, truncated
-
-
-def main():
-    args = parse_args()
-    engine = Omni(model=args.model, deploy_config=args.deploy_config, trust_remote_code=True)
-
-    tokenizer = None
-    if args.cfg_scale > 1.0:
-        tokenizer = load_audex_tokenizer(args.model, profile="full", folder="checkpoint_folder_full")
-
-    if args.audio_file:
-        audio, sr = load_audio_file(args.audio_file)
-    else:
-        audio, sr = AudioAsset("mary_had_lamb").audio_and_sample_rate
-
-    # Pass 1 — ASR (text-final; never touches the speech decoder).
-    transcript = _clean(
-        _text_of(
-            engine.generate(
-                [
-                    {
-                        "prompt": _asr_prompt(),
-                        "multi_modal_data": {"audio": (audio, sr)},
-                        "modalities": ["text"],
-                    }
-                ]
-            )
-        )
-    )
-    print(f"[1/3] transcript : {transcript}")
-    if not transcript:
-        raise SystemExit("ASR pass produced an empty transcript; aborting cascade")
-
-    # Pass 2 — chat answer (text-final). The recipe is size-conditional (see
-    # the TEXT_INSTRUCTION comment above): the 30B requires the official
-    # recipe, the 2B regresses under it.
-    if is_30b(args.model):
-        # Official recipe. The raw completion is
-        # <reasoning...></think><answer>; speak only the final response.
-        # Retry with the next seed only when the budget ran out before the
-        # reasoning block closed (a natural stop without </think> is a
-        # complete answer); if every seed spins, fall back to the official
-        # demo's behavior of keeping the full text.
-        raw = ""
-        for seed in CHAT_SEEDS:
-            raw, truncated = _chat_once(engine, _chat_prompt(transcript), seed)
-            if "</think>" in raw or not truncated:
-                break
-            print(f"[2/3] chat seed {seed} hit the token budget mid-reasoning; retrying")
-        answer = _extract_final_response(raw)
-    else:
-        # 2B: bare prompt with a closed think block, engine-default sampling
-        # (yaml near-greedy), no retry; the answer is used as-is.
-        chat_prompt = f"<|im_start|>user\n{transcript}<|im_end|>\n<|im_start|>assistant\n<think></think>"
-        answer = _text_of(engine.generate([{"prompt": chat_prompt, "modalities": ["text"]}]))
-    print(f"[2/3] answer     : {answer}")
-    if not answer:
-        raise SystemExit("Chat pass produced an empty answer; aborting cascade")
-
-    # Pass 3 — TTS (audio-final; streams through code2wav). Both final
-    # stages of the pipeline may emit an output record for the request;
-    # collect the audio-bearing one.
-    # The formatting instruction puts every sentence on its own line;
-    # flatten to plain prose for the TTS thinker.
-    tts_prompt = build_cond_prompt(re.sub(r"\s+", " ", answer).strip())
-    outputs = engine.generate(
-        [{"prompt": tts_prompt, "modalities": ["audio"]}],
-        guided_stage_params(engine, tts_prompt, tokenizer, args.cfg_scale, "s2s-tts-pass"),
-    )
-    chunks: list[torch.Tensor] = []
-    for req_output in outputs:
-        mm = getattr(req_output, "multimodal_output", None) or {}
-        if not mm and req_output.outputs:
-            mm = getattr(req_output.outputs[0], "multimodal_output", None) or {}
-        audio_val = mm.get("model_outputs", mm.get("audio"))
-        if audio_val is None:
-            continue
-        if isinstance(audio_val, list):
-            chunks.extend(torch.as_tensor(a).float().cpu().reshape(-1) for a in audio_val if a is not None)
-        else:
-            chunks.append(torch.as_tensor(audio_val).float().cpu().reshape(-1))
-    pcm = torch.cat(chunks) if chunks else torch.empty(0)
-    if pcm.numel() == 0:
-        raise SystemExit("TTS pass produced empty audio")
-
-    out_path = Path(args.output)
-    write_wav(out_path, pcm)
-    print(f"[3/3] speech     : {pcm.numel() / SAMPLE_RATE:.2f}s -> {out_path}")
-
-
-if __name__ == "__main__":
-    main()

examples/offline_inference/audex/text_to_audio.py

@@ -1,157 +0,0 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
-"""Offline Audex (Nemotron-Labs-Audex-2B) text-to-audio inference example.
-
-Caption → general audio through the vLLM-Omni TTA pipeline: the audiogen
-thinker generates interleaved 4-codebook <audiocodec_N> RVQ tokens under an
-RVQ phase mask, and the external XCodec1 checkpoint decodes them to a 16 kHz
-mono WAV.
-
-(For DIFFUSION text-to-audio models — Stable Audio Open, AudioX — use
-``examples/offline_inference/text_to_audio/text_to_audio.py`` instead; Audex
-TTA is a two-stage autoregressive pipeline with a different interface.)
-
-Classifier-free guidance is effectively mandatory for TTA quality (official
-default scale 3.0), so every request submits a cond/uncond pair; requests
-run one at a time.
-
-Pass the HF repo ROOT as --model. XCodec1 resolves from --xcodec1-path /
-XCODEC1_PATH / the default hf-audio repo.
-
-Example:
-
-    python examples/offline_inference/audex/text_to_audio.py \\
-        --captions "Heavy rain falling on a tin roof." \\
-        --output-dir results/audex_tta_wavs
-"""
-
-from __future__ import annotations
-
-import argparse
-import os
-import time
-from pathlib import Path
-
-from common import (
-    SAMPLE_RATE,
-    default_deploy_config,
-    extract_pcm,
-    guided_stage_params,
-    load_audex_tokenizer,
-    load_corpus,
-    write_wav,
-)
-
-from vllm_omni import Omni
-from vllm_omni.model_executor.models.audex.prompt import build_tta_cond_prompt, build_tta_null_prompt
-from vllm_omni.model_executor.models.audex.tta import build_tta_phase_token_ids
-
-DEFAULT_CAPTIONS = (
-    "Heavy rain falling on a tin roof.",
-    "A dog barking in the distance while birds chirp.",
-    "Ocean waves crashing on a rocky shore.",
-)
-DEFAULT_CODEC_CAP = 4000
-
-
-def parse_args():
-    parser = argparse.ArgumentParser(description="Offline Audex TTA inference")
-    parser.add_argument("--model", type=str, default="nvidia/Nemotron-Labs-Audex-2B")
-    parser.add_argument("--captions", type=str, nargs="+", default=None, help="Audio captions.")
-    parser.add_argument(
-        "--captions-file",
-        type=str,
-        default=None,
-        help="TSV corpus: one 'utt_id<TAB>caption' per line (overrides --captions).",
-    )
-    parser.add_argument("--output-dir", type=str, default="results/audex_tta_wavs")
-    parser.add_argument(
-        "--deploy-config",
-        type=str,
-        # The model root's default deploy yaml is the TTS pipeline; TTA
-        # prompts and RVQ sampling params require the dedicated TTA pipeline.
-        default=default_deploy_config("audex_tta.yaml"),
-        help="Deploy yaml (defaults to the audex_tta pipeline).",
-    )
-    parser.add_argument(
-        "--xcodec1-path",
-        type=str,
-        default=None,
-        help="Local XCodec1 checkpoint (defaults to $XCODEC1_PATH or the hf-audio repo).",
-    )
-    parser.add_argument(
-        "--cfg-scale",
-        type=float,
-        default=3.0,
-        help="Classifier-free guidance strength (official TTA default 3.0; 1.0 disables).",
-    )
-    parser.add_argument(
-        "--codec-cap",
-        type=int,
-        default=DEFAULT_CODEC_CAP,
-        help="Max generated codec tokens before the phase mask forces <audiogen_end>.",
-    )
-    return parser.parse_args()
-
-
-def main():
-    args = parse_args()
-    if args.xcodec1_path:
-        os.environ["XCODEC1_PATH"] = args.xcodec1_path
-    output_dir = Path(args.output_dir)
-    output_dir.mkdir(parents=True, exist_ok=True)
-    corpus = load_corpus(args.captions_file, args.captions, DEFAULT_CAPTIONS)
-
-    tokenizer = load_audex_tokenizer(args.model, profile="tta")
-    phase_token_ids, start_tid, end_tid = build_tta_phase_token_ids(tokenizer)
-    tta_rvq = {
-        "phase_token_ids": phase_token_ids,
-        "start_tid": start_tid,
-        "end_tid": end_tid,
-        "codec_cap": args.codec_cap,
-        # The prompt ends with <audiogen_start>, so generation begins at
-        # RVQ phase 0 immediately.
-        "start_in_prompt": True,
-    }
-
-    engine = Omni(model=args.model, deploy_config=args.deploy_config, trust_remote_code=True)
-
-    print(f"Model       : {args.model}")
-    print(f"Captions    : {len(corpus)}")
-    print(f"CFG scale   : {args.cfg_scale}")
-    print(f"Codec cap   : {args.codec_cap}")
-    print(f"Output dir  : {output_dir}")
-
-    total_elapsed = 0.0
-    total_dur = 0.0
-    for utt, caption in corpus:
-        cond_prompt = build_tta_cond_prompt(caption)
-        sampling = guided_stage_params(
-            engine,
-            cond_prompt,
-            tokenizer,
-            args.cfg_scale,
-            f"tta-{utt}",
-            extra_args={"tta_rvq": tta_rvq},
-            null_prompt_builder=build_tta_null_prompt,
-        )
-        t_start = time.perf_counter()
-        outputs = engine.generate([cond_prompt], sampling)
-        total_elapsed += time.perf_counter() - t_start
-
-        (req_output,) = outputs
-        pcm = extract_pcm(req_output.outputs[0].multimodal_output)
-        dur = pcm.numel() / SAMPLE_RATE
-        if dur <= 0:
-            raise RuntimeError(f"empty audio for {utt}")
-        out_path = output_dir / f"{utt}.wav"
-        write_wav(out_path, pcm)
-        total_dur += dur
-        print(f"  {utt:<48} dur={dur:6.2f}s  -> {out_path}")
-
-    rtf = total_elapsed / total_dur if total_dur > 0 else float("inf")
-    print(f"Total infer : {total_elapsed:.2f}s  total audio: {total_dur:.2f}s  RTF: {rtf:.3f}")
-
-
-if __name__ == "__main__":
-    main()

examples/offline_inference/audex/text_to_speech.py

@@ -1,162 +0,0 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
-"""Offline Audex (Nemotron-Labs-Audex-2B) TTS inference example.
-
-Runs Stage 0 (thinker) + Stage 1 (streaming causal speech decoder) end-to-end
-through the vLLM-Omni engine and saves a 16 kHz mono WAV per prompt.
-
-Pass the HF repo ROOT as --model; per-stage subfolders resolve automatically.
-
-Examples:
-
-    python examples/offline_inference/audex/text_to_speech.py \\
-        --texts "The weather is so good today." \\
-        --output-dir results/audex_wavs
-
-    # TSV corpus (utt_id<TAB>text per line), all requests in flight at once:
-    python examples/offline_inference/audex/text_to_speech.py \\
-        --texts-file corpus.tsv --output-dir results/audex_wavs
-
-    # One request at a time:
-    python examples/offline_inference/audex/text_to_speech.py \\
-        --texts-file corpus.tsv --sequential --output-dir results/audex_wavs
-"""
-
-from __future__ import annotations
-
-import argparse
-import time
-from pathlib import Path
-
-from common import (
-    GUIDED_TEMPERATURE,
-    SAMPLE_RATE,
-    extract_pcm,
-    guided_stage_params,
-    load_audex_tokenizer,
-    load_corpus,
-    request_index,
-    write_wav,
-)
-
-from vllm_omni import Omni
-from vllm_omni.model_executor.models.audex.prompt import build_cond_prompt
-
-DEFAULT_TEXTS = (
-    "Hello world.",
-    "The quick brown fox jumps over the lazy dog.",
-    "Today is a beautiful day for a walk in the park.",
-)
-
-
-def parse_args():
-    parser = argparse.ArgumentParser(description="Offline Audex TTS inference")
-    parser.add_argument(
-        "--model",
-        type=str,
-        default="nvidia/Nemotron-Labs-Audex-2B",
-        help="HF repo root (or local snapshot root).",
-    )
-    parser.add_argument("--texts", type=str, nargs="+", default=None, help="Plain-text prompts.")
-    parser.add_argument(
-        "--texts-file",
-        type=str,
-        default=None,
-        help="TSV corpus: one 'utt_id<TAB>text' per line (overrides --texts).",
-    )
-    parser.add_argument("--output-dir", type=str, default="results/audex_wavs")
-    parser.add_argument("--deploy-config", type=str, default=None, help="Override the deploy config path.")
-    parser.add_argument(
-        "--sequential",
-        action="store_true",
-        help="Submit one request at a time instead of all at once.",
-    )
-    parser.add_argument(
-        "--cfg-scale",
-        type=float,
-        default=1.0,
-        help=(
-            "Classifier-free guidance strength. 1.0 disables guidance (default, "
-            "matches the v1 baseline); 1.5 is the official quality setting. "
-            "Guided runs submit one request at a time (each carries its own "
-            "length-matched null prompt and pair id)."
-        ),
-    )
-    parser.add_argument(
-        "--temperature",
-        type=float,
-        default=None,
-        help="Override the deploy yaml's stage-0 sampling temperature.",
-    )
-    return parser.parse_args()
-
-
-def _write_outputs(outputs, batch: list[tuple[str, str]], output_dir: Path) -> float:
-    ordered = sorted(outputs, key=request_index)
-    if len(ordered) != len(batch):
-        raise RuntimeError(f"expected {len(batch)} outputs, got {len(ordered)}")
-
-    total_dur = 0.0
-    for (utt, _text), req_output in zip(batch, ordered):
-        pcm = extract_pcm(req_output.outputs[0].multimodal_output)
-        dur = pcm.numel() / SAMPLE_RATE
-        if dur <= 0:
-            raise RuntimeError(f"empty audio for {utt}")
-        out_path = output_dir / f"{utt}.wav"
-        write_wav(out_path, pcm)
-        total_dur += dur
-        print(f"  {utt:<40} dur={dur:6.2f}s  -> {out_path}")
-    return total_dur
-
-
-def main():
-    args = parse_args()
-    output_dir = Path(args.output_dir)
-    output_dir.mkdir(parents=True, exist_ok=True)
-    corpus = load_corpus(args.texts_file, args.texts, DEFAULT_TEXTS)
-
-    engine = Omni(model=args.model, deploy_config=args.deploy_config, trust_remote_code=True)
-
-    cfg_enabled = args.cfg_scale > 1.0
-    tokenizer = load_audex_tokenizer(args.model) if cfg_enabled else None
-
-    print(f"Model       : {args.model}")
-    print(f"Prompts     : {len(corpus)}")
-    print(f"Mode        : {'sequential' if args.sequential or cfg_enabled else 'batched'}")
-    print(f"CFG scale   : {args.cfg_scale}" + (" (guidance off)" if not cfg_enabled else ""))
-    print(f"Output dir  : {output_dir}")
-
-    total_elapsed = 0.0
-    total_dur = 0.0
-    # CFG runs one request at a time: each request needs its own pair id and
-    # text-specific null prompt in the stage-0 sampling params.
-    batches = [[item] for item in corpus] if (args.sequential or cfg_enabled) else [corpus]
-    for batch in batches:
-        prompts = [build_cond_prompt(text) for _utt, text in batch]
-        sampling_params_list = None
-        if cfg_enabled:
-            sampling_params_list = guided_stage_params(
-                engine,
-                prompts[0],
-                tokenizer,
-                args.cfg_scale,
-                f"cfg-{batch[0][0]}",
-                temperature=GUIDED_TEMPERATURE,
-            )
-        if args.temperature is not None:
-            import copy
-
-            if sampling_params_list is None:
-                sampling_params_list = copy.deepcopy(engine.resolve_sampling_params_list(None))
-            sampling_params_list[0].temperature = args.temperature
-        t_start = time.perf_counter()
-        outputs = engine.generate(prompts, sampling_params_list)
-        total_elapsed += time.perf_counter() - t_start
-        total_dur += _write_outputs(outputs, batch, output_dir)
-
-    rtf = total_elapsed / total_dur if total_dur > 0 else float("inf")
-    print(f"Total infer : {total_elapsed:.2f}s  total audio: {total_dur:.2f}s  RTF: {rtf:.3f}")
-
-
-if __name__ == "__main__":
-    main()

examples/offline_inference/image_to_video/README.md

@@ -1,7 +1,7 @@
 # Image-To-Video
 
-This shared example generates videos from images with VACE, Wan2.2, LTX-2,
-HunyuanVideo-1.5, Cosmos3, and other compatible pipelines.
+This shared example generates videos from images with VACE, Wan2.2,
+LTX-2/LTX-2.3, HunyuanVideo-1.5, Cosmos3, and other compatible pipelines.
 
 - `image_to_video.py`: command-line script for single video generation with advanced options.
 
@@ -26,7 +26,7 @@ This folder provides a unified CLI script for image-to-video generation using vL
 | `Wan-AI/Wan2.2-I2V-A14B-Diffusers` | 480 x 832 | 81 | 50 | 5.0 | Around 60 GiB BF16 for basic single-card usage |
 | `Wan-AI/Wan2.2-TI2V-5B-Diffusers` | 480 x 832 | 81 | 50 | 4.0 | Around 20–25 GiB BF16, smallest I2V model |
 | `hunyuanvideo-community/HunyuanVideo-1.5-Diffusers-480p_i2v` | 480 x 832 | 121 | 50 | 6.0 | Around 100 GiB at default settings; the example enables `--enable-cpu-offload` + VAE tiling/slicing to fit an 80 GiB card |
-| `Lightricks/LTX-2` | 512 x 768 | 121 | 40 | video 3.0 / audio 7.0 | Memory use depends on frame count and tensor parallelism |
+| LTX2 (local path + `--model-class-name LTX2ImageToVideoPipeline`) | 512 x 768 | 121 | 40 | 4.0 | Memory use depends on frame count and tensor parallelism |
 
 !!! info
     Peak VRAM: based on basic single-card usage, batch size = 1, without any acceleration/optimization features. Some model weights cannot fit into one card with 80 GiB VRAM, which may need to use CPU offloading.
@@ -105,7 +105,7 @@ python image_to_video.py \
 | Argument | Type | Default | Description |
 | -------- | ---- | ------- | ----------- |
 | `--model` | str | `Wan-AI/Wan2.2-I2V-A14B-Diffusers` | Diffusers I2V model ID or local path |
-| `--model-class-name` | str | `None` | Optional pipeline override |
+| `--model-class-name` | str | `None` | Override model class name (e.g., `LTX2ImageToVideoPipeline`) |
 | `--image` | str | (required) | Path to input image |
 | `--prompt` | str | `""` | Text description of desired motion/animation |
 | `--negative-prompt` | str | `""` | Optional list of artifacts to suppress |
@@ -195,19 +195,24 @@ python image_to_video.py \
   --output hunyuan_i2v.mp4
 ```
 
-### LTX-2
+### LTX2 Image-to-Video
 
 ```bash
 python image_to_video.py \
-  --model Lightricks/LTX-2 \
+  --model /path/to/LTX-2 \
+  --model-class-name LTX2ImageToVideoPipeline \
   --image cherry_blossom.jpg \
   --prompt "A cinematic dolly shot of cherry blossoms" \
+  --height 512 \
+  --width 768 \
+  --num-frames 121 \
+  --num-inference-steps 40 \
+  --guidance-scale 4.0 \
+  --frame-rate 24 \
+  --fps 24 \
   --output ltx2_i2v.mp4
 ```
 
-See the [LTX-2 recipe](../../../recipes/LTX/LTX-2.md) for all checkpoints,
-pipeline selection, T2V, defaults, and advanced options.
-
 ## Advanced Features
 
 ### CFG Parallel
@@ -228,6 +233,30 @@ python image_to_video.py \
   --output i2v_cached.mp4
 ```
 
+### LTX-2.3
+
+```bash
+python image_to_video.py \
+  --model diffusers/LTX-2.3-Diffusers \
+  --model-class-name LTX23ImageToVideoPipeline \
+  --image cherry_blossom.jpg \
+  --prompt "Cherry blossoms swaying gently in the breeze with synchronized ambient sound" \
+  --negative-prompt "worst quality, inconsistent motion, blurry, jittery, distorted" \
+  --height 384 \
+  --width 512 \
+  --num-frames 25 \
+  --guidance-scale 4.0 \
+  --num-inference-steps 20 \
+  --frame-rate 24 \
+  --fps 24 \
+  --output ltx23_i2v_output.mp4
+```
+
+Use the Diffusers-format checkpoint `diffusers/LTX-2.3-Diffusers`; the
+upstream `Lightricks/LTX-2.3` raw safetensors repo is not directly loadable by
+this pipeline. Pass `--model-class-name LTX23ImageToVideoPipeline` to select
+the LTX-2.3 image-to-video pipeline.
+
 ### Cosmos3
 
 ```bash
@@ -245,8 +274,8 @@ python image_to_video.py \
 
 Key arguments:
 
-- `--model`: Model ID (I2V-A14B for MoE, TI2V-5B for unified T2V+I2V, LTX-2,
-  Cosmos3, or VACE).
+- `--model`: Model ID (I2V-A14B for MoE, TI2V-5B for unified T2V+I2V,
+  LTX-2/LTX-2.3, Cosmos3, or VACE).
 - `--image`: Path to the first-frame or source image.
 - `--last-image`: Optional last-frame condition for models such as VACE.
 - `--mask-image`: Optional inpainting mask. White pixels are regenerated and black pixels are preserved.
@@ -261,7 +290,7 @@ Key arguments:
 - `--guidance-scale` and `--guidance-scale-high`: CFG scale (applied to low/high-noise stages for MoE).
 - `--negative-prompt`: Optional list of artifacts to suppress.
 - `--boundary-ratio`: Boundary split ratio for two-stage MoE models.
-- `--flow-shift`: Scheduler flow shift (default: Wan 5.0, Cosmos3 10.0). LTX does not consume this option.
+- `--flow-shift`: Scheduler flow shift (default: model-specific — Wan/LTX2 5.0, Cosmos3 10.0).
 - `--sample-solver`: Wan2.2 sampling solver. Use `unipc` for the default multistep solver, or `euler` for Lightning/Distill checkpoints.
 - `--num-inference-steps`: Number of denoising steps (default: model-specific — Wan 50, LTX2 40, Cosmos3 35).
 - `--fps`: Frames per second for the saved MP4 (requires `diffusers` export_to_video).

examples/offline_inference/image_to_video/image_to_video.py

@@ -23,14 +23,16 @@ Usage:
 
     # LTX2 image-to-video
     python image_to_video.py --model /path/to/LTX-2 \
+        --model-class-name LTX2ImageToVideoPipeline \
         --image input.jpg --prompt "A cinematic dolly shot of a boat" \
-        --num-frames 121 --num-inference-steps 40 \
+        --num-frames 121 --num-inference-steps 40 --guidance-scale 4.0 \
         --frame-rate 24 --fps 24 --output ltx2_i2v.mp4
 
     # LTX-2.3 image-to-video
     python image_to_video.py --model diffusers/LTX-2.3-Diffusers \
+        --model-class-name LTX23ImageToVideoPipeline \
         --image input.jpg --prompt "A cinematic dolly shot of a boat" \
-        --height 512 --width 768 --num-frames 121 --num-inference-steps 30 \
+        --height 384 --width 512 --num-frames 25 --num-inference-steps 20 \
         --frame-rate 24 --fps 24 --output ltx23_i2v.mp4
 
     # HunyuanVideo-1.5 I2V (480p)
@@ -98,7 +100,7 @@ def parse_args() -> argparse.Namespace:
     parser.add_argument(
         "--model-class-name",
         default=None,
-        help="Override model class name (LTX checkpoints default to LTX2Pipeline).",
+        help="Override model class name (e.g., LTX2ImageToVideoPipeline or LTX23ImageToVideoPipeline).",
     )
     parser.add_argument(
         "--deploy-config",
@@ -115,7 +117,7 @@ def parse_args() -> argparse.Namespace:
         help="Path to a reference image. Repeat to provide multiple references.",
     )
     parser.add_argument("--prompt", default="", help="Text prompt describing the desired motion.")
-    parser.add_argument("--negative-prompt", default=None, help="Negative prompt. Default: model-specific.")
+    parser.add_argument("--negative-prompt", default="", help="Negative prompt.")
     parser.add_argument("--seed", type=int, default=42, help="Random seed.")
     parser.add_argument("--guidance-scale", type=float, default=None, help="CFG scale. Default: model-specific.")
     parser.add_argument(
@@ -329,10 +331,7 @@ def main():
     generator = torch.Generator(device=current_omni_platform.device_type).manual_seed(args.seed)
     model_name = str(args.model).lower() if args.model is not None else ""
     model_class_name = args.model_class_name
-    model_class_name_lower = (model_class_name or "").lower()
-    is_ltx2_distilled = "distilled" in model_class_name_lower or "distilled" in model_name
-    is_ltx23 = "ltx23" in model_class_name_lower or "ltx-2.3" in model_name
-    is_ltx2 = is_ltx2_distilled or is_ltx23 or "ltx2" in model_class_name_lower or "ltx-2" in model_name
+    is_ltx2 = model_class_name in {"LTX2ImageToVideoPipeline", "LTX23ImageToVideoPipeline"}
     is_cosmos = "cosmos" in model_name or (model_class_name is not None and "cosmos" in model_class_name.lower())
 
     image = PIL.Image.open(args.image).convert("RGB") if args.image else None
@@ -357,23 +356,13 @@ def main():
             1280 * 720,
             16,
         )
-    elif is_ltx2_distilled:
-        d_fps, d_guidance, d_num_frames, d_steps, d_flow_shift, d_max_area, d_mod = (
-            24,
-            None,
-            121,
-            8,
-            None,
-            1024 * 1536,
-            64,
-        )
     elif is_ltx2:
         d_fps, d_guidance, d_num_frames, d_steps, d_flow_shift, d_max_area, d_mod = (
             24,
-            None,
+            4.0,
             121,
-            30 if is_ltx23 else 40,
-            None,
+            40,
+            5.0,
             512 * 768,
             32,
         )
@@ -456,6 +445,7 @@ def main():
         vae_use_slicing=args.vae_use_slicing,
         vae_use_tiling=args.vae_use_tiling,
         boundary_ratio=args.boundary_ratio,
+        flow_shift=flow_shift,
         diffusion_kv_cache_dtype=args.diffusion_kv_cache_dtype,
         diffusion_kv_cache_skip_steps=args.diffusion_kv_cache_skip_steps,
         diffusion_kv_cache_skip_layers=args.diffusion_kv_cache_skip_layers,
@@ -470,8 +460,6 @@ def main():
     )
     if args.deploy_config:
         omni_kwargs["deploy_config"] = args.deploy_config
-    if flow_shift is not None:
-        omni_kwargs["flow_shift"] = flow_shift
     if args.quantization is not None:
         omni_kwargs["quantization"] = args.quantization
     # Cosmos3 loads its (gated) guardrail models at build time, so the guardrails
@@ -504,14 +492,10 @@ def main():
     print(f"  Video size: {width}x{height}")
     print(f"{'=' * 60}\n")
 
-    negative_prompt = args.negative_prompt
-    if negative_prompt is None and not is_ltx2:
-        # Preserve the historical empty-prompt behavior for non-LTX examples.
-        negative_prompt = ""
     prompt_dict = build_image_to_video_prompt(
         model_class_name=model_class_name,
         prompt=args.prompt,
-        negative_prompt=negative_prompt,
+        negative_prompt=args.negative_prompt,
         media_inputs=media_inputs,
         height=height,
         width=width,
@@ -527,10 +511,11 @@ def main():
         num_inference_steps=num_inference_steps,
         num_frames=num_frames,
         frame_rate=frame_rate,
-        extra_args={"sample_solver": args.sample_solver},
+        extra_args={
+            "sample_solver": args.sample_solver,
+            "flow_shift": flow_shift,
+        },
     )
-    if flow_shift is not None:
-        sampling_params.extra_args["flow_shift"] = flow_shift
 
     # Route model-specific knobs through extra_body, filtered against the model's
     # declared extra_body_params. Models without a declaration only forward explicit

examples/offline_inference/minicpmo/README.md

@@ -1,8 +1,9 @@
 # MiniCPM-o 4.5: Offline inference
 
-Two-stage pipeline: **thinker** (multimodal understanding) → **talker + Token2Wav**
+Three-stage pipeline: **Thinker** → **Talker** (codec) → **Code2Wav**
 (24 kHz speech). Deploy config auto-loads from
-`vllm_omni/deploy/minicpmo_4_5.yaml` (2-GPU default).
+`vllm_omni/deploy/minicpmo_4_5.yaml` (single-GPU default). For higher
+throughput, pass `--deploy-config vllm_omni/deploy/minicpmo_4_5_2gpu.yaml`.
 
 ## Setup
 
@@ -72,7 +73,7 @@ Supported `--query-type` values:
 
 ```bash
 python end2end.py --query-type text \
-    --deploy-config /path/to/vllm_omni/deploy/minicpmo_4_5_8x4090.yaml
+    --deploy-config /path/to/vllm_omni/deploy/minicpmo_4_5_2gpu.yaml
 ```
 
 ## Notes
@@ -83,7 +84,8 @@ python end2end.py --query-type text \
 - Output WAV is **24 kHz mono**.
 - Placeholders in the prompt are MiniCPM-style:
   `(<image>./</image>)`, `(<audio>./</audio>)`, `(<video>./</video>)`.
-- Default layout needs **2 GPUs**. Async chunking is off in the bundled YAMLs.
+- Default layout needs **1 GPU**. Use `minicpmo_4_5_2gpu.yaml` for the
+  recommended two-GPU continuous-batching layout.
 
 ## Online serving
 

examples/offline_inference/minicpmo/end2end.py

@@ -16,7 +16,6 @@ import numpy as np
 import soundfile as sf
 import vllm
 from PIL import Image
-from vllm import SamplingParams
 from vllm.assets.audio import AudioAsset
 from vllm.assets.image import ImageAsset
 from vllm.assets.video import VideoAsset, video_to_ndarrays
@@ -26,8 +25,6 @@ from vllm.multimodal.media.audio import load_audio
 from vllm_omni.entrypoints.omni import Omni
 from vllm_omni.utils.tracking_parser import TrackingArgumentParser
 
-SEED = 42
-
 default_system = (
     "You are MiniCPM-o, a helpful multimodal assistant that can "
     "understand images, audio and video, and respond in text and speech."
@@ -285,27 +282,17 @@ def main(args):
     omni_kwargs["trust_remote_code"] = True
     omni = Omni(**omni_kwargs)
 
-    # Stage 0 (thinker): multimodal understanding → text (+ TTS span when enabled).
-    thinker_sampling_params = SamplingParams(
-        temperature=0.0,
-        top_p=1.0,
-        top_k=-1,
-        max_tokens=2048,
-        seed=SEED,
-        detokenize=True,
-        repetition_penalty=1.1,
-    )
-    # Stage 1 (talker + Token2Wav): max_tokens=1 satisfies the scheduler;
-    # waveform is produced in-process by Token2Wav.
-    talker_sampling_params = SamplingParams(
-        temperature=0.0,
-        top_p=1.0,
-        top_k=-1,
-        max_tokens=1,
-        seed=SEED,
-        detokenize=False,
-    )
-    sampling_params_list = [thinker_sampling_params, talker_sampling_params][: omni.num_stages]
+    # Per-stage sampling defaults live in the deploy YAML
+    # (``vllm_omni/deploy/minicpmo_4_5*.yaml``, ``default_sampling_params``),
+    # which is the single source of truth that tracks the pipeline topology.
+    # Passing None makes ``resolve_sampling_params_list`` load exactly those,
+    # so adding or removing a stage cannot desynchronize this example.
+    #
+    # This used to hardcode a two-element list built for the old two-stage
+    # topology (thinker → talker+Token2Wav). Since Code2Wav became its own
+    # Stage 2, that list was both the wrong length and wrong in content
+    # (talker ``max_tokens=1`` truncated the codec stream to one frame).
+    sampling_params_list = None
 
     if args.txt_prompts is None:
         prompts = [query_result.inputs for _ in range(args.num_prompts)]

examples/offline_inference/text_to_video/text_to_video.md

@@ -8,7 +8,7 @@ A unified script for text-to-video generation. Supports multiple models with mod
 |---|---|---|---|---|---|
 | `Wan-AI/Wan2.1-VACE-1.3B-diffusers` | 480x832 | 81 | 30 | 5.0 | ~20 GiB (RTX 5090, VAE tiling) |
 | `Wan-AI/Wan2.2-T2V-A14B-Diffusers` | 720x1280 | 81 | 40 | 4.0 | ~60 GiB |
-| `Lightricks/LTX-2` | 512x768 | 121 | 40 | video 3.0 / audio 7.0 | Model-dependent |
+| `diffusers/LTX-2.3-Diffusers` | 384x512 | 25 | 20 | 4.0 | 96GB-class GPU |
 | `hunyuanvideo-community/HunyuanVideo-1.5-Diffusers-480p_t2v` | 480x832 | 121 | 50 | 6.0 | 1×A100 80GB |
 | `hunyuanvideo-community/HunyuanVideo-1.5-Diffusers-720p_t2v` | 720x1280 | 121 | 50 | 6.0 | FP8 + VAE tiling required |
 | `nvidia/Cosmos3-Nano` | 720x1280 | 189 | 35 | 6.0 | ~46 GiB (peak, 720p) |
@@ -55,17 +55,45 @@ python text_to_video.py \
   --output vace_t2v_output.mp4
 ```
 
-### LTX-2
+LTX2 example:
 
 ```bash
 python text_to_video.py \
-  --model Lightricks/LTX-2 \
-  --prompt "Cherry blossoms swaying gently in the breeze with synchronized ambient sound" \
-  --output ltx2_output.mp4
+  --model "Lightricks/LTX-2" \
+  --prompt "A cinematic close-up of ocean waves at golden hour." \
+  --negative-prompt "worst quality, inconsistent motion, blurry, jittery, distorted" \
+  --height 512 \
+  --width 768 \
+  --num-frames 121 \
+  --num-inference-steps 40 \
+  --guidance-scale 4.0 \
+  --frame-rate 24 \
+  --output ltx2_out.mp4
 ```
 
-See the [LTX-2 recipe](../../../recipes/LTX/LTX-2.md) for all checkpoints,
-pipeline selection, I2V, defaults, and advanced options.
+### LTX-2.3
+
+```bash
+python text_to_video.py \
+  --model diffusers/LTX-2.3-Diffusers \
+  --model-class-name LTX23Pipeline \
+  --prompt "Cherry blossoms swaying gently in the breeze with synchronized ambient sound" \
+  --negative-prompt "worst quality, inconsistent motion, blurry, jittery, distorted" \
+  --height 384 \
+  --width 512 \
+  --num-frames 25 \
+  --num-inference-steps 20 \
+  --guidance-scale 4.0 \
+  --frame-rate 24 \
+  --fps 24 \
+  --audio-sample-rate 48000 \
+  --output ltx23_t2v_output.mp4
+```
+
+Use the Diffusers-format checkpoint `diffusers/LTX-2.3-Diffusers`; the
+upstream `Lightricks/LTX-2.3` raw safetensors repo is not directly loadable by
+this pipeline. Pass `--model-class-name LTX23Pipeline` to select the LTX-2.3
+text-to-video pipeline explicitly.
 
 ### HunyuanVideo-1.5 (480p)
 
@@ -184,7 +212,8 @@ python text_to_video.py \
 ### Common
 
 - `--model`: Diffusers model ID or local path.
-- `--model-class-name`: Optional explicit pipeline override.
+- `--model-class-name`: Optional explicit pipeline class. Use `LTX23Pipeline`
+  for LTX-2.3 text-to-video.
 - `--prompt`: text description (string).
 - `--height/--width`: output resolution. Default depends on model.
 - `--num-frames`: number of frames. Default depends on model.
@@ -200,7 +229,8 @@ python text_to_video.py \
 - `--enable-cpu-offload`: enable CPU offloading for diffusion models.
 - `--enable-layerwise-offload`: enable layerwise offloading on DiT modules.
 - `--frame-rate`: generation FPS for pipelines that require it (e.g., LTX2).
-- `--audio-sample-rate`: fallback audio sample rate when the pipeline returns audio.
+- `--audio-sample-rate`: audio sample rate for embedded audio (when the
+  pipeline returns audio; LTX-2.3 outputs 48kHz audio).
 - `--quantization`: quantization method (such as `fp8` for FP8).
 - `--flow-shift`: scheduler flow_shift parameter.
 - `--extra-body`: JSON object of model-specific generation params, filtered against the model's declared `extra_body_params` (see [`vllm_omni/model_extras`](../../../vllm_omni/model_extras)). Used by Cosmos3 (see above).

examples/offline_inference/text_to_video/text_to_video.py

@@ -65,42 +65,11 @@ _MODEL_PRESETS = {
         "fps": 16,
         "output": "helios_output.mp4",
     },
-    "ltx2": {
-        "height": 512,
-        "width": 768,
-        "num_frames": 121,
-        "num_inference_steps": 40,
-        "fps": 24,
-        "output": "ltx2_output.mp4",
-    },
-    "ltx2_distilled": {
-        "height": 1024,
-        "width": 1536,
-        "num_frames": 121,
-        "num_inference_steps": 8,
-        "fps": 24,
-        "output": "ltx2_distilled_output.mp4",
-    },
-    "ltx23": {
-        "height": 512,
-        "width": 768,
-        "num_frames": 121,
-        "num_inference_steps": 30,
-        "fps": 24,
-        "output": "ltx23_output.mp4",
-    },
 }
 
 
-def _detect_preset(model: str, model_class_name: str | None = None) -> dict:
+def _detect_preset(model: str) -> dict:
     model_lower = model.lower()
-    class_lower = (model_class_name or "").lower()
-    if "distilled" in class_lower or "distilled" in model_lower:
-        return _MODEL_PRESETS["ltx2_distilled"]
-    if "ltx23" in class_lower or "ltx-2.3" in model_lower or "ltx_2.3" in model_lower:
-        return _MODEL_PRESETS["ltx23"]
-    if "ltx2" in class_lower or "ltx-2" in model_lower or "ltx_2" in model_lower:
-        return _MODEL_PRESETS["ltx2"]
     if "vace" in model_lower:
         return _MODEL_PRESETS["vace"]
     if "cosmos" in model_lower:
@@ -153,7 +122,7 @@ def parse_args() -> argparse.Namespace:
     parser.add_argument(
         "--model-class-name",
         default=None,
-        help="Override model class name (e.g., LTX2Pipeline).",
+        help="Override model class name (e.g., LTX2TwoStagesVideoPipeline).",
     )
     parser.add_argument(
         "--deploy-config",
@@ -161,7 +130,7 @@ def parse_args() -> argparse.Namespace:
         help="Optional deploy config YAML to use for pipeline-backed runs.",
     )
     parser.add_argument("--prompt", default="A serene lakeside sunrise with mist over the water.", help="Text prompt.")
-    parser.add_argument("--negative-prompt", default=None, help="Negative prompt. Default: model-specific.")
+    parser.add_argument("--negative-prompt", default="", help="Negative prompt.")
     parser.add_argument(
         "--extra-body",
         type=parse_extra_body,
@@ -362,7 +331,7 @@ def main():
     args = parse_args()
     model_class_name = args.model_class_name
 
-    preset = _detect_preset(args.model, model_class_name)
+    preset = _detect_preset(args.model)
     for key, default_val in preset.items():
         if getattr(args, key.replace("-", "_"), None) is None:
             setattr(args, key.replace("-", "_"), default_val)
@@ -453,11 +422,8 @@ def main():
     print(f"{'=' * 60}\n")
 
     prompt_dict = {"prompt": args.prompt}
-    if args.negative_prompt is not None:
+    if args.negative_prompt:
         prompt_dict["negative_prompt"] = args.negative_prompt
-    elif preset not in (_MODEL_PRESETS["ltx2"], _MODEL_PRESETS["ltx23"]):
-        # Preserve the historical empty-prompt behavior for non-LTX examples.
-        prompt_dict["negative_prompt"] = ""
 
     sampling_kwargs = dict(
         height=args.height,

examples/online_serving/audex/README.md

@@ -1,89 +0,0 @@
-# Audex (Nemotron-Labs-Audex-2B) online serving
-
-One checkpoint, four deployment modes. `run_server.sh` starts the server
-for a given `MODE`; `client.py --mode <mode>` tests it. The capability
-matrix:
-
-| mode (`vllm_omni/deploy/audex_<mode>.yaml`) | audio in | text out | speech out | general audio out | endpoint |
-|---|---|---|---|---|---|
-| `tts` | ❌ | ❌ | ✅ | ❌ | `/v1/audio/speech` |
-| `tta` | ❌ | ❌ | ❌ | ✅ | `/v1/audio/speech` |
-| `thinker_only` | ✅ | ✅ | ❌ | ❌ | `/v1/chat/completions` |
-| `s2s` | ✅ | ✅ | ✅ | ❌ | both |
-
-Audio-input modes of `client.py` fall back to vLLM's public
-`mary_had_lamb` asset when `--audio-file` is omitted; CFG defaults follow
-the official settings per mode (tts/s2s 1.5, tta 3.0; `--cfg-scale 1.0`
-disables).
-
-## tts — text → speech
-
-Text-only thinker (`checkpoint_folder_audiogen`) emits `<speechcodec_N>`
-tokens; the streaming causal decoder produces 16 kHz WAVs. This is also
-the DEFAULT pipeline when the repo root is served without a deploy config.
-
-```bash
-./run_server.sh                       # MODE=tts, port 8097
-
-python client.py --mode tts --text "Hello there." --output hello.wav
-# or raw curl:
-curl -s http://localhost:8097/v1/audio/speech \
-    -H 'Content-Type: application/json' \
-    -d '{"model": "nvidia/Nemotron-Labs-Audex-2B", "input": "Hello there.", "response_format": "wav", "extra_params": {"cfg_scale": 1.5}}' \
-    -o hello.wav
-```
-
-## tta — caption → general audio
-
-Same thinker over the 4-codebook `<audiocodec_N>` RVQ block, decoded by
-the external XCodec1 checkpoint (auto-downloaded; override with
-`XCODEC1_PATH`). Clips are capped at 10 s.
-
-```bash
-MODE=tta ./run_server.sh
-
-python client.py --mode tta --caption "Heavy rain falling on a tin roof." \
-    --output rain.wav
-```
-
-## thinker_only — audio (+ instruction) → text
-
-Single-stage audio understanding on the full checkpoint (NV-Whisper
-encoder + projector + LM). Send the clip as `input_audio` chat content.
-
-```bash
-MODE=thinker_only ./run_server.sh
-
-python client.py --mode thinker_only                      # transcribe the demo asset
-python client.py --mode thinker_only --audio-file a.wav \
-    --question "What language is being spoken?"
-```
-
-## s2s — spoken question → spoken answer
-
-The audio-capable thinker plus the speech decoder in one deployment.
-`client.py --mode s2s` runs the official three passes: ASR (chat with
-`input_audio`) → chat on the transcript → TTS on the answer. Text-final
-passes carry `"modalities": ["text"]`; only the TTS pass streams through
-the decoder.
-
-```bash
-MODE=s2s PORT=8098 ./run_server.sh
-
-python client.py --mode s2s --port 8098 \
-    --audio-file question.wav --output answer.wav
-```
-
-Offline counterparts (no HTTP, one `Omni` engine per script) live in
-`examples/offline_inference/audex/`.
-
-## 30B-A3B (nvidia/Nemotron-Labs-Audex-30B-A3B)
-
-`SIZE=30b` switches run_server.sh to the 30B yamls and model id:
-
-    SIZE=30b ./run_server.sh                 # 30B tts on port 8097
-    SIZE=30b MODE=s2s PORT=8098 ./run_server.sh
-
-Pass `--model nvidia/Nemotron-Labs-Audex-30B-A3B` to the client — the server
-validates the model id and 404s on a mismatch. Single-H100 defaults; see the
-yaml comments for the TP2 fallback. First launch downloads ~60 GB.

examples/online_serving/audex/client.py

@@ -1,319 +0,0 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
-"""Audex online client for all four deployment modes.
-
-Pick ``--mode`` to match the server's deploy yaml (see run_server.sh):
-
-  tts           /v1/audio/speech: text -> speech WAV
-  tta           /v1/audio/speech: caption -> general-audio WAV
-  thinker_only  /v1/chat/completions: audio (+ question) -> text
-  s2s           the official three-pass cascade against one audex_s2s
-                server: ASR (chat, input_audio) -> chat -> TTS
-
-Audio-input modes fall back to vLLM's public ``mary_had_lamb`` asset when
-``--audio-file`` is omitted.
-
-Examples:
-
-    python examples/online_serving/audex/client.py --mode tts \\
-        --text "Hello there." --output hello.wav
-    python examples/online_serving/audex/client.py --mode tta \\
-        --caption "Heavy rain falling on a tin roof." --output rain.wav
-    python examples/online_serving/audex/client.py --mode thinker_only
-    python examples/online_serving/audex/client.py --mode s2s \\
-        --audio-file question.wav --output answer.wav
-"""
-
-from __future__ import annotations
-
-import argparse
-import base64
-import re
-from pathlib import Path
-
-import requests
-
-ASR_QUESTION = "Transcribe the input speech."
-SYSTEM_PROMPT = "You are a helpful and harmless assistant.\n\nYou are not allowed to use any tools."
-# The official chat-pass recipe below (SYSTEM_PROMPT + TEXT_INSTRUCTION
-# prefix, reasoning enabled, temp 1.0/top_p 0.95, seed retry) applies to the
-# 30B-A3B ONLY (see is_30b() below and
-# tests/e2e/offline_inference/test_audex_s2s.py, which mirrors this same
-# split). It is load-bearing on the 30B: without the formatting-instruction
-# prefix a bare transcript turn is treated as a TTS request and the model
-# emits literal <speechcodec_N> tokens instead of a text answer (the system
-# prompt alone does not prevent this). The 2B must NOT use this recipe: the
-# official temp-1.0 seed-0 draw measurably hallucinates off-language text on
-# the 2B (yet closes </think> cleanly, so the seed retry never fires) and
-# fails the round-trip WER gate — measured, reproduced byte-identically. The
-# 2B instead keeps its original bare-message / enable_thinking=False /
-# temperature=0.0 recipe.
-TEXT_INSTRUCTION = (
-    "SYSTEM & FORMATTING INSTRUCTION: You are Audex, created by NVIDIA based on the Nemotron-Cascade-2 architecture. "
-    "You may output your reasoning, followed by your final response. "
-    "You may format your reasoning block however you like. "
-    "However, your final response must strictly follow these rules: "
-    "* Write in plain, unformatted prose like a book or newspaper article. "
-    "* Do not use markdown, bullet points, lists, or headers. "
-    "* Standard numbers, abbreviations, and symbols are acceptable. "
-    "* [CRITICAL] You must press enter after every single sentence, placing each sentence on its own separate line."
-)
-DEFAULT_TEXT = "Hello there! This is the Audex text to speech pipeline."
-DEFAULT_CAPTION = "Heavy rain falling on a tin roof."
-# Official quality settings; 1.0 disables guidance.
-DEFAULT_CFG = {"tts": 1.5, "tta": 3.0, "s2s": 1.5}
-
-
-def is_30b(model: str) -> bool:
-    # Users pass arbitrary repo IDs or local snapshot paths, so this is a
-    # heuristic (not an exact match against the official 30B-A3B repo ID)
-    # — good enough for an example script; keep in sync with the same
-    # branch in examples/offline_inference/audex/speech_to_speech.py and
-    # tests/e2e/offline_inference/test_audex_s2s.py.
-    return "30b" in model.lower()
-
-
-def parse_args():
-    parser = argparse.ArgumentParser(description="Audex online client (tts / tta / thinker_only / s2s)")
-    parser.add_argument("--mode", choices=("tts", "tta", "thinker_only", "s2s"), default="s2s")
-    parser.add_argument("--host", type=str, default="127.0.0.1")
-    parser.add_argument("--port", type=int, default=8097)
-    parser.add_argument("--model", type=str, default="nvidia/Nemotron-Labs-Audex-2B")
-    parser.add_argument("--text", type=str, default=DEFAULT_TEXT, help="TTS input text (mode=tts).")
-    parser.add_argument("--caption", type=str, default=DEFAULT_CAPTION, help="TTA caption (mode=tta).")
-    parser.add_argument(
-        "--audio-file",
-        type=str,
-        default=None,
-        help="Input WAV for thinker_only/s2s. Defaults to vLLM's mary_had_lamb asset.",
-    )
-    parser.add_argument("--question", type=str, default=ASR_QUESTION, help="Instruction (mode=thinker_only).")
-    parser.add_argument("--output", type=str, default="results/audex_online.wav")
-    parser.add_argument(
-        "--cfg-scale",
-        type=float,
-        default=None,
-        help="CFG strength; defaults per mode (tts/s2s 1.5, tta 3.0). 1.0 disables.",
-    )
-    parser.add_argument("--max-tokens", type=int, default=512)
-    return parser.parse_args()
-
-
-def _audio_b64(audio_file: str | None) -> str:
-    if audio_file:
-        return base64.b64encode(Path(audio_file).read_bytes()).decode()
-    # Fall back to vLLM's public asset, re-encoded as WAV for the chat API.
-    import io
-
-    import numpy as np
-    import soundfile as sf
-    from vllm.assets.audio import AudioAsset
-
-    audio, sr = AudioAsset("mary_had_lamb").audio_and_sample_rate
-    buf = io.BytesIO()
-    sf.write(buf, np.asarray(audio, dtype="float32"), int(sr), format="WAV")
-    return base64.b64encode(buf.getvalue()).decode()
-
-
-def _chat(
-    base_url: str,
-    model: str,
-    messages: list[dict],
-    max_tokens: int,
-    enable_thinking: bool = False,
-    temperature: float = 0.0,
-    top_p: float = 1.0,
-    seed: int = 0,
-) -> tuple[str, str]:
-    resp = requests.post(
-        f"{base_url}/v1/chat/completions",
-        json={
-            "model": model,
-            "messages": messages,
-            "max_tokens": max_tokens,
-            "temperature": temperature,
-            "top_p": top_p,
-            "seed": seed,
-            # Text-final passes must say so explicitly: without "modalities"
-            # the server falls back to the deployment's configured output
-            # modalities, which for the s2s pipeline include the audio final
-            # stage — routing ASR/chat through code2wav instead of stopping
-            # at stage 0.
-            "modalities": ["text"],
-            # The checkpoint's chat template thinking toggle. The official
-            # cascaded demo transcribes with enable_thinking=False but runs
-            # the chat pass with reasoning ENABLED (its web UI ships the
-            # "Enable reasoning" checkbox checked), priming the generation
-            # prompt with an open "<think>\n" so the model's reasoning stays
-            # inside the think block; _extract_final_response then drops it.
-            # With thinking disabled the 30B leaks its reasoning as plain
-            # text in front of the final response.
-            "chat_template_kwargs": {"enable_thinking": enable_thinking},
-        },
-        timeout=300,
-    )
-    resp.raise_for_status()
-    choice = resp.json()["choices"][0]
-    return choice["message"]["content"].strip(), choice.get("finish_reason") or ""
-
-
-def _speech(base_url: str, model: str, text: str, cfg_scale: float, out_path: Path) -> None:
-    payload: dict = {"model": model, "input": text, "response_format": "wav"}
-    if cfg_scale > 1.0:
-        payload["extra_params"] = {"cfg_scale": cfg_scale}
-    resp = requests.post(f"{base_url}/v1/audio/speech", json=payload, timeout=600)
-    resp.raise_for_status()
-    out_path.parent.mkdir(parents=True, exist_ok=True)
-    out_path.write_bytes(resp.content)
-
-
-def _audio_question(audio_b64: str, question: str) -> list[dict]:
-    return [
-        {
-            "role": "user",
-            "content": [
-                {"type": "input_audio", "input_audio": {"data": audio_b64, "format": "wav"}},
-                {"type": "text", "text": question},
-            ],
-        }
-    ]
-
-
-def _tts_text(answer: str) -> str:
-    # The formatting instruction's CRITICAL rule puts each final-response
-    # sentence on its own line, and the official demo TTS-es those sentence
-    # segments (split_sentence_segments). Mirror that per-line segmentation
-    # but keep a single /v1/audio/speech request: speak the leading whole
-    # sentences of the CLEANED final response up to ~200 chars, never
-    # cutting mid-sentence. Markdown markup is stripped as before (it makes
-    # the TTS thinker stop almost immediately).
-    sentences = []
-    for line in answer.splitlines():
-        line = re.sub(r"[*#_`]+", "", line)
-        line = re.sub(r"\s+", " ", line).strip()
-        if line:
-            sentences.append(line)
-    picked: list[str] = []
-    total = 0
-    for sentence in sentences:
-        if picked and total + len(sentence) > 200:
-            break
-        picked.append(sentence)
-        total += len(sentence) + 1
-    return " ".join(picked)
-
-
-def _extract_final_response(text: str) -> str:
-    # Mirror the official demo's extract_spoken_answer
-    # (cascaded_s2s_web_server.py): the chat pass reasons inside the primed
-    # <think> block, so keep only what follows the LAST closing tag; fall
-    # back to the full text when the model skipped reasoning or never
-    # closed the block. Benign no-op for the 2B's clean answers.
-    if "</think>" in text:
-        answer = text.rsplit("</think>", 1)[-1].strip()
-        if answer:
-            return answer
-    return text.strip()
-
-
-def _clean_transcript(text: str) -> str:
-    # The full checkpoint prefixes transcripts with a language-tag sentence
-    # and quotes the content; extract the quoted payload when present. Span
-    # first-to-last quote (not split on every apostrophe) so contractions
-    # inside the transcript (e.g. "don't") survive intact.
-    text = text.strip()
-    match = re.search(r"'(.*)'", text, flags=re.S)
-    if match and match.group(1).strip():
-        return match.group(1).strip()
-    return text
-
-
-def main():
-    args = parse_args()
-    base_url = f"http://{args.host}:{args.port}"
-    cfg_scale = args.cfg_scale if args.cfg_scale is not None else DEFAULT_CFG.get(args.mode, 1.0)
-    out_path = Path(args.output)
-
-    if args.mode == "tts":
-        _speech(base_url, args.model, args.text, cfg_scale, out_path)
-        print(f"speech: {out_path.stat().st_size} bytes -> {out_path}")
-        return
-
-    if args.mode == "tta":
-        _speech(base_url, args.model, args.caption, cfg_scale, out_path)
-        print(f"audio : {out_path.stat().st_size} bytes -> {out_path}")
-        return
-
-    audio_b64 = _audio_b64(args.audio_file)
-
-    if args.mode == "thinker_only":
-        content, _ = _chat(base_url, args.model, _audio_question(audio_b64, args.question), args.max_tokens)
-        print(f"answer: {_extract_final_response(content)}")
-        return
-
-    # s2s — the official three-pass cascade.
-    asr_content, _ = _chat(base_url, args.model, _audio_question(audio_b64, ASR_QUESTION), args.max_tokens)
-    transcript = _clean_transcript(asr_content)
-    print(f"[1/3] transcript : {transcript}")
-    if not transcript:
-        raise SystemExit("ASR pass produced an empty transcript; aborting cascade")
-
-    # The chat-pass recipe is size-conditional (see the TEXT_INSTRUCTION
-    # comment above): the 30B requires the official recipe, the 2B keeps its
-    # original bare-message recipe.
-    if is_30b(args.model):
-        # Official recipe: system prompt + formatting instruction prefixed
-        # to the transcript, reasoning enabled, and the official demo's
-        # text-pass sampling (its defaults: --text-temperature 1.0,
-        # --text-top-p 0.95, --text-max-tokens 2048). Near-greedy decoding
-        # makes the 30B loop inside its reasoning block until the token
-        # budget runs out without ever closing </think>. Even at the
-        # official temperature some trajectories spin past the budget, so
-        # retry a few deterministic seeds; a natural stop without </think>
-        # is a complete answer, and if every seed spins we keep the full
-        # text like the official demo does.
-        chat_messages = [
-            {"role": "system", "content": SYSTEM_PROMPT},
-            {"role": "user", "content": f"{TEXT_INSTRUCTION}\n\n{transcript}"},
-        ]
-        raw = ""
-        for seed in (0, 1, 2):
-            raw, finish_reason = _chat(
-                base_url,
-                args.model,
-                chat_messages,
-                max(args.max_tokens, 2048),
-                enable_thinking=True,
-                temperature=1.0,
-                top_p=0.95,
-                seed=seed,
-            )
-            if "</think>" in raw or finish_reason != "length":
-                break
-            print(f"[2/3] chat seed {seed} hit the token budget mid-reasoning; retrying")
-        answer = _extract_final_response(raw)
-    else:
-        # 2B: bare user message (no system/instruction prefix), reasoning
-        # disabled, temperature 0.0, no retry — its original recipe. The
-        # official temp-1.0 seed-0 recipe measurably makes the 2B
-        # hallucinate off-language answers (see the module-level comment).
-        chat_messages = [{"role": "user", "content": transcript}]
-        raw, _ = _chat(
-            base_url,
-            args.model,
-            chat_messages,
-            args.max_tokens,
-            enable_thinking=False,
-            temperature=0.0,
-        )
-        answer = _extract_final_response(raw)
-    print(f"[2/3] answer     : {answer}")
-    if not answer:
-        raise SystemExit("Chat pass produced an empty answer; aborting cascade")
-
-    _speech(base_url, args.model, _tts_text(answer), cfg_scale, out_path)
-    print(f"[3/3] speech     : {out_path.stat().st_size} bytes -> {out_path}")
-
-
-if __name__ == "__main__":
-    main()

examples/online_serving/audex/run_server.sh

@@ -1,57 +0,0 @@
-#!/bin/bash
-# Launch the vLLM-Omni server for Audex (Nemotron-Labs-Audex-2B).
-#
-# MODE picks the deployment (see README.md for the capability matrix):
-#   tts           text -> speech            /v1/audio/speech   (default)
-#   tta           caption -> general audio  /v1/audio/speech
-#   thinker_only  audio -> text             /v1/chat/completions
-#   s2s           cascaded speech-to-speech (both endpoints)
-#
-# Pass the HF repo ROOT as MODEL — per-stage subfolders resolve
-# automatically.
-#
-# Usage:
-#   ./run_server.sh                       # 2B tts on port 8097, GPU 0
-#   SIZE=30b ./run_server.sh              # 30B-A3B tts (explicit 30B yaml)
-#   SIZE=30b MODE=s2s PORT=8098 ./run_server.sh
-#   MODE=tta ./run_server.sh              # needs XCodec1 (auto-downloaded)
-#   MODEL=/path/to/local/snapshot ./run_server.sh
-
-set -e
-
-MODE="${MODE:-${1:-tts}}"
-SIZE="${SIZE:-2b}"
-if [ "$SIZE" = "30b" ]; then
-    MODEL="${MODEL:-nvidia/Nemotron-Labs-Audex-30B-A3B}"
-    YAML_SUFFIX="_30b"
-elif [ "$SIZE" = "2b" ]; then
-    MODEL="${MODEL:-nvidia/Nemotron-Labs-Audex-2B}"
-    YAML_SUFFIX=""
-else
-    echo "Unknown SIZE '$SIZE'; expected 2b|30b" >&2
-    exit 1
-fi
-PORT="${PORT:-8097}"
-GPUS="${GPUS:-0}"
-
-REPO_ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/../../.." && pwd)"
-DEPLOY_YAML="$REPO_ROOT/vllm_omni/deploy/audex_${MODE}${YAML_SUFFIX}.yaml"
-if [ ! -f "$DEPLOY_YAML" ]; then
-    echo "Unknown MODE '$MODE' / SIZE '$SIZE' (no $DEPLOY_YAML); expected MODE=tts|tta|thinker_only|s2s, SIZE=2b|30b" >&2
-    exit 1
-fi
-
-echo "Starting Audex server"
-echo "  MODE=$MODE ($DEPLOY_YAML)"
-echo "  SIZE=$SIZE"
-echo "  MODEL=$MODEL"
-echo "  PORT=$PORT"
-echo "  CUDA_VISIBLE_DEVICES=$GPUS"
-
-CUDA_VISIBLE_DEVICES="$GPUS" \
-vllm-omni serve "$MODEL" \
-    --host 0.0.0.0 \
-    --port "$PORT" \
-    --trust-remote-code \
-    --stage-configs-path "$DEPLOY_YAML" \
-    --omni

examples/online_serving/minicpmo/README.md

@@ -23,17 +23,17 @@ including `librosa`.
 
 The deploy config auto-loads via `--omni`.
 The default `vllm_omni/deploy/minicpmo_4_5.yaml` keeps all three stages on
-logical device 0 with memory budgets of 55%, 22%, and 22%. The profile admits
-at most four sequences per stage and bounds startup video profiling to 32
-frames per video. For throughput,`minicpmo_4_5_batching.yaml` gives the Thinker
-GPU 0 (90%) and colocates the Talker (55%) and Code2Wav (35%) on GPU 1. That
-profile admits at most four concurrent sequences per stage.
+logical device 0 with memory budgets of 55%, 15%, and 18%, leaving headroom
+for runtime kernels such as the HiFi-GAN vocoder's cuDNN workspace. The
+profile admits at most four sequences per stage. For throughput,
+`minicpmo_4_5_2gpu.yaml` gives the Thinker GPU 0 (90%) and colocates the
+Talker (55%) and Code2Wav (35%) on GPU 1. That profile admits at most four
+concurrent sequences per stage.
 
 | deploy config | GPUs | Notes |
 |---|---|---|
 | `minicpmo_4_5.yaml` (default) | 1 | Memory-constrained compatibility layout. |
-| `minicpmo_4_5_batching.yaml` | 2 | Recommended continuous-batching layout; Talker and Code2Wav share GPU 1. |
-| `minicpmo_4_5_2gpu.yaml` | 2 | Backward-compatible alias for `minicpmo_4_5_batching.yaml`. |
+| `minicpmo_4_5_2gpu.yaml` | 2 | Recommended continuous-batching layout; Talker and Code2Wav share GPU 1. |
 | `minicpmo_4_5_3gpu.yaml` | 3 | One GPU per stage. |
 | `minicpmo_4_5_8x4090.yaml` | 8 | Full 8x4090 layout. |
 | `minicpmo_4_5_duplex.yaml` | 1 | Experimental native full-duplex overlay. |
@@ -89,17 +89,18 @@ python openai_chat_completion_client_for_multimodal_generation.py \
     --prompt "Briefly introduce yourself."
 ```
 
-Streaming text + audio:
+Streaming text + audio (use `--stream`):
 
 ```bash
-python streaming_chat_completion.py \
-    --base-url http://localhost:8099/v1 \
-    --output minicpmo_stream.wav
+python openai_chat_completion_client_for_multimodal_generation.py \
+    --query-type text \
+    --prompt "Briefly introduce yourself." \
+    --port 8099 \
+    --stream
 ```
 
-The example prints text deltas immediately and joins the independently encoded
-24 kHz WAV audio deltas into one valid WAV file. Add `--text-only` to skip
-Talker and Code2Wav.
+The client prints text deltas as they arrive and saves streamed audio chunks
+to WAV files.
 
 Shared helpers also work if you pass MiniCPM defaults yourself:
 

examples/online_serving/minicpmo/realtime_web/app/static/playback_worklet.js

@@ -6,6 +6,7 @@ class FullDuplexPcmPlayback extends AudioWorkletProcessor {
     this.playedFrames = 0;
     this.underrunFrames = 0;
     this.drain = null;
+    this.terminalFadeFrames = 0;
     this.started = false;
     this.activeResponseId = null;
     this.initialBufferFrames = Math.round(sampleRate * 0.2);
@@ -31,6 +32,7 @@ class FullDuplexPcmPlayback extends AudioWorkletProcessor {
       }
     } else if (message.type === 'drain') {
       this.drain = { responseId: message.responseId || null };
+      this.terminalFadeFrames = Math.min(this.fadeFrames, this.bufferedFrames());
       if (!this.started && this.bufferedFrames() > 0) {
         this.bufferWaitFrames = 0;
         this.startPlayback();
@@ -42,6 +44,7 @@ class FullDuplexPcmPlayback extends AudioWorkletProcessor {
       this.playedFrames = 0;
       this.underrunFrames = 0;
       this.drain = null;
+      this.terminalFadeFrames = 0;
       this.started = false;
       this.activeResponseId = null;
       this.bufferWaitFrames = this.initialBufferFrames;
@@ -76,6 +79,7 @@ class FullDuplexPcmPlayback extends AudioWorkletProcessor {
     this.playedFrames = 0;
     this.underrunFrames = 0;
     this.drain = null;
+    this.terminalFadeFrames = 0;
     this.started = false;
     this.activeResponseId = null;
     this.bufferWaitFrames = this.initialBufferFrames;
@@ -116,6 +120,7 @@ class FullDuplexPcmPlayback extends AudioWorkletProcessor {
     while (target < output.length && this.queue.length > 0) {
       const pcm = this.queue[0];
       const count = Math.min(output.length - target, pcm.length - this.offset);
+      const remainingBeforeChunk = this.bufferedFrames();
       for (let index = 0; index < count; index += 1) {
         let sample = pcm[this.offset + index] / 32768;
         if (this.fadeInFrames > 0) {
@@ -123,6 +128,16 @@ class FullDuplexPcmPlayback extends AudioWorkletProcessor {
           sample *= elapsed / this.fadeFrames;
           this.fadeInFrames -= 1;
         }
+        const remainingFrames = remainingBeforeChunk - index;
+        if (
+          this.drain
+          && this.terminalFadeFrames > 0
+          && remainingFrames <= this.terminalFadeFrames
+        ) {
+          sample *= this.terminalFadeFrames === 1
+            ? 0
+            : (remainingFrames - 1) / (this.terminalFadeFrames - 1);
+        }
         output[target + index] = sample;
       }
       target += count;

examples/online_serving/minicpmo/streaming_chat_completion.py

@@ -1,71 +0,0 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
-"""Stream MiniCPM-o 4.5 text and audio from /v1/chat/completions."""
-
-from __future__ import annotations
-
-import argparse
-import base64
-import io
-import wave
-from pathlib import Path
-
-from openai import OpenAI
-
-
-def append_wav_chunk(encoded: str, pcm_parts: list[bytes]) -> tuple[int, int, int]:
-    """Decode one base64 WAV delta and append its PCM frames."""
-    with wave.open(io.BytesIO(base64.b64decode(encoded)), "rb") as wav:
-        channels = wav.getnchannels()
-        sample_width = wav.getsampwidth()
-        sample_rate = wav.getframerate()
-        pcm_parts.append(wav.readframes(wav.getnframes()))
-    return channels, sample_width, sample_rate
-
-
-def main() -> None:
-    parser = argparse.ArgumentParser()
-    parser.add_argument("--base-url", default="http://localhost:8099/v1")
-    parser.add_argument("--model", default="openbmb/MiniCPM-o-4_5")
-    parser.add_argument("--prompt", default="Say hello, then introduce vLLM in one sentence.")
-    parser.add_argument("--output", type=Path, default=Path("minicpmo_stream.wav"))
-    parser.add_argument("--text-only", action="store_true")
-    args = parser.parse_args()
-
-    modalities = ["text"] if args.text_only else ["text", "audio"]
-    stream = OpenAI(base_url=args.base_url, api_key="EMPTY").chat.completions.create(
-        model=args.model,
-        messages=[{"role": "user", "content": args.prompt}],
-        modalities=modalities,
-        stream=True,
-        extra_body={"chat_template_kwargs": {"use_tts_template": not args.text_only}},
-    )
-
-    pcm_parts: list[bytes] = []
-    audio_format: tuple[int, int, int] | None = None
-    for chunk in stream:
-        for choice in chunk.choices:
-            content = getattr(choice.delta, "content", None)
-            if not content:
-                continue
-            if getattr(chunk, "modality", None) == "audio":
-                current_format = append_wav_chunk(content, pcm_parts)
-                if audio_format is not None and current_format != audio_format:
-                    raise ValueError(f"Audio format changed: {audio_format} -> {current_format}")
-                audio_format = current_format
-            elif getattr(chunk, "modality", None) == "text":
-                print(content, end="", flush=True)
-
-    print()
-    if pcm_parts and audio_format is not None:
-        channels, sample_width, sample_rate = audio_format
-        with wave.open(str(args.output), "wb") as wav:
-            wav.setnchannels(channels)
-            wav.setsampwidth(sample_width)
-            wav.setframerate(sample_rate)
-            wav.writeframes(b"".join(pcm_parts))
-        print(f"Audio saved to {args.output} ({sample_rate} Hz)")
-
-
-if __name__ == "__main__":
-    main()

examples/online_serving/text_to_video/README.md

@@ -248,12 +248,82 @@ done
 
 ## LTX-2
 
+### Start Server
+
+#### Basic Start
+
 ```bash
-vllm serve Lightricks/LTX-2 --omni --port 8098
+vllm serve Lightricks/LTX-2 --omni --port 8098 \
+    --enforce-eager --boundary-ratio 1.0
 ```
 
-See the [LTX-2 recipe](../../../recipes/LTX/LTX-2.md) for all checkpoints,
-pipeline selection, requests, defaults, and advanced options.
+#### Start with Optimization Presets
+
+Use the LTX-2 startup script with built-in optimization presets:
+
+```bash
+# Baseline (1 GPU, eager)
+bash run_server_ltx2.sh baseline
+
+# 4-GPU Ulysses sequence parallelism (lossless)
+bash run_server_ltx2.sh ulysses4
+
+# Cache-DiT lossy acceleration (1 GPU, ~1.4× speedup)
+bash run_server_ltx2.sh cache-dit
+
+# Best combo: 4-GPU Ulysses SP + Cache-DiT (~2.2× speedup)
+bash run_server_ltx2.sh best-combo
+```
+
+#### Optimization Benchmarks
+
+Benchmarked on H800, online serving (480×768, 41 frames, 20 steps, `seed=42`).
+"Inference" is the server-reported inference time; excludes HTTP/poll overhead.
+
+| Preset | Server Command | Inference (s) | Speedup | Type |
+|--------|---------------|---------------|---------|------|
+| `baseline` | `--enforce-eager` | 10.3 | 1.00× | — |
+| `compile` | *(default, no --enforce-eager)* | ~10.3 (warm) | ~1.00× | Lossless |
+| `ulysses4` | `--enforce-eager --usp 4` | ~10.3 | ~1.00× | Lossless |
+| `cache-dit` | `--enforce-eager --cache-backend cache_dit` | 7.4 avg | ~1.4× | Lossy |
+| `best-combo` | `--enforce-eager --usp 4 --cache-backend cache_dit` | 4.7 avg | **~2.2×** | Lossless + Lossy |
+
+**Observations**:
+- **torch.compile**: On H800, warm-request inference time matches the eager baseline (~10.3s).
+  The first request pays ~6s compilation overhead. Benefit depends on model architecture and GPU.
+- **Ulysses SP (4 GPU)**: No measurable speedup alone for 41-frame generation at this resolution.
+  Communication overhead outweighs gains at this sequence length.
+- **Cache-DiT**: Inference varies per request (6–10s) due to dynamic caching decisions.
+  Average is ~7.4s (~1.4× speedup) with slight quality tradeoff.
+- **Best combo**: 4-GPU Ulysses SP + Cache-DiT synergize well — Cache-DiT reduces per-step
+  computation, making the communication overhead of Ulysses SP worthwhile. Average ~4.7s
+  (~2.2× speedup).
+- **FP8 quantization**: Reduces VRAM but does not speed up LTX-2 on H800 (compute-bound).
+
+**Deployment Recommendations**:
+- For **production with quality priority**: use `baseline` with `--enforce-eager`
+- For **maximum throughput** (4 GPUs, quality tradeoff): use `best-combo` (~2.2× speedup)
+- For **single-GPU throughput**: use `cache-dit` (~1.4× speedup)
+- `--enforce-eager` is recommended to avoid torch.compile warmup latency on first request
+
+### Send Requests (curl)
+
+```bash
+# Using the provided script
+bash run_curl_ltx2.sh
+
+# Or directly
+curl -sS -X POST http://localhost:8098/v1/videos \
+  -H "Accept: application/json" \
+  -F "prompt=A serene lakeside sunrise with mist over the water." \
+  -F "width=768" \
+  -F "height=480" \
+  -F "num_frames=41" \
+  -F "fps=24" \
+  -F "num_inference_steps=20" \
+  -F "guidance_scale=3.0" \
+  -F "seed=42"
+```
 
 ## Helios
 

examples/online_serving/text_to_video/run_curl_ltx2.sh

@@ -22,6 +22,7 @@ create_response=$(
     -F "num_frames=41" \
     -F "fps=24" \
     -F "num_inference_steps=20" \
+    -F "guidance_scale=3.0" \
     -F "seed=42"
 )
 

examples/online_serving/text_to_video/run_server_ltx2.sh

@@ -21,6 +21,8 @@ set -euo pipefail
 
 MODEL="${MODEL:-Lightricks/LTX-2}"
 PORT="${PORT:-8098}"
+FLOW_SHIFT="${FLOW_SHIFT:-1.0}"
+BOUNDARY_RATIO="${BOUNDARY_RATIO:-1.0}"
 
 PRESET="${1:-baseline}"
 
@@ -64,13 +66,19 @@ case "$PRESET" in
         echo "Environment variables:"
         echo "  MODEL           - Model path (default: Lightricks/LTX-2)"
         echo "  PORT            - Server port (default: 8098)"
+        echo "  FLOW_SHIFT      - Scheduler flow shift (default: 1.0)"
+        echo "  BOUNDARY_RATIO  - Boundary ratio (default: 1.0)"
         exit 1
         ;;
 esac
 
 echo "Model: $MODEL"
 echo "Port: $PORT"
+echo "Flow shift: $FLOW_SHIFT"
+echo "Boundary ratio: $BOUNDARY_RATIO"
 
 vllm serve "$MODEL" --omni \
     --port "$PORT" \
+    --flow-shift "$FLOW_SHIFT" \
+    --boundary-ratio "$BOUNDARY_RATIO" \
     "${EXTRA_ARGS[@]}"

pyproject.toml

@@ -243,13 +243,16 @@ markers = [
     "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)
-    "H100: Tests that require H100 GPU",
-    "L4: Tests that require L4 GPU",
-    "MI325: Tests that require MI325 GPU (AMD/ROCm)",
-    "B60: Tests that require Intel Arc Pro B60 XPU",
-    "S5000: Tests that require S5000 GPU (Moore Threads/MUSA)",
-    "A2: Tests that require A2 NPU",
-    "A3: Tests that require A3 NPU",
+    # Tag with [hardware-resource] for tests.helpers.mark.get_hardware_mark_list().
+    "H100: [hardware-resource] Tests that require H100 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",

recipes/LTX/LTX-2.3.md

@@ -0,0 +1,154 @@
+# LTX-2.3
+
+> 22B parameter text-to-video + audio generation model served via vLLM-Omni
+
+## Summary
+
+- Vendor: Lightricks
+- Model: `diffusers/LTX-2.3-Diffusers`
+- Task: Text-to-video and image-to-video with synchronized audio generation
+- Mode: Online serving (pure diffusion)
+- Maintainer: @oglok
+
+## When to use this recipe
+
+Use this recipe when you want to serve LTX-2.3 for text-to-video or
+image-to-video generation with audio. The model generates videos up to 20+
+seconds at 768x512 resolution with 48kHz audio from text prompts, and can
+condition video generation on an initial image. Start with a 96GB-class GPU for
+validation because the model combines a 22B parameter transformer with text
+encoder, VAE, and vocoder components.
+
+## References
+
+- Upstream raw checkpoints: <https://huggingface.co/Lightricks/LTX-2.3>
+- Diffusers-format checkpoint: <https://huggingface.co/diffusers/LTX-2.3-Diffusers>
+- Requires `diffusers >= 0.38.0` (install from git: `pip install git+https://github.com/huggingface/diffusers.git`)
+
+## Serving
+
+### Command
+
+```bash
+vllm serve diffusers/LTX-2.3-Diffusers \
+  --omni \
+  --model-class-name LTX23Pipeline \
+  --stage-init-timeout 600
+```
+
+For image-to-video:
+
+```bash
+vllm serve diffusers/LTX-2.3-Diffusers \
+  --omni \
+  --model-class-name LTX23ImageToVideoPipeline \
+  --stage-init-timeout 600
+```
+
+### T2V Verification
+
+```bash
+# Health check
+curl http://localhost:8000/health
+
+# Generate a 3-second video (81 frames at 24fps)
+curl -X POST http://localhost:8000/v1/videos \
+  -F "prompt=A majestic bald eagle soaring over a misty mountain valley at dawn, golden sunlight breaking through clouds" \
+  -F "negative_prompt=blurry, low quality, distorted, watermark" \
+  -F "model=diffusers/LTX-2.3-Diffusers" \
+  -F "num_frames=81" \
+  -F "fps=24" \
+  -F "size=768x512" \
+  -F "num_inference_steps=30" \
+  -F "guidance_scale=4.0" \
+  -F "seed=42"
+
+# Generate a 10-second video (241 frames)
+curl -X POST http://localhost:8000/v1/videos \
+  -F "prompt=A cozy Japanese ramen shop at night in the rain, steam rising from bowls, neon signs reflecting on wet cobblestone streets" \
+  -F "model=diffusers/LTX-2.3-Diffusers" \
+  -F "num_frames=241" \
+  -F "fps=24" \
+  -F "size=768x512" \
+  -F "num_inference_steps=30" \
+  -F "guidance_scale=4.0"
+
+# Generate a 20-second video (481 frames)
+curl -X POST http://localhost:8000/v1/videos \
+  -F "prompt=An underwater coral reef teeming with tropical fish, sea turtles gliding gracefully, National Geographic documentary style" \
+  -F "model=diffusers/LTX-2.3-Diffusers" \
+  -F "num_frames=481" \
+  -F "fps=24" \
+  -F "size=768x512" \
+  -F "num_inference_steps=30" \
+  -F "guidance_scale=4.0"
+```
+
+### I2V Verification
+
+Start the server with `--model-class-name LTX23ImageToVideoPipeline`, then pass
+an image reference. The I2V pipeline requires `input_reference` or
+`image_reference` because the first frame is the conditioning input.
+
+```bash
+curl http://localhost:8000/health
+
+curl -X POST http://localhost:8000/v1/videos/sync \
+  -F "prompt=A plush toy astronaut gently waving while the camera slowly pushes in." \
+  -F "input_reference=@/absolute/path/to/reference.png" \
+  -F "model=diffusers/LTX-2.3-Diffusers" \
+  -F "num_frames=81" \
+  -F "fps=24" \
+  -F "size=768x512" \
+  -F "num_inference_steps=30" \
+  -F "guidance_scale=4.0" \
+  -F "seed=42" \
+  -o ltx23_i2v.mp4
+```
+
+### Notes
+
+- Memory usage: run a latest-head serving smoke or DFX benchmark on the target
+  GPU before publishing load/peak VRAM numbers.
+- Checkpoint format: use a Diffusers-format checkpoint such as
+  `diffusers/LTX-2.3-Diffusers`; the upstream `Lightricks/LTX-2.3` repository ships
+  raw safetensors and does not contain the subfolder configs required by this
+  pipeline loader.
+- Key flags:
+  - `--stage-init-timeout 600`: Gives the initial `torch.compile` warmup enough time on large checkpoints
+  - `--model-class-name LTX23Pipeline`: Selects the LTX-2.3 text-to-video pipeline (not LTX-2)
+  - `--model-class-name LTX23ImageToVideoPipeline`: Selects the LTX-2.3 image-to-video pipeline
+  - `input_reference=@...`: Uploads the I2V reference image directly
+- Audio: 48kHz AAC via BWE vocoder, automatically synced with video
+- CPU offloading: Text encoder (Gemma-3-12B), connectors, VAE, audio VAE, and vocoder stay on CPU and are moved to GPU only when needed
+- Supported resolutions: 768x512, 512x384 (must be divisible by 32)
+- Frame rate: 24 fps
+- Duration: Controlled by `num_frames` (frames = duration_seconds * 24 + 1)
+- Known limitations:
+  - LTX-2.3 I2V currently supports first-frame image conditioning, matching the LTX image-conditioning contract.
+  - Requires `diffusers >= 0.38.0` (not yet on PyPI, install from git)
+
+## Hardware Support
+
+## GPU
+
+### 1x NVIDIA RTX PRO 6000 Blackwell (96GB)
+
+#### Environment
+
+- OS: Ubuntu 22.04
+- Python: 3.10+
+- Driver / runtime: CUDA 13.0, Driver 580.126.09
+- vLLM version: Match the PR or release checkout being validated
+- vLLM-Omni version or commit: Use the commit you are deploying from
+
+### Benchmarking
+
+Do not copy latency or VRAM numbers between commits or machines. Record them
+only after running a latest-head sweep on the target hardware.
+
+For formal PR benchmarking, reuse
+`tests/dfx/perf/tests/test_ltx2_vllm_omni.json` with
+`tests/dfx/perf/scripts/run_diffusion_benchmark.py`. Each case in that file includes a JSON **`mark`** (`hardware_marks` + `full_model` + `diffusion`) for local `-m` filtering. That config captures the
+single-device eager baseline and CFG-parallel=2 case for a small 384x512,
+25-frame, 20-step workload.

recipes/LTX/LTX-2.md

@@ -1,303 +1,121 @@
-# LTX-2 Family
+# LTX-2
 
-> LTX-2 and LTX-2.3 text-to-video and image-to-video generation with synchronized audio
+> Text-to-video and image-to-video serving
 
-## Pipelines
+## Summary
 
-| Model | Checkpoint | Task | `--model-class-name` | Batching |
-|---|---|---|---|---|
-| LTX-2 | `Lightricks/LTX-2` | One-stage T2V/I2V | `LTX2Pipeline` | Yes |
-| LTX-2 distilled | `rootonchair/LTX-2-19b-distilled` | Two-stage T2V/I2V | `LTX2DistilledPipeline` | No |
-| LTX-2.3 | `diffusers/LTX-2.3-Diffusers` | One-stage T2V/I2V | `LTX2Pipeline` | Yes |
+- Vendor: Lightricks
+- Model: `Lightricks/LTX-2`
+- Task: Text-to-video generation and image-to-video generation
+- Mode: Online serving with the OpenAI-compatible video API
+- Maintainer: Community
 
-`LTX2Pipeline` is the unified one-stage entry. Checkpoint metadata selects the
-LTX-2 or LTX-2.3 profile; omitting an image selects T2V, while one initial
-image selects I2V. Both one-stage repositories declare this class, so
-`--model-class-name` is optional. LTX-2.3 requires the Diffusers checkpoint;
-the raw `Lightricks/LTX-2.3` safetensors repository is not directly loadable.
+## When to use this recipe
 
-`LTX2DistilledPipeline` is the unified distilled entry: omit `image` for T2V
-or provide one initial image for I2V. As in the official LTX pipelines,
-distilled mode is selected explicitly rather than inferred from the checkpoint
-name or metadata; always pass `--model-class-name LTX2DistilledPipeline`.
-
-## API Migration
-
-Only `req` may be passed positionally to `LTX2Pipeline`; every optional
-`forward` argument is keyword-only:
-
-```python
-# No longer supported
-pipe(req, prompt)
-pipe(req, image, prompt)
-
-# Supported
-pipe(req, prompt=prompt)
-pipe(req, image=image, prompt=prompt)
-```
-
-The consolidation also removes these registry names without aliases:
-
-| Removed name | Replacement |
-|---|---|
-| `LTX23Pipeline` | `LTX2Pipeline`; checkpoint metadata selects LTX-2.3 |
-| `LTX2ImageToVideoPipeline` | `LTX2Pipeline` with `image=` |
-| `LTX23ImageToVideoPipeline` | `LTX2Pipeline` with `image=`; checkpoint metadata selects LTX-2.3 |
-| `LTX2TwoStagesPipeline` | `LTX2DistilledPipeline` |
-| `LTX2ImageToVideoTwoStagesPipeline` | `LTX2DistilledPipeline` with `image=` |
-
-Passing any second positional argument now raises `TypeError`. These changes
-affect direct Python callers and explicit `--model-class-name` overrides;
-offline and serving entrypoints already use named fields and are unaffected.
-
-## One-Stage Defaults
-
-| Parameter | LTX-2 | LTX-2.3 |
-|---|---:|---:|
-| Width × height | 768 × 512 | 768 × 512 |
-| Frames / frame rate | 121 / 24 | 121 / 24 |
-| Denoise steps | 40 | 30 |
-| Video/audio CFG | 3.0 / 7.0 | 3.0 / 7.0 |
-| Video/audio STG | 1.0 / 1.0 | 1.0 / 1.0 |
-| Video/audio modality guidance | 3.0 / 3.0 | 3.0 / 3.0 |
-| Video/audio rescale | 0.7 / 0.7 | 0.7 / 0.7 |
-| Video/audio STG blocks | `[29]` / `[29]` | `[28]` / `[28]` |
-
-The model recipe also supplies the default negative prompt. Top-level
-`guidance_scale` overrides both video and audio CFG values. The online video
-API defaults `num_frames` to `1`, so set it explicitly; the offline LTX scripts
-default to `121`.
-
-## Distilled Two-Stage Defaults
-
-| Parameter | Value |
-|---|---:|
-| Final width × height | 1536 × 1024 |
-| Stage 1 width × height | 768 × 512 |
-| Frames / frame rate | 121 / 24 |
-| Stage 1 / Stage 2 denoise steps | 8 / 3 |
-| Guidance | Fixed positive-only |
-
-The dimensions passed to the API are the final output dimensions and must be
-divisible by 64. Both one- and two-stage requests require `num_frames = 8k+1`.
-The distilled recipe fixes both sigma schedules; `num_inference_steps` must be
-`8` when supplied.
-
-## Serving
-
-Start either checkpoint:
-
-```bash
-vllm serve Lightricks/LTX-2 --omni --stage-init-timeout 600
-```
-
-```bash
-vllm serve diffusers/LTX-2.3-Diffusers --omni --stage-init-timeout 600
-```
-
-Start the distilled checkpoint with its two-stage entry:
-
-```bash
-vllm serve rootonchair/LTX-2-19b-distilled --omni \
-  --model-class-name LTX2DistilledPipeline --stage-init-timeout 600
-```
-
-The same server handles T2V and I2V. A T2V request using the selected recipe's
-default guidance is:
-
-```bash
-curl -X POST http://localhost:8000/v1/videos/sync \
-  -F "prompt=A cinematic close-up of ocean waves at golden hour." \
-  -F "negative_prompt=worst quality, inconsistent motion, blurry, jittery, distorted" \
-  -F "size=768x512" \
-  -F "num_frames=121" \
-  -F "fps=24" \
-  -F "seed=42" \
-  -o ltx_t2v.mp4
-```
-
-For I2V, provide exactly one initial image:
-
-```bash
-curl -X POST http://localhost:8000/v1/videos/sync \
-  -F "prompt=A plush toy astronaut gently waving while the camera slowly pushes in." \
-  -F "negative_prompt=worst quality, inconsistent motion, blurry, jittery, distorted" \
-  -F "input_reference=@/absolute/path/to/reference.png" \
-  -F "size=768x512" \
-  -F "num_frames=121" \
-  -F "fps=24" \
-  -F "seed=42" \
-  -o ltx_i2v.mp4
-```
-
-Use `image_reference` for a URL or JSON-safe image reference. Do not provide it
-together with `input_reference`.
-
-## Guidance
-
-LTX supports independent video/audio CFG, spatio-temporal guidance (STG),
-cross-modality guidance, and rescaling:
-
-| Parameter | Default | Effect | Alias |
-|---|---:|---|---|
-| `video_cfg_scale` | 3.0 | Video text CFG; `1.0` disables it | `video_cfg_guidance_scale` |
-| `audio_cfg_scale` | 7.0 | Audio text CFG; `1.0` disables it | `audio_cfg_guidance_scale` |
-| `video_stg_scale` | 1.0 | Video STG; `0.0` disables it | `video_stg_guidance_scale` |
-| `audio_stg_scale` | 1.0 | Audio STG; `0.0` disables it | `audio_stg_guidance_scale` |
-| `video_modality_scale` | 3.0 | Audio-to-video guidance; `1.0` disables it | `a2v_guidance_scale` |
-| `audio_modality_scale` | 3.0 | Video-to-audio guidance; `1.0` disables it | `v2a_guidance_scale` |
-| `video_rescale_scale` | 0.7 | Video guidance rescale; `0.0` disables it | — |
-| `audio_rescale_scale` | 0.7 | Audio guidance rescale; `0.0` disables it | — |
-| `video_stg_blocks` | `[29]` / `[28]` | Perturbed video transformer blocks | — |
-| `audio_stg_blocks` | `[29]` / `[28]` | Perturbed audio transformer blocks | — |
-
-The STG block defaults are shown as LTX-2 / LTX-2.3. A canonical name wins
-over its alias. Explicit top-level `guidance_scale` wins over both CFG fields;
-omit it for independent video/audio CFG.
-
-Pass these fields in online `extra_params` or offline `--extra-body`:
-
-```bash
-# Add to the curl request above
--F 'extra_params={"video_cfg_scale":3.0,"audio_cfg_scale":7.0}'
-
-# Add to text_to_video.py or image_to_video.py
---extra-body '{"video_cfg_scale":3.0,"audio_cfg_scale":7.0}'
-```
-
-### Python API
-
-Normal per-request values belong in `OmniDiffusionSamplingParams`. Pass the
-resulting `DiffusionRequestBatch` as the only positional argument:
-
-```python
-from vllm_omni.diffusion.request import OmniDiffusionRequest
-from vllm_omni.diffusion.worker.request_batch import DiffusionRequestBatch
-from vllm_omni.inputs.data import OmniDiffusionSamplingParams
-
-params = OmniDiffusionSamplingParams(
-    width=768,
-    height=512,
-    num_frames=121,
-    frame_rate=24.0,
-    num_inference_steps=30,
-    seed=42,
-    extra_args={"video_cfg_scale": 3.0, "audio_cfg_scale": 7.0},
-)
-req = DiffusionRequestBatch([
-    OmniDiffusionRequest(
-        prompt={"prompt": "Cherry blossoms moving in a light breeze"},
-        sampling_params=params,
-        request_id="ltx-example",
-    )
-])
-
-t2v_output = pipe(req)
-i2v_output = pipe(req, image=image)
-```
-
-Direct keyword arguments are low-level fallbacks, for example
-`pipe(req, prompt=prompt, width=768)`. Request fields take precedence unless
-noted below.
-
-### Complete `forward` Surface
-
-| Argument | Type/default | Meaning and constraints |
-|---|---|---|
-| `req` | `DiffusionRequestBatch`, required | Only positional argument; contains prompts and per-request sampling parameters. |
-| `image` | image or batch, `None` | Direct value wins over request images; no image selects T2V. I2V accepts one image per prompt, and a batch cannot mix T2V/I2V. |
-| `prompt` | string or list, `None` | Positive-text fallback; request prompts win. Mutually exclusive with `prompt_embeds`. |
-| `negative_prompt` | string or list, `None` | Fallback after request values and before the recipe default; mutually exclusive with negative embeddings. |
-| `height` | `int`, `None` | Request → direct value → recipe default; divisible by 32 one-stage or 64 distilled. |
-| `width` | `int`, `None` | Same precedence and alignment as `height`. |
-| `num_frames` | `int`, `None` | Request → direct value → recipe default; must be `8k+1` and also determines audio duration with `frame_rate`. |
-| `frame_rate` | `float`, `None` | Request `frame_rate` → request `fps` → direct value → recipe default. |
-| `num_inference_steps` | `int`, `None` | Request → direct value → recipe default; minimum 2 one-stage, fixed at 8 for distilled Stage 1. Custom one-stage `sigmas` determine actual steps. |
-| `sigmas` | list of float, `None` | One-stage only. Request values win; every request in a fused batch must use the same schedule. |
-| `timesteps` | list of int, `None` | Compatibility slot; LTX accepts only `None`. Use `sigmas`. |
-| `guidance_scale` | `float`, `None` | One-stage common video/audio CFG fallback; an explicit request value wins. Distilled two-stage uses fixed positive-only guidance. |
-| `guidance_rescale` | `float`, `None` | Accepts only `None` or `0.0`; use the modality rescale fields. |
-| `noise_scale` | `float`, `0.0` | Compatibility slot; LTX accepts only `0.0`. |
-| `num_videos_per_prompt` | `int`, `1` | Output-count fallback; positive request `num_outputs_per_prompt` wins. |
-| `generator` | generator or list, `None` | Explicit RNG; otherwise request generators/seeds are collated. Lists must match the effective output batch. |
-| `latents` | tensor, `None` | Request tensors win. One-stage accepts packed `[B, S, C]` and validated unpacked video layouts. |
-| `audio_latents` | tensor, `None` | One-stage initial audio latents; request tensors win and are collated. |
-| `prompt_embeds` | tensor, `None` | Precomputed positive conditioning; requires `prompt_attention_mask` and cannot accompany `prompt`. |
-| `negative_prompt_embeds` | tensor, `None` | One-stage precomputed negative conditioning; requires its mask and cannot accompany `negative_prompt`. |
-| `prompt_attention_mask` | tensor, `None` | Mask for positive embeddings; request `prompt_attention_mask` or `attention_mask` wins. |
-| `negative_prompt_attention_mask` | tensor, `None` | Mask for negative embeddings; request negative mask fields win. |
-| `decode_timestep` | float or list, `0.0` | Video-VAE decode timestep; request value wins. Lists may match 1, prompt batch, or output batch. |
-| `decode_noise_scale` | float or list, `None` | Same list rules; request value wins. Defaults to `decode_timestep`. |
-| `output_type` | string, `"np"` | Request value wins. `"np"` decodes output; `"latent"` skips VAE/vocoder decode. |
-| `return_dict` | `bool`, `True` | Compatibility slot; only `True` is accepted. |
-| `attention_kwargs` | dict, `None` | Per-call values are unsupported; configure attention at engine startup. |
-| `max_sequence_length` | `int`, `None` | Request → direct value → tokenizer limit. |
-
-Request-object naming differs only in a few places: `num_videos_per_prompt`
-maps to `num_outputs_per_prompt`; images and prompt text/embeddings live in the
-request prompt payload; LTX guidance fields live in sampling `extra_args`.
-
-### Recipe-Specific Request Capabilities
-
-| Override | One-stage | Distilled two-stage |
-|---|---|---|
-| Guidance | Supported | Fixed positive-only |
-| Negative prompt/embeddings | Supported | Rejected |
-| `num_inference_steps` | Supported | Fixed at 8 for Stage 1; Stage 2 uses 3 |
-| Custom `sigmas` | Supported | Rejected; both phases use fixed schedules |
-| Video/audio latents | Supported | Rejected |
-
-These capability checks apply equally to direct `forward` keywords and
-values in `OmniDiffusionSamplingParams`; unsupported values fail instead of
-being silently ignored.
-
-### Custom Sigma Schedules
-
-One-stage Python requests may set final scheduler boundaries directly:
-
-```python
-params = OmniDiffusionSamplingParams(sigmas=[1.0, 0.75, 0.5, 0.25])
-```
-
-Each nonterminal value produces one denoise step, and a terminal `0.0` is
-appended when omitted. This schedule overrides `num_inference_steps`. All
-requests in a fused batch must use the same list. The video form API and
-bundled offline CLI do not currently expose `sigmas`.
-
-### Constraints
-
-- LTX rejects non-default `timesteps`, `flow_shift`, `guidance_rescale`,
-  `noise_scale`, `attention_kwargs`, and `return_dict=False`. Use final
-  `sigmas`, modality rescale fields, startup attention configuration, and the
-  standard `DiffusionOutput` respectively.
-- Fused one-stage requests must resolve to identical LTX guidance and sigma
-  schedules. Keep concurrent requests guidance-homogeneous or use
-  `--max-num-seqs 1`.
-- `--cfg-parallel-size 2` supports only the CFG-only plan: disable STG and
-  modality guidance and set both rescale fields to `0.0`.
-
-## Operational Notes
-
-- LTX-2 one-stage previously loaded and peaked at about 73.5 GiB on one H200
-  141GB; remeasure on the deployed commit and hardware.
-- LTX-2.3 includes a 22B transformer, Gemma encoder, two VAEs, and a vocoder.
-  Start on a 96GB-class GPU or use CPU/layerwise offload on smaller devices.
-- The distilled upsampler uses the configured pipeline dtype and participates
-  in component discovery/device placement; it remains resident during denoise
-  offload because it is used only at the phase boundary.
-- The output audio sample rate comes from the loaded components and is not a
-  request parameter.
-- For benchmarks, use `tests/dfx/perf/tests/test_ltx2_vllm_omni.json` with
-  `tests/dfx/perf/scripts/run_diffusion_benchmark.py`.
-- Current two-stage support is limited to the distilled checkpoint; official
-  LTX two-stage/HQ execution remains out of scope.
+Use this recipe when you want to deploy the base `Lightricks/LTX-2` model
+for both text-to-video and image-to-video generation. The same model ID is used for
+both tasks; switch the pipeline class with `--model-class-name` depending on
+whether you want T2V or I2V.
 
 ## References
 
-- <https://huggingface.co/Lightricks/LTX-2>
-- <https://huggingface.co/Lightricks/LTX-2.3>
-- <https://huggingface.co/diffusers/LTX-2.3-Diffusers>
-- [Online video generation](../../docs/user_guide/examples/online_serving/text_to_video.md)
-- [Request batching](../../docs/user_guide/diffusion/request_batching.md)
-- [T2V offline example](../../examples/offline_inference/text_to_video/text_to_video.md)
-- [I2V offline example](../../examples/offline_inference/image_to_video/README.md)
+- Model: <https://huggingface.co/Lightricks/LTX-2>
+- User guide:
+  [`docs/user_guide/examples/online_serving/text_to_video.md`](../../docs/user_guide/examples/online_serving/text_to_video.md)
+- Example guide:
+  [`examples/online_serving/text_to_video/README.md`](../../examples/online_serving/text_to_video/README.md)
+  [`examples/online_serving/image_to_video/README.md`](../../examples/online_serving/image_to_video/README.md)
+
+## Hardware Support
+
+## GPU
+
+### 1x H200 141GB (Text-to-Video)
+
+#### Environment
+
+- OS: Ubuntu 22.04.5 LTS
+- Python: 3.12+
+- Driver / runtime: NVIDIA CUDA environment with H200 141GB GPU
+- vLLM version: Match the repository requirements from your current checkout
+- vLLM-Omni version or commit: Use the commit you are deploying from
+
+#### Command
+
+```bash
+vllm serve Lightricks/LTX-2 \
+  --omni \
+  --model-class-name LTX2Pipeline
+```
+
+#### Verification
+
+```bash
+curl http://localhost:8000/health
+
+curl -X POST http://localhost:8000/v1/videos \
+  -F "prompt=A cinematic close-up of ocean waves at golden hour." \
+  -F "negative_prompt=worst quality, inconsistent motion, blurry, jittery, distorted" \
+  -F "width=512" \
+  -F "height=768" \
+  -F "num_frames=121" \
+  -F "fps=24" \
+  -F "num_inference_steps=40" \
+  -F "guidance_scale=4.0" \
+  -o ltx2_t2v_h200.mp4
+```
+
+#### Notes
+
+- Memory usage: Model loads at ~73.5GB, peaks at ~73.5 GiB during inference.
+- Key flags:
+  - `--model-class-name LTX2Pipeline` selects the text-to-video pipeline explicitly.
+- Known limitations:
+  - This recipe is tested on 1x H200 validation; add measured latency and VRAM after running it on your target machine.
+
+### 1x H200 141GB (Image-to-Video)
+
+#### Environment
+
+- OS: Ubuntu 22.04.5 LTS
+- Python: 3.12+
+- Driver / runtime: NVIDIA CUDA environment with H200 141GB GPU
+- vLLM version: Match the repository requirements from your current checkout
+- vLLM-Omni version or commit: Use the commit you are deploying from
+
+#### Command
+
+```bash
+vllm serve Lightricks/LTX-2 \
+  --omni \
+  --model-class-name LTX2ImageToVideoPipeline
+```
+
+#### Verification
+
+```bash
+curl http://localhost:8000/health
+
+curl -X POST http://localhost:8000/v1/videos/sync \
+  -F "prompt=A plush toy astronaut gently waving while the camera slowly pushes in." \
+  -F "input_reference=@/absolute/path/to/reference.png" \
+  -F "width=512" \
+  -F "height=768" \
+  -F "num_frames=121" \
+  -F "fps=24" \
+  -F "num_inference_steps=40" \
+  -F "guidance_scale=4.0" \
+  -o ltx2_i2v_h200.mp4
+```
+
+#### Notes
+
+- Memory usage: Model loads at ~73.5GB, peaks at ~73.5 GiB during inference.
+- Key flags:
+  - `--model-class-name LTX2ImageToVideoPipeline` switches the same `Lightricks/LTX-2` weights into image-to-video mode.
+  - `input_reference=@...` uploads the reference image directly
+- Known limitations:
+  - Use `image_reference` when you want to pass a URL or JSON-safe image reference
+instead of uploading a file.
+  - Do not send `input_reference` and `image_reference` in the same request.
+  - This recipe is tested on 1x H200 validation; add measured latency and VRAM after running it on your target machine.

recipes/NVIDIA/Nemotron-Labs-Audex.md

@@ -1,142 +0,0 @@
-# Nemotron-Labs Audex (2B / 30B-A3B)
-
-> Speech suite on 1×H100: TTS, text-to-audio, audio understanding, and cascaded speech-to-speech
-
-## Summary
-
-- Vendor: NVIDIA
-- Model: `nvidia/Nemotron-Labs-Audex-2B`, `nvidia/Nemotron-Labs-Audex-30B-A3B`
-- Task: Text-to-speech (English), text-to-audio (sound effects), audio understanding (ASR / audio QA), cascaded speech-to-speech
-- Mode: Online serving (`/v1/audio/speech`, `/v1/chat/completions`) and offline batch inference
-- Maintainer: Community
-
-## When to use this recipe
-
-Use this recipe to serve either Audex checkpoint through one of its four
-deployment pipelines. Both sizes share the same repo layout, codec token
-space, and streaming causal speech decoder (16 kHz output); they differ in
-the thinker: the 2B is a dense LM, the 30B-A3B is a hybrid Mamba + MoE
-NemotronH (~3B active parameters). Pick the pipeline by task:
-
-| pipeline (`vllm_omni/deploy/<name>[_30b].yaml`) | audio in | text out | speech out | general audio out | endpoint |
-|---|---|---|---|---|---|
-| `audex_tts` | ❌ | ❌ | ✅ | ❌ | `/v1/audio/speech` |
-| `audex_tta` | ❌ | ❌ | ❌ | ✅ | `/v1/audio/speech` |
-| `audex_thinker_only` | ✅ | ✅ | ❌ | ❌ | `/v1/chat/completions` |
-| `audex_s2s` | ✅ | ✅ | ✅ | ❌ | both |
-
-## References
-
-- Model cards: [Nemotron-Labs-Audex-2B](https://huggingface.co/nvidia/Nemotron-Labs-Audex-2B), [Nemotron-Labs-Audex-30B-A3B](https://huggingface.co/nvidia/Nemotron-Labs-Audex-30B-A3B)
-- Offline examples: [`examples/offline_inference/audex/`](../../examples/offline_inference/audex/)
-- Online examples: [`examples/online_serving/audex/`](../../examples/online_serving/audex/)
-
-## Hardware Support
-
-## GPU
-
-### 1×H100 80GB — Audex-2B
-
-#### Environment
-
-- OS: Linux
-- Python: 3.12+
-- CUDA: 12.x
-- vLLM version: 0.24.0
-- vLLM-Omni version or commit: PR #4976 (branch `audex`)
-
-#### Command
-
-**Online serving** (TTS is the default pipeline; the repo root's
-`model_type` auto-resolves `audex_tts.yaml`):
-
-```bash
-vllm-omni serve nvidia/Nemotron-Labs-Audex-2B \
-    --host 0.0.0.0 --port 8097 \
-    --trust-remote-code --omni
-# other modes: add --stage-configs-path vllm_omni/deploy/audex_{tta,thinker_only,s2s}.yaml
-# or use the launcher: MODE=s2s PORT=8098 examples/online_serving/audex/run_server.sh
-```
-
-**Offline batch inference** (one script per task, each defaults to its
-correct deploy yaml):
-
-```bash
-python examples/offline_inference/audex/text_to_speech.py \
-    --texts "Hello world." --cfg-scale 1.5 --output-dir results/audex_wavs
-python examples/offline_inference/audex/text_to_audio.py     # sound effects
-python examples/offline_inference/audex/audio_qa.py          # ASR / audio QA
-python examples/offline_inference/audex/speech_to_speech.py  # 3-pass cascade
-```
-
-#### Verification
-
-```bash
-curl -X POST http://localhost:8097/v1/audio/speech \
-    -H "Content-Type: application/json" \
-    -d '{
-        "model": "nvidia/Nemotron-Labs-Audex-2B",
-        "input": "Hello, how are you?",
-        "response_format": "wav",
-        "extra_params": {"cfg_scale": 1.5}
-    }' --output hello.wav
-# or the universal client (all four modes):
-python examples/online_serving/audex/client.py --mode tts --output hello.wav
-```
-
-#### Notes
-
-- Memory usage: TTS deploy — stage 0 (thinker) 0.4, stage 1 (decoder) 0.25
-  GPU-memory fraction; both stages share GPU 0.
-- Key flags: `--trust-remote-code` and `--omni` are required.
-- Classifier-free guidance: `extra_params.cfg_scale` (1.0 = off; 1.5 is the
-  official quality setting; TTA effectively requires its default 3.0).
-  Measured en-24 self-transcribed CER: 6.87% guided vs 7.24% unguided.
-- Output: 16 kHz mono WAV. English, single built-in voice (no cloning).
-- TTA decodes through the external XCodec1 checkpoint
-  (`hf-audio/xcodec-hubert-general-balanced`, auto-downloaded; override
-  with `XCODEC1_PATH`).
-
-### 1×H100 80GB — Audex-30B-A3B
-
-#### Environment
-
-Same as the 2B row above.
-
-#### Command
-
-The 30B REQUIRES its explicit deploy yaml (the shared HF `model_type`
-auto-resolves to the 2B-tuned config otherwise):
-
-```bash
-vllm-omni serve nvidia/Nemotron-Labs-Audex-30B-A3B \
-    --host 0.0.0.0 --port 8097 \
-    --trust-remote-code --omni \
-    --stage-configs-path vllm_omni/deploy/audex_tts_30b.yaml
-# or: SIZE=30b MODE=tts examples/online_serving/audex/run_server.sh
-```
-
-Offline: same four scripts with
-`--model nvidia/Nemotron-Labs-Audex-30B-A3B --deploy-config vllm_omni/deploy/audex_<mode>_30b.yaml`.
-
-#### Verification
-
-Same curl / `client.py` as the 2B with
-`"model": "nvidia/Nemotron-Labs-Audex-30B-A3B"` (the server validates the
-model id). First launch downloads ~60 GB.
-
-#### Notes
-
-- Memory usage: ~61 GiB weights on one H100 80 GB (stage 0 at 0.85 GPU
-  fraction + decoder at 0.08 on the same card); verified with healthy KV
-  headroom. If long sequences OOM, set `tensor_parallel_size: 2` on
-  stage 0 (documented in the yaml headers).
-- `enable_prefix_caching: false` on stage 0 — hybrid Mamba constraint.
-- CUDA graphs stay ON for the thinker: the NemotronH stage captures
-  cleanly and eager mode costs ~7× single-stream TTS RTF (review-measured
-  0.30 vs 2.08).
-- Quality (en-24 self-transcribed CER): 0.72% unguided baseline; 0.72% at
-  cfg 1.5 with guided temperature 0.1.
-- Known limitation: the S2S chat pass needs the official
-  formatting-instruction prompt (the bundled examples/client send it);
-  a bare transcript turn makes the 30B answer in speech-codec tokens.

recipes/OpenBMB/MiniCPM-o-4_5.md

@@ -31,9 +31,8 @@ also provided.
   - Default single-GPU compatibility layout (auto-loaded):
     [`vllm_omni/deploy/minicpmo_4_5.yaml`](../../vllm_omni/deploy/minicpmo_4_5.yaml)
   - Recommended 2-GPU continuous-batching layout:
-    [`vllm_omni/deploy/minicpmo_4_5_batching.yaml`](../../vllm_omni/deploy/minicpmo_4_5_batching.yaml)
-  - 2-GPU and 3-GPU layouts:
-    [`vllm_omni/deploy/minicpmo_4_5_2gpu.yaml`](../../vllm_omni/deploy/minicpmo_4_5_2gpu.yaml),
+    [`vllm_omni/deploy/minicpmo_4_5_2gpu.yaml`](../../vllm_omni/deploy/minicpmo_4_5_2gpu.yaml)
+  - 3-GPU layout:
     [`vllm_omni/deploy/minicpmo_4_5_3gpu.yaml`](../../vllm_omni/deploy/minicpmo_4_5_3gpu.yaml)
   - 8x RTX 4090 layout:
     [`vllm_omni/deploy/minicpmo_4_5_8x4090.yaml`](../../vllm_omni/deploy/minicpmo_4_5_8x4090.yaml)
@@ -68,8 +67,8 @@ Code2Wav consumes them through a shared-memory async connector.
 MiniCPM-o 4.5 now requires the three-stage topology: the Talker owns
 request-local codec generation and Code2Wav owns waveform state and
 reference-voice prompt features. `minicpmo_4_5.yaml` remains the stable
-single-GPU entry point; `minicpmo_4_5_batching.yaml` is the recommended
-two-GPU profile. The removed fused two-stage implementation is not retained as
+single-GPU entry point; `minicpmo_4_5_2gpu.yaml` is the recommended
+two-GPU continuous-batching profile. The removed fused two-stage implementation is not retained as
 a fallback because it would duplicate state machines and correctness paths.
 
 ## GPU
@@ -79,9 +78,9 @@ a fallback because it would duplicate state machines and correctness paths.
 The default
 [`vllm_omni/deploy/minicpmo_4_5.yaml`](../../vllm_omni/deploy/minicpmo_4_5.yaml)
 co-locates Thinker, codec-only Talker, and Code2Wav on GPU 0. Their
-`gpu_memory_utilization` budgets are 0.55, 0.22, and 0.22. All stages admit
-up to four sequences. Startup video profiling is bounded to 32 frames per
-video. Use the two-GPU profile for production throughput.
+`gpu_memory_utilization` budgets are 0.55, 0.15, and 0.18 so concurrent
+Code2Wav activations keep residual headroom. All stages admit up to four
+sequences. Use the two-GPU profile for production throughput.
 
 #### Environment
 
@@ -105,7 +104,7 @@ The deploy config is auto-loaded by the model registry — no
 For the recommended two-GPU layout, add:
 
 ```bash
---deploy-config vllm_omni/deploy/minicpmo_4_5_batching.yaml
+--deploy-config vllm_omni/deploy/minicpmo_4_5_2gpu.yaml
 ```
 
 #### Performance comparison
@@ -175,16 +174,18 @@ in another choice's `message.audio.data` (24 kHz mono, see Notes). With
 `modalities: ["text", "audio"]` you typically get two `choices` entries
 (one text, one audio).
 
-**Streaming text + speech**:
+**Streaming text + speech** (use `--stream`):
 
 ```bash
-python examples/online_serving/minicpmo/streaming_chat_completion.py \
-    --base-url http://localhost:8099/v1 \
-    --output minicpmo_stream.wav
+python examples/online_serving/minicpmo/openai_chat_completion_client_for_multimodal_generation.py \
+    --query-type text \
+    --prompt "Say hello, then introduce vLLM in one sentence." \
+    --port 8099 \
+    --stream
 ```
 
-The client prints text deltas as they arrive and reconstructs one valid WAV
-from the independently encoded audio deltas.
+The client prints text deltas as they arrive and saves streamed audio chunks
+to WAV files.
 
 **Gradio demo (text + image + audio + video UI)**:
 
@@ -202,8 +203,8 @@ speech output (TTS)"** checkbox on / off.
 
 #### Notes
 
-- Memory budget: Thinker, Talker, and Code2Wav reserve 0.55, 0.22, and
-  0.22 of GPU 0. The larger Thinker share protects its multimodal KV cache;
+- Memory budget: Thinker, Talker, and Code2Wav reserve 0.55, 0.15, and
+  0.18 of GPU 0 so concurrent Code2Wav activations keep residual headroom;
   all three model processes still share one CUDA device.
 - `--trust-remote-code` is required — the HF repo ships a custom
   `MiniCPMO` config / model class.
@@ -214,14 +215,13 @@ speech output (TTS)"** checkbox on / off.
   contention. Talker AR
   state and Code2Wav caches are request-owned; Code2Wav batches only
   exact-shape-compatible chunks and does not fall back to serial decode.
-- `limit_mm_per_prompt.video.num_frames: 32` bounds startup dummy profiling,
-  not runtime media decoding. Use `media_io_kwargs.video.num_frames` when a
-  matching URL/file video sampling limit is required.
+- Stage 0 pins `interleave_mm_strings=true` and
+  `media_io_kwargs.video={fps:1,num_frames:128}` for Daily-Omni MiniCPM packs.
 - `StageRequestStats.batch_size` is a request-scoped placeholder, not the
   scheduler's execution batch.
 - Single-GPU co-location trades throughput for hardware density: Stage 0/1
   CUDA Graph replay and eager Stage 2 vocoder kernels compete across three
-  CUDA contexts. Use the 8x4090 config or a custom multi-GPU mapping for
+  CUDA contexts. Use `minicpmo_4_5_2gpu.yaml` or a custom multi-GPU mapping for
   throughput-sensitive serving.
 
 ### 8 x RTX 4090 24GB (consumer-GPU layout)

recipes/README.md

@@ -46,7 +46,8 @@ recipes/
 | [`inclusionAI/Ming-omni-tts.md`](./inclusionAI/Ming-omni-tts.md) | Offline + online dense Ming TTS/audio generation | 1x H100 80GB / 1x AMD MI300X (ROCm 7.2) |
 | [`IndexTeam/IndexTTS-2.md`](./IndexTeam/IndexTTS-2.md) | Online serving for voice-cloned TTS with optional emotion control | 1x L4 24GB or larger CUDA GPU |
 | [`krea/Krea-2.md`](./krea/Krea-2.md) | Text-to-image (Turbo + Raw), offline + online, with LoRA | 1x H100 80GB |
-| [`LTX/LTX-2.md`](./LTX/LTX-2.md) | LTX-2/LTX-2.3 text-to-video and image-to-video with synchronized audio | H200 141GB / 96GB-class GPU |
+| [`LTX/LTX-2.md`](./LTX/LTX-2.md) | Text-to-video and image-to-video serving | 1x H200 141GB |
+| [`LTX/LTX-2.3.md`](./LTX/LTX-2.3.md) | Text-to-video with audio generation (22B) | 1x GPU (96GB VRAM) |
 | [`MammothModa2/MammothModa2.md`](./MammothModa2/MammothModa2.md) | Preview and Dev text-to-image (AR → DiT); Dev text/image understanding | Preview: 1x L40S 48GB / 1x ≥40GB GPU; Dev: 1x NVIDIA GPU with sufficient cache headroom |
 | [`mistralai/Voxtral-TTS.md`](./mistralai/Voxtral-TTS.md) | Online serving for TTS | 1x RTX 4090 24GB |
 | [`Robbyant/LingBot-Video.md`](./Robbyant/LingBot-Video.md) | Native dense and MoE text-to-video serving | Dense: 1x L20X; MoE: 1x L20X (~67.7 GiB peak) |
@@ -56,7 +57,6 @@ recipes/
 | [`OpenBMB/MiniCPM-o-4_5.md`](./OpenBMB/MiniCPM-o-4_5.md) | Online serving for omni multimodal chat (text / image / audio / video → text + 24 kHz speech) | 2x A100/H100 80GB / 3x mid-tier GPU / 8x RTX 4090 24GB |
 | [`OpenBMB/VoxCPM2.md`](./OpenBMB/VoxCPM2.md) | Online + offline TTS with native AR pipeline (48 kHz, 30+ languages) | 1x RTX 4090 24GB |
 | [`OpenMOSS/MOSS-TTS.md`](./OpenMOSS/MOSS-TTS.md) | Online + offline multilingual TTS (MOSS-TTS family, 8B) | 1x H100 80GB |
-| [`NVIDIA/Nemotron-Labs-Audex.md`](./NVIDIA/Nemotron-Labs-Audex.md) | TTS / text-to-audio / audio understanding / cascaded S2S (Audex 2B + 30B-A3B) | 1x H100 80GB |
 | [`Qwen/Qwen-Image.md`](./Qwen/Qwen-Image.md) | Text-to-image serving with step-wise continuous batching replay and ModelOpt mixed FP8/NVFP4 | 1x A100 80GB / 2x B200 |
 | [`Qwen/Qwen-Image-2512.md`](./Qwen/Qwen-Image-2512.md) | Text-to-image serving with step-wise continuous batching replay and ModelOpt FP8 / mixed FP8/NVFP4 | 1x A800 80GB / 2x B200 |
 | [`Qwen/Qwen-Image-Edit.md`](./Qwen/Qwen-Image-Edit.md) | Text-guided single-image editing | 1x or 2x H200 141GB |

recipes/Tencent/HunyuanImage-3.0-Instruct.md

@@ -164,10 +164,11 @@ curl -s http://localhost:8091/v1/chat/completions \
 PR [#2495](https://github.com/vllm-project/vllm-omni/pull/2495) adds
 performance CI configs for the same DiT-only settings. The CI step is
 currently opt-in (gated by `RUN_HUNYUAN_IMAGE3_PERF=1`) with `soft_fail`
-enabled, intended for initial data collection. Performance assertions are
-skipped (`skip-performance-assertion: true`); the baseline values in the
-JSON configs are reference-only and will be promoted to regression gates
-once enough nightly data has been collected.
+enabled, intended for initial data collection. Perf runners no longer
+compare metrics against JSON baselines at assert time; the `baseline`
+blocks in the configs are reference-only and are copied into result JSON
+for reporting. Regression gates can be reintroduced once enough nightly
+data has been collected.
 
 The user-facing equivalent is to launch one of the CLI commands above and
 generate 1024x1024 images with 50 denoising steps.

tests/attention/test_fish_kvcache_attn.py

@@ -100,9 +100,8 @@ def _decode_kv_cache(
     head_size: int = 128,
     dtype: torch.dtype = torch.float16,
 ) -> torch.Tensor:
-    # vLLM >=0.23.0 KV cache layout: (num_blocks, num_kv_heads, block_size,
-    # 2*head_size) with K and V packed in the last dimension.
-    return torch.zeros((num_blocks, num_kv_heads, block_size, 2 * head_size), dtype=dtype)
+    # vLLM >=0.23.0 KV cache layout: key/value live on dim 1.
+    return torch.zeros((num_blocks, 2, block_size, num_kv_heads, head_size), dtype=dtype)
 
 
 def test_fish_kvcache_enabled_by_default(monkeypatch):
@@ -222,11 +221,10 @@ def _vllm_kv_cache_shape(num_blocks, block_size, num_kv_heads, head_size):
 
 
 def test_fish_kvcache_backend_unbinds_kv_on_vllm_cache_layout(monkeypatch):
-    # vLLM >=0.23.0 lays out the KV cache as (num_blocks, num_kv_heads,
-    # block_size, 2*head_size) with K and V packed in the last dimension.
-    # The backend must split on the content dim so each of key/value comes
-    # back as the 4-D (num_blocks, block_size, num_kv_heads, head_size)
-    # tensor the decode kernel expects.
+    # vLLM >=0.23.0 lays the KV cache out as (num_blocks, 2, block_size,
+    # num_kv_heads, head_size). The backend must unbind on dim 1 so each of
+    # key/value comes back as the 4-D (num_blocks, block_size, num_kv_heads,
+    # head_size) tensor the decode kernel expects.
     monkeypatch.setenv("VLLM_OMNI_FISH_KVCACHE_ATTN", "1")
     monkeypatch.setattr(fish_kvcache_backend, "is_available", lambda: True)
     monkeypatch.setattr(fish_kvcache_backend, "load_error", lambda: None)
@@ -234,8 +232,7 @@ def test_fish_kvcache_backend_unbinds_kv_on_vllm_cache_layout(monkeypatch):
 
     num_blocks, block_size, num_kv_heads, head_size = 8, 16, 8, 128
     cache_shape = _vllm_kv_cache_shape(num_blocks, block_size, num_kv_heads, head_size)
-    # Upstream now returns a 4-D packed shape: (B, H, N, 2*D).
-    assert cache_shape == (num_blocks, num_kv_heads, block_size, 2 * head_size)
+    assert cache_shape == (num_blocks, 2, block_size, num_kv_heads, head_size)
 
     captured = {}
 
@@ -265,21 +262,20 @@ def test_fish_kvcache_backend_unbinds_kv_on_vllm_cache_layout(monkeypatch):
 def test_fish_kvcache_vllm_cache_unbinds_then_falls_back_until_contiguous(monkeypatch):
     # End-to-end with the real vLLM cache layout and the real (unmocked) guard.
     #
-    # The point of the transpose+split fix: unbind(1) on the old 5-D layout
-    # raised "too many values to unpack" on the vLLM 0.23.0 layout, and the
-    # upstream now packs K/V into the last dimension (4-D shape). The
-    # transpose(1, 2) + split correctly unpacks into separate key/value tensors
-    # and the backend gracefully falls back to the original attention with
-    # correct output. This test guards that fix and must NOT depend on the fast
-    # path firing.
+    # The point of the unbind(1) fix: unbind(0) raised "too many values to
+    # unpack" on the 5-D vLLM 0.23.0 layout, so the request crashed outright.
+    # unbind(1) makes the request *work* again -- it unpacks cleanly and the
+    # backend gracefully falls back to the original attention with correct
+    # output. This test guards that fix and must NOT depend on the fast path
+    # firing.
     #
-    # It also documents a known limitation (tracked separately):
-    # transpose(1, 2) on the interleaved (B, H, N, 2*D) layout yields
-    # *non-contiguous* key/value views, which the guard rejects on the
-    # is_contiguous() check. So the dimensions are compatible (4-D,
-    # block_size=16, head_size=128) -- the only thing keeping the fast path
-    # from engaging is contiguity, not shape. Making the fast path actually
-    # fire would require a .contiguous() copy and is out of scope here.
+    # It also documents a known limitation (tracked separately): unbind(1) on
+    # the interleaved (num_blocks, 2, ...) layout yields *non-contiguous*
+    # key/value views, which the guard rejects on the is_contiguous() check.
+    # So the dimensions are compatible (4-D, block_size=16, head_size=128) --
+    # the only thing keeping the fast path from engaging is contiguity, not
+    # shape. Making the fast path actually fire would require a .contiguous()
+    # copy and is out of scope here.
     fish_kvcache_backend.reset_fish_kvcache_attn_stats()
     monkeypatch.setenv("VLLM_OMNI_FISH_KVCACHE_ATTN", "1")
     monkeypatch.setattr(fish_kvcache_backend, "is_available", lambda: True)
@@ -304,14 +300,13 @@ def test_fish_kvcache_vllm_cache_unbinds_then_falls_back_until_contiguous(monkey
     output = torch.zeros_like(query)
     kv_cache = torch.zeros(cache_shape, dtype=torch.float16)
 
-    # Root cause record: transpose(1,2) + split gives the right 4-D shape but
-    # a non-contiguous view.
-    key_cache, value_cache = kv_cache.transpose(1, 2).split(head_size, dim=-1)
+    # Root cause record: unbind(1) gives the right 4-D shape but a non-contiguous view.
+    key_cache, value_cache = kv_cache.unbind(1)
     assert tuple(key_cache.shape) == (num_blocks, block_size, num_kv_heads, head_size)
     assert not key_cache.is_contiguous()
 
-    # transpose+split no longer crashes (unbind(1) would have raised here) and
-    # the request completes via the correct fallback path.
+    # unbind(1) no longer crashes (unbind(0) would have raised here) and the
+    # request completes via the correct fallback path.
     result = model.layers[0].self_attn.attn.impl.forward(None, query, None, None, kv_cache, _metadata(), output)
 
     assert result is output

tests/benchmarks/patch/test_patch.py

@@ -6,13 +6,21 @@ Unit tests for patch.py
 """
 
 import asyncio
+import base64
 import json
+import time
+from types import SimpleNamespace
 
 import pytest
 from pytest_mock import MockerFixture
 from vllm.benchmarks.lib.endpoint_request_func import RequestFuncInput
 
-from vllm_omni.benchmarks.patch.patch import MixRequestFuncOutput, async_request_openai_chat_omni_completions
+from vllm_omni.benchmarks.patch.patch import (
+    MixRequestFuncOutput,
+    async_request_openai_chat_omni_completions,
+    async_request_openai_realtime_duplex,
+)
+from vllm_omni.experimental.fullduplex.client import RealtimeEventCollector
 
 pytestmark = [pytest.mark.core_model, pytest.mark.benchmark, pytest.mark.cpu]
 
@@ -40,6 +48,156 @@ class MockResponse:
         pass
 
 
+@pytest.mark.asyncio
+async def test_seed_tts_realtime_duplex_exports_per_request_metrics(monkeypatch):
+    class FakeRealtimeClient:
+        last_instance = None
+
+        def __init__(self, url):
+            assert url == "ws://localhost:8000/v1/realtime?duplex=1"
+            self.events = RealtimeEventCollector()
+            self.configure_kwargs = None
+            self.sent = []
+            self.response_count = 0
+            self.ack_count = 0
+            type(self).last_instance = self
+
+        async def __aenter__(self):
+            return self
+
+        async def __aexit__(self, *_args):
+            return None
+
+        async def configure(self, model, **kwargs):
+            assert model == "openbmb/MiniCPM-o-4_5"
+            self.configure_kwargs = kwargs
+
+        async def send(self, event):
+            self.sent.append(event)
+            if event["type"] != "response.create":
+                return
+            self.response_count += 1
+            response_id = f"resp-{self.response_count}"
+            now = time.monotonic()
+            self.events.add(
+                {"type": "response.created", "response": {"id": response_id}},
+                received_at_s=now + 0.01,
+            )
+            self.events.add(
+                {
+                    "type": "response.audio_transcript.delta",
+                    "response_id": response_id,
+                    "delta": f"turn {self.response_count}",
+                },
+                received_at_s=now + 0.02,
+            )
+            self.events.add(
+                {
+                    "type": "response.audio.delta",
+                    "response_id": response_id,
+                    "delta": base64.b64encode(b"\x00\x00" * 2400).decode(),
+                    "sample_rate_hz": 24_000,
+                    "metadata": {"audio_duration_ms": 100},
+                },
+                received_at_s=now + 0.03,
+            )
+            self.events.add(
+                {"type": "response.done", "response": {"id": response_id}},
+                received_at_s=now + 0.04,
+            )
+
+        async def acknowledge_playback(self):
+            self.ack_count += 1
+
+        async def close_session(self, **_kwargs):
+            return None
+
+    monkeypatch.setattr(
+        "vllm_omni.benchmarks.patch.patch.RealtimeDuplexClient",
+        FakeRealtimeClient,
+    )
+    request_input = RequestFuncInput(
+        model="openbmb/MiniCPM-o-4_5",
+        model_name="openbmb/MiniCPM-o-4_5",
+        prompt="hello",
+        api_url="http://localhost:8000/v1/realtime",
+        prompt_len=1,
+        output_len=20,
+        logprobs=None,
+        multi_modal_content=None,
+        ignore_eos=False,
+        extra_body={"save_duplex_request_metrics": True},
+    )
+    request_input.seed_tts_speech_extra = {"ref_audio": "data:audio/wav;base64,AAAA"}
+    request_input.seed_tts_system_prompt = "Speak exactly."
+    request_input.seed_tts_turns = tuple(
+        SimpleNamespace(utterance_id=f"utt-{index}", target_text=f"text {index}") for index in range(4)
+    )
+
+    output = await async_request_openai_realtime_duplex(
+        request_input,
+        session=None,
+    )
+
+    client = FakeRealtimeClient.last_instance
+    assert client.configure_kwargs["native_duplex"] is False
+    assert client.configure_kwargs["extra_body"] == {
+        "ref_audio": "data:audio/wav;base64,AAAA",
+    }
+    assert client.sent == [
+        event
+        for index in range(4)
+        for event in (
+            {
+                "type": "conversation.item.create",
+                "item": {
+                    "type": "message",
+                    "role": "user",
+                    "content": [{"type": "input_text", "text": f"text {index}"}],
+                },
+            },
+            {"type": "response.create"},
+        )
+    ]
+    assert client.ack_count == 4
+    assert output.success is True
+    assert output.generated_text == "turn 1 turn 2 turn 3 turn 4"
+    assert output.audio_duration == pytest.approx(0.4)
+    assert output.ttft > 0
+    assert output.audio_ttfp > output.ttft
+    assert output.audio_rtf > 0
+    assert output.latency > 0
+    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"]
+    assert len(output.duplex_request_metrics) == 4
+    for index, metrics in enumerate(output.duplex_request_metrics):
+        assert metrics == {
+            "session_id": session_id,
+            "request_index": index,
+            "utterance_id": f"utt-{index}",
+            "response_id": f"resp-{index + 1}",
+            "source": "client_monotonic_receive",
+            "measurement_origin": {
+                "ttft": "conversation.item.create client send to first non-empty text delta",
+                "ttfp": "conversation.item.create client send to first audio packet",
+                "rtf": "request-start-to-last-audio receive time divided by emitted audio duration",
+            },
+            "ttft_ms": pytest.approx(20.0, abs=2.0),
+            "ttfp_ms": pytest.approx(30.0, abs=2.0),
+            "rtf": pytest.approx(0.3, abs=0.03),
+            "audio_generation_ms": pytest.approx(30.0, abs=2.0),
+            "audio_duration_ms": 100.0,
+        }
+    assert output.duplex_session_metrics == {
+        "session_id": session_id,
+        "audio_turn_count": 4,
+        "mean_ttft_ms": pytest.approx(20.0, abs=2.0),
+        "mean_ttfp_ms": pytest.approx(30.0, abs=2.0),
+        "mean_rtf": pytest.approx(0.3, abs=0.03),
+    }
+
+
 def create_sse_chunk(data_dict):
     """Helper to create SSE formatted chunk"""
     return f"data: {json.dumps(data_dict)}\n\n".encode()

tests/benchmarks/test_accuracy_bench_utils.py

@@ -42,7 +42,10 @@ from benchmarks.accuracy.text_to_image.gbench import (
     summarize_generated_records as summarize_gebench_generated_records,
     summarize_gebench_results,
 )
-from tests.e2e.accuracy.qwen3_omni.qwen3_omni_acc_bench_core import seed_tts_bench_argv
+from tests.e2e.accuracy.qwen3_omni.qwen3_omni_acc_bench_core import (
+    build_serve_common_argv,
+    seed_tts_bench_argv,
+)
 from tests.e2e.accuracy.qwen3_omni.run_qwen_omni_acc_benchmark import sync_dataset_env_from_ns
 from vllm_omni.benchmarks.data_modules.seed_tts_dataset import resolve_seed_tts_root
 
@@ -57,6 +60,25 @@ def test_seed_tts_bench_argv_preserves_hf_repo_id_from_env(monkeypatch):
     assert argv[dataset_idx + 1] == "zhaochenyang20/seed-tts-eval"
 
 
+def test_build_serve_common_argv_accepts_realtime_seed_tts_route(tmp_path):
+    argv = build_serve_common_argv(
+        host="127.0.0.1",
+        port=8000,
+        model="openbmb/MiniCPM-o-4_5",
+        num_prompts=4,
+        max_concurrency=1,
+        num_warmups=0,
+        percentile_metrics="ttft,audio_ttfp,audio_rtf",
+        result_dir=tmp_path,
+        result_filename="result.json",
+        backend="openai-realtime-tts",
+        endpoint="/v1/realtime",
+    )
+
+    assert argv[argv.index("--backend") + 1] == "openai-realtime-tts"
+    assert argv[argv.index("--endpoint") + 1] == "/v1/realtime"
+
+
 def test_sync_dataset_env_preserves_seed_tts_hf_repo_id(monkeypatch):
     ns = argparse.Namespace(
         daily_omni_repo=None,

tests/benchmarks/test_daily_omni_local_source.py

@@ -0,0 +1,152 @@
+# SPDX-License-Identifier: Apache-2.0
+"""Offline Daily-Omni sources: local mirrors must not fall back to the Hub id."""
+
+import json
+import os
+import tarfile
+from pathlib import Path
+
+import pytest
+
+from vllm_omni.benchmarks.data_modules.daily_omni_dataset import (
+    DailyOmniDataset,
+    daily_omni_local_qa_json,
+    daily_omni_local_videos_dir,
+    ensure_daily_omni_hub_root,
+    resolve_daily_omni_local_root,
+)
+
+pytestmark = [pytest.mark.core_model, pytest.mark.cpu, pytest.mark.benchmark]
+
+_HUB_ID = "liarliar/Daily-Omni"
+
+
+def _write_qa_json(root: Path) -> Path:
+    qa = root / "qa.json"
+    qa.write_text(json.dumps([{"Question": "q", "Answer": "A", "video_id": "vid0"}]), encoding="utf-8")
+    return qa
+
+
+def _stage_offline_hub_cache(tmp_path: Path, monkeypatch: pytest.MonkeyPatch, *, with_qa: bool = True) -> Path:
+    """Build a hub cache holding ``liarliar/Daily-Omni`` and cut off all network access.
+
+    Reproduces the air-gapped NPU runner: the dataset is pre-staged with ``hf download``, so it
+    only exists in the *hub* cache -- never in the ``datasets`` cache that ``load_dataset`` reads.
+    """
+    constants = pytest.importorskip("huggingface_hub").constants
+
+    cache = tmp_path / "hub"
+    repo_dir = cache / "datasets--liarliar--Daily-Omni"
+    snapshot = repo_dir / "snapshots" / "deadbeef"
+    snapshot.mkdir(parents=True)
+    if with_qa:
+        _write_qa_json(snapshot)
+    (repo_dir / "refs").mkdir()
+    (repo_dir / "refs" / "main").write_text("deadbeef", encoding="utf-8")
+
+    monkeypatch.setattr(constants, "HF_HUB_CACHE", str(cache))
+    monkeypatch.setattr(constants, "HF_HUB_OFFLINE", True)
+    return snapshot
+
+
+def test_resolve_local_root_returns_none_for_hub_id() -> None:
+    assert resolve_daily_omni_local_root("liarliar/Daily-Omni") is None
+    assert resolve_daily_omni_local_root("") is None
+    assert resolve_daily_omni_local_root(None) is None
+
+
+def test_plain_local_mirror(tmp_path: Path) -> None:
+    _write_qa_json(tmp_path)
+    (tmp_path / "Videos" / "vid0").mkdir(parents=True)
+
+    root = resolve_daily_omni_local_root(str(tmp_path))
+    assert root == tmp_path.resolve()
+    assert daily_omni_local_qa_json(root) == (tmp_path / "qa.json").resolve()
+    assert daily_omni_local_videos_dir(root) == root / "Videos"
+
+
+def test_hf_cache_dir_resolves_to_snapshot(tmp_path: Path) -> None:
+    cache_dir = tmp_path / "datasets--liarliar--Daily-Omni"
+    snapshot = cache_dir / "snapshots" / "deadbeef"
+    snapshot.mkdir(parents=True)
+    _write_qa_json(snapshot)
+    (cache_dir / "refs").mkdir()
+    (cache_dir / "refs" / "main").write_text("deadbeef", encoding="utf-8")
+
+    root = resolve_daily_omni_local_root(str(cache_dir))
+    assert root == snapshot.resolve()
+    assert daily_omni_local_qa_json(root) == (snapshot / "qa.json").resolve()
+
+
+def test_hf_cache_dir_without_refs_picks_newest_snapshot(tmp_path: Path) -> None:
+    cache_dir = tmp_path / "datasets--liarliar--Daily-Omni"
+    older = cache_dir / "snapshots" / "aaa"
+    newer = cache_dir / "snapshots" / "bbb"
+    older.mkdir(parents=True)
+    newer.mkdir(parents=True)
+    os.utime(older, (1, 1))
+    os.utime(newer, (10, 10))
+
+    assert resolve_daily_omni_local_root(str(cache_dir)) == newer.resolve()
+
+
+def test_videos_tar_is_extracted(tmp_path: Path, monkeypatch: pytest.MonkeyPatch) -> None:
+    monkeypatch.setenv("HF_HOME", str(tmp_path / "hf_home"))
+    root = tmp_path / "mirror"
+    (root / "Videos" / "vid0").mkdir(parents=True)
+    (root / "Videos" / "vid0" / "vid0_video.mp4").write_bytes(b"fake")
+    with tarfile.open(root / "Videos.tar", "w") as tf:
+        tf.add(root / "Videos", arcname="Videos")
+    # Only the tarball is left in the mirror.
+    for child in (root / "Videos").rglob("*"):
+        if child.is_file():
+            child.unlink()
+    (root / "Videos" / "vid0").rmdir()
+    (root / "Videos").rmdir()
+
+    videos_dir = daily_omni_local_videos_dir(root)
+    assert videos_dir is not None
+    assert (videos_dir / "vid0" / "vid0_video.mp4").is_file()
+
+    # Second call reuses the extraction instead of redoing it.
+    assert daily_omni_local_videos_dir(root) == videos_dir
+
+
+def test_missing_qa_json_reports_none(tmp_path: Path) -> None:
+    root = resolve_daily_omni_local_root(str(tmp_path))
+    assert daily_omni_local_qa_json(root) is None
+    assert daily_omni_local_videos_dir(root) is None
+
+
+def test_hub_id_resolves_to_offline_cache(tmp_path: Path, monkeypatch: pytest.MonkeyPatch) -> None:
+    snapshot = _stage_offline_hub_cache(tmp_path, monkeypatch)
+
+    root = ensure_daily_omni_hub_root(_HUB_ID)
+
+    assert root == snapshot.resolve()
+    assert daily_omni_local_qa_json(root) == (snapshot / "qa.json").resolve()
+
+
+def test_dataset_loads_hub_id_without_network(tmp_path: Path, monkeypatch: pytest.MonkeyPatch) -> None:
+    """The regression this file exists for: a Hub id must not require ``load_dataset``."""
+    snapshot = _stage_offline_hub_cache(tmp_path, monkeypatch)
+    # Any use of `datasets` here means we went back to the path that cannot read the hub cache.
+    monkeypatch.setattr("vllm_omni.benchmarks.data_modules.daily_omni_dataset.load_dataset", None)
+
+    dataset = DailyOmniDataset(dataset_path=_HUB_ID, random_seed=0)
+
+    assert dataset.qa_json_path == (snapshot / "qa.json").resolve()
+    assert [row["video_id"] for row in dataset.data] == ["vid0"]
+
+
+def test_uncached_hub_id_reports_both_attempts(tmp_path: Path, monkeypatch: pytest.MonkeyPatch) -> None:
+    _stage_offline_hub_cache(tmp_path, monkeypatch, with_qa=False)
+    monkeypatch.setattr("vllm_omni.benchmarks.data_modules.daily_omni_dataset.load_dataset", None)
+
+    with pytest.raises(RuntimeError) as excinfo:
+        DailyOmniDataset(dataset_path="someone-else/Daily-Omni", random_seed=0)
+
+    message = str(excinfo.value)
+    assert "someone-else/Daily-Omni" in message
+    assert "snapshot_download failed" in message
+    assert "--daily-omni-qa-json" in message

tests/benchmarks/test_daily_omni_pack_modes.py

@@ -0,0 +1,237 @@
+"""Tests for Daily-Omni ``--daily-omni-pack-mode`` packing recipes.
+
+The ``minicpm-interleave`` mode is an accuracy-critical protocol: MiniCPM-o only reaches
+OpenBMB's reported Daily-Omni score when frames and 1s audio segments arrive strictly
+interleaved, with ``max_slice_nums=1`` / ``use_image_id=False`` and no Qwen system prompt.
+These tests pin that contract so a refactor cannot silently regress the score.
+
+vllm stubs are installed by tests/benchmarks/conftest.py before collection.
+"""
+
+from __future__ import annotations
+
+import importlib.util
+import json
+import sys
+from pathlib import Path
+
+import numpy as np
+import pytest
+
+pytestmark = [pytest.mark.core_model, pytest.mark.cpu]
+
+# Load the data module directly (bypasses vllm_omni.__init__ heavy imports).
+_REPO_ROOT = Path(__file__).resolve().parents[2]
+_MODULE_PATH = _REPO_ROOT / "vllm_omni" / "benchmarks" / "data_modules" / "daily_omni_dataset.py"
+_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)
+    _mod = importlib.util.module_from_spec(_spec)
+    sys.modules[_MODULE_NAME] = _mod
+    _spec.loader.exec_module(_mod)
+
+from vllm_omni.benchmarks.data_modules.daily_omni_dataset import (  # noqa: E402
+    DailyOmniDataset,
+    _uniform_sample_indices,
+)
+
+_VIDEO_ID = "vid001"
+_NUM_FRAMES = 5
+
+
+@pytest.fixture
+def video_dir(tmp_path: Path) -> Path:
+    """Official Daily-Omni ``Videos/`` layout with placeholder media files."""
+    per_video = tmp_path / "Videos" / _VIDEO_ID
+    per_video.mkdir(parents=True)
+    (per_video / f"{_VIDEO_ID}_video.mp4").write_bytes(b"fake-mp4")
+    (per_video / f"{_VIDEO_ID}_audio.wav").write_bytes(b"fake-wav")
+    return tmp_path / "Videos"
+
+
+@pytest.fixture
+def qa_json(tmp_path: Path) -> Path:
+    path = tmp_path / "qa.json"
+    path.write_text(
+        json.dumps(
+            [
+                {
+                    "video_id": _VIDEO_ID,
+                    "question": "What happens first?",
+                    "choice": {"A": "a", "B": "b", "C": "c", "D": "d"},
+                    "answer": "A",
+                    "task_type": "AV Event Alignment",
+                    "video_duration": "30s",
+                }
+            ]
+        ),
+        encoding="utf-8",
+    )
+    return path
+
+
+def _make_dataset(qa_json: Path, video_dir: Path, **kwargs) -> DailyOmniDataset:
+    return DailyOmniDataset(
+        qa_json_path=str(qa_json),
+        video_dir=str(video_dir),
+        disable_shuffle=True,
+        **kwargs,
+    )
+
+
+def _stub_extract(monkeypatch, dataset: DailyOmniDataset, counter: list[int]) -> None:
+    """Replace decord/ffmpeg extraction with deterministic in-memory frames + segments."""
+    from PIL import Image
+
+    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 = [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)
+
+
+# ---------------------------------------------------------------------------
+# _uniform_sample_indices
+# ---------------------------------------------------------------------------
+
+
+@pytest.mark.parametrize(("n", "k"), [(10, 4), (100, 64), (7, 7), (3, 5), (1, 1)])
+def test_uniform_sample_indices_matches_numpy_linspace(n: int, k: int) -> None:
+    got = _uniform_sample_indices(n, k)
+    if n <= k:
+        assert got == list(range(n))
+    else:
+        assert got == [int(i) for i in np.linspace(0, n - 1, k, dtype=int)]
+
+
+def test_uniform_sample_indices_non_positive_k() -> None:
+    assert _uniform_sample_indices(10, 0) == []
+    assert _uniform_sample_indices(10, -1) == []
+
+
+# ---------------------------------------------------------------------------
+# minicpm-interleave packing
+# ---------------------------------------------------------------------------
+
+
+def test_minicpm_interleave_alternates_image_and_audio(monkeypatch, qa_json, video_dir) -> None:
+    ds = _make_dataset(qa_json, video_dir, pack_mode="minicpm-interleave", input_mode="all")
+    _stub_extract(monkeypatch, ds, [])
+
+    parts, extra, position = ds._compose_daily_omni_multimodal(_VIDEO_ID, None)
+
+    assert position == "first"
+    assert [p["type"] for p in parts] == ["image_url", "audio_url"] * _NUM_FRAMES
+    assert extra["mm_processor_kwargs"] == {
+        "use_audio_in_video": False,
+        "max_slice_nums": 1,
+        "use_image_id": False,
+    }
+
+
+def test_minicpm_interleave_visual_mode_emits_frames_only(monkeypatch, qa_json, video_dir) -> None:
+    ds = _make_dataset(qa_json, video_dir, pack_mode="minicpm-interleave", input_mode="visual")
+    _stub_extract(monkeypatch, ds, [])
+
+    parts, _extra, _position = ds._compose_daily_omni_multimodal(_VIDEO_ID, None)
+
+    assert [p["type"] for p in parts] == ["image_url"] * _NUM_FRAMES
+
+
+def test_minicpm_interleave_writes_file_urls_under_video_dir(monkeypatch, qa_json, video_dir) -> None:
+    ds = _make_dataset(qa_json, video_dir, pack_mode="minicpm-interleave", input_mode="all")
+    _stub_extract(monkeypatch, ds, [])
+
+    parts, _extra, _position = ds._compose_daily_omni_multimodal(_VIDEO_ID, None)
+
+    segment_dir = video_dir / ".minicpm_daily_omni_interleave" / _VIDEO_ID
+    for part in parts:
+        url = part[part["type"]]["url"]
+        assert url.startswith("file://")
+        assert Path(url[len("file://") :]).parent == segment_dir.resolve()
+    assert sorted(p.name for p in segment_dir.glob("frame_*.jpg")) == [f"frame_{i:04d}.jpg" for i in range(_NUM_FRAMES)]
+
+
+def test_minicpm_interleave_inline_mode_emits_data_urls(monkeypatch, qa_json, video_dir) -> None:
+    ds = _make_dataset(
+        qa_json,
+        video_dir,
+        pack_mode="minicpm-interleave",
+        input_mode="all",
+        inline_local_video=True,
+    )
+    _stub_extract(monkeypatch, ds, [])
+
+    parts, _extra, _position = ds._compose_daily_omni_multimodal(_VIDEO_ID, None)
+
+    assert parts[0]["image_url"]["url"].startswith("data:image/jpeg;base64,")
+    assert parts[1]["audio_url"]["url"].startswith("data:audio/wav;base64,")
+    assert not (video_dir / ".minicpm_daily_omni_interleave").exists()
+
+
+def test_minicpm_interleave_caches_file_parts_but_not_inline_payloads(monkeypatch, qa_json, video_dir) -> None:
+    """Inline base64 must not be retained: ~700 Daily-Omni videos would pin GBs of RSS."""
+    file_calls: list[int] = []
+    ds_file = _make_dataset(qa_json, video_dir, pack_mode="minicpm-interleave", input_mode="all")
+    _stub_extract(monkeypatch, ds_file, file_calls)
+    ds_file._compose_daily_omni_multimodal(_VIDEO_ID, None)
+    ds_file._compose_daily_omni_multimodal(_VIDEO_ID, None)
+    assert len(file_calls) == 1
+    assert ds_file._minicpm_interleave_cache
+
+    inline_calls: list[int] = []
+    ds_inline = _make_dataset(
+        qa_json,
+        video_dir,
+        pack_mode="minicpm-interleave",
+        input_mode="all",
+        inline_local_video=True,
+    )
+    _stub_extract(monkeypatch, ds_inline, inline_calls)
+    ds_inline._compose_daily_omni_multimodal(_VIDEO_ID, None)
+    ds_inline._compose_daily_omni_multimodal(_VIDEO_ID, None)
+    assert len(inline_calls) == 2
+    assert not ds_inline._minicpm_interleave_cache
+
+
+def test_minicpm_interleave_missing_local_video_returns_none(qa_json, tmp_path) -> None:
+    empty = tmp_path / "empty_videos"
+    empty.mkdir()
+    ds = _make_dataset(qa_json, empty, pack_mode="minicpm-interleave", input_mode="all")
+
+    parts, extra, _position = ds._compose_daily_omni_multimodal(_VIDEO_ID, None)
+
+    assert parts is None
+    assert extra is None
+
+
+# ---------------------------------------------------------------------------
+# System prompt selection
+# ---------------------------------------------------------------------------
+
+
+def test_minicpm_interleave_omits_qwen_system_message(monkeypatch, qa_json, video_dir) -> None:
+    ds = _make_dataset(qa_json, video_dir, pack_mode="minicpm-interleave", input_mode="all")
+    _stub_extract(monkeypatch, ds, [])
+    parts, _extra, _position = ds._compose_daily_omni_multimodal(_VIDEO_ID, None)
+
+    messages = ds._build_daily_omni_openai_messages(parts, "q?", {"A": "a"})
+
+    assert [m["role"] for m in messages] == ["user"]
+
+
+def test_qwen_pack_mode_keeps_system_message(qa_json, video_dir) -> None:
+    ds = _make_dataset(qa_json, video_dir, pack_mode="qwen", input_mode="all")
+
+    messages = ds._build_daily_omni_openai_messages([], "q?", {"A": "a"})
+
+    assert [m["role"] for m in messages] == ["system", "user"]
+    assert "Qwen" in messages[0]["content"][0]["text"]
+
+
+def test_invalid_pack_mode_rejected(qa_json, video_dir) -> None:
+    with pytest.raises(ValueError, match="pack_mode must be"):
+        _make_dataset(qa_json, video_dir, pack_mode="bogus")

tests/benchmarks/test_seed_tts_dataset_variants.py

@@ -28,10 +28,13 @@ if _MODULE_NAME not in sys.modules:
     _spec.loader.exec_module(_mod)
 
 from vllm_omni.benchmarks.data_modules.seed_tts_dataset import (  # noqa: E402
+    SeedTTSDataset,
     SeedTTSDesignDataset,
     SeedTTSDesignSampleRequest,
+    SeedTTSSampleRequest,
     SeedTTSTextDataset,
     SeedTTSTextSampleRequest,
+    SeedTTSTurn,
 )
 
 # ---------------------------------------------------------------------------
@@ -70,6 +73,94 @@ def mock_tokenizer(mocker):
 # ---------------------------------------------------------------------------
 
 
+def test_seed_tts_dataset_groups_four_turns_with_one_reference(seed_tts_root, mock_tokenizer):
+    ds = SeedTTSDataset(
+        dataset_path=str(seed_tts_root),
+        random_seed=0,
+        locale="en",
+        disable_shuffle=True,
+    )
+
+    requests = ds.sample(
+        mock_tokenizer,
+        num_requests=1,
+        turns_per_session=4,
+        no_oversample=True,
+    )
+
+    assert len(requests) == 1
+    request = requests[0]
+    assert [turn.utterance_id for turn in request.seed_tts_turns] == [
+        "utt000",
+        "utt001",
+        "utt002",
+        "utt003",
+    ]
+    assert [turn.target_text for turn in request.seed_tts_turns] == [
+        "target text 0",
+        "target text 1",
+        "target text 2",
+        "target text 3",
+    ]
+    assert request.seed_tts_speech_extra["ref_text"] == "ref text 0"
+    assert request.seed_tts_ref_wav_path.endswith("utt000.wav")
+
+
+def test_seed_tts_eval_expands_grouped_pcm_by_turn():
+    from vllm_omni.benchmarks.data_modules.seed_tts_eval import (
+        _expand_seed_tts_turn_outputs,
+    )
+
+    request = SeedTTSSampleRequest(
+        prompt="target 0",
+        prompt_len=4,
+        expected_output_len=100,
+        multi_modal_data=None,
+        request_id="session-0",
+        seed_tts_turns=tuple(
+            SeedTTSTurn(utterance_id=f"utt-{index}", target_text=f"target {index}") for index in range(4)
+        ),
+    )
+    output = types.SimpleNamespace(
+        success=True,
+        tts_turn_pcm_bytes=[bytes([index]) for index in range(4)],
+    )
+
+    requests, outputs = _expand_seed_tts_turn_outputs([request], [output])
+
+    assert [item.prompt for item in requests] == [f"target {index}" for index in range(4)]
+    assert [item.seed_tts_utterance_id for item in requests] == [f"utt-{index}" for index in range(4)]
+    assert [item.tts_output_pcm_bytes for item in outputs] == [bytes([index]) for index in range(4)]
+
+
+def test_seed_tts_eval_keeps_session_pcm_for_single_turn_chat():
+    """Chat-omni Seed-TTS sets session PCM only; expand must not wipe it."""
+    from vllm_omni.benchmarks.data_modules.seed_tts_eval import (
+        _expand_seed_tts_turn_outputs,
+    )
+
+    request = SeedTTSSampleRequest(
+        prompt="target 0",
+        prompt_len=4,
+        expected_output_len=100,
+        multi_modal_data=None,
+        request_id="session-0",
+        seed_tts_turns=(SeedTTSTurn(utterance_id="utt-0", target_text="target 0"),),
+    )
+    pcm = b"\x01\x02\x03\x04"
+    output = types.SimpleNamespace(
+        success=True,
+        tts_output_pcm_bytes=pcm,
+        tts_turn_pcm_bytes=None,
+    )
+
+    requests, outputs = _expand_seed_tts_turn_outputs([request], [output])
+
+    assert len(requests) == 1
+    assert requests[0].prompt == "target 0"
+    assert outputs[0].tts_output_pcm_bytes == pcm
+
+
 def test_seed_tts_text_dataset_omits_ref_audio(seed_tts_root, mock_tokenizer):
     ds = SeedTTSTextDataset(
         dataset_path=str(seed_tts_root),

tests/config/test_omni_config.py

@@ -392,11 +392,6 @@ def test_sub_config_fields_match_rfc_scopes():
         "enable_multithread_weight_load",
         "num_weight_load_threads",
         "disable_autocast",
-        # Per-stage checkpoint resolution for repos whose stages live in
-        # subfolders (e.g. Audex): mirrors StagePipelineConfig on the
-        # legacy engine-args path.
-        "model_subdir",
-        "tokenizer_subdir",
     }
     assert {f.name for f in fields(OmniStageLoadConfig)} == {
         "load_format",

tests/core/sched/test_omni_ar_scheduler_logprobs.py

@@ -1,182 +0,0 @@
-"""Tests for the AR sampled-token logprob contract."""
-
-from __future__ import annotations
-
-from collections import defaultdict
-from types import SimpleNamespace
-
-import numpy as np
-import pytest
-from vllm.v1.engine import FinishReason
-from vllm.v1.outputs import LogprobsLists
-from vllm.v1.request import RequestStatus
-
-from vllm_omni.core.sched.omni_ar_scheduler import OmniARScheduler, _slice_sampled_logprobs
-
-pytestmark = [pytest.mark.core_model, pytest.mark.cpu]
-
-
-def _logprobs(
-    token_ids: list[list[int]],
-    values: list[list[float]],
-    *,
-    cumulative_rows: list[int] | None = None,
-) -> LogprobsLists:
-    return LogprobsLists(
-        logprob_token_ids=np.asarray(token_ids, dtype=np.int32),
-        logprobs=np.asarray(values, dtype=np.float32),
-        sampled_token_ranks=np.zeros(len(token_ids), dtype=np.int32),
-        cu_num_generated_tokens=cumulative_rows,
-    )
-
-
-def test_slices_requested_rows_and_preserves_sampled_token_values() -> None:
-    all_logprobs = _logprobs(
-        [[3, 8], [4, 9], [11, 2], [12, 1]],
-        [[-0.3, -1.0], [-0.4, -1.1], [-0.11, -2.0], [-0.12, -2.1]],
-        cumulative_rows=[0, 2],
-    )
-
-    sliced = _slice_sampled_logprobs(all_logprobs, req_index=1, sampled_token_ids=[11, 12])
-
-    np.testing.assert_array_equal(sliced.logprob_token_ids[:, 0], [11, 12])
-    np.testing.assert_allclose(sliced.logprobs[:, 0], [-0.11, -0.12])
-
-
-def test_rejects_missing_logprobs() -> None:
-    with pytest.raises(RuntimeError, match="model runner returned none"):
-        _slice_sampled_logprobs(None, req_index=0, sampled_token_ids=[7])
-
-
-def test_rejects_row_count_mismatch() -> None:
-    logprobs = _logprobs([[7, 1]], [[-0.7, -1.0]])
-
-    with pytest.raises(RuntimeError, match="row count does not match"):
-        _slice_sampled_logprobs(logprobs, req_index=0, sampled_token_ids=[7, 8])
-
-
-def test_rejects_sampled_token_mismatch() -> None:
-    logprobs = _logprobs([[9, 7]], [[-0.9, -1.0]])
-
-    with pytest.raises(RuntimeError, match="generated_token=7 logprob_token=9"):
-        _slice_sampled_logprobs(logprobs, req_index=0, sampled_token_ids=[7])
-
-
-def test_rejects_non_finite_sampled_logprob() -> None:
-    logprobs = _logprobs([[7, 1]], [[float("nan"), -1.0]])
-
-    with pytest.raises(RuntimeError, match="non-finite values at rows \\[0\\]"):
-        _slice_sampled_logprobs(logprobs, req_index=0, sampled_token_ids=[7])
-
-
-class _Request:
-    def __init__(self, request_id: str) -> None:
-        self.request_id = request_id
-        self.client_index = 0
-        self.status = RequestStatus.RUNNING
-        self.sampling_params = SimpleNamespace(num_logprobs=1)
-        self.num_computed_tokens = 0
-        self.num_output_placeholders = 0
-        self.has_encoder_inputs = False
-        self.pooling_params = None
-        self.resumable = True
-        self.stop_reason = None
-        self.trace_headers = None
-        self.num_nans_in_logits = 0
-
-    def is_finished(self) -> bool:
-        return RequestStatus.is_finished(self.status)
-
-    def get_finished_reason(self):
-        return RequestStatus.get_finished_reason(self.status)
-
-    def take_events(self):
-        return None
-
-    def take_prefill_stats(self):
-        return None
-
-
-class _RequestQueue(list):
-    def remove_requests(self, requests) -> None:
-        for request in requests:
-            if request in self:
-                self.remove(request)
-
-
-@pytest.mark.skip(reason="#issue 5484")
-def test_invalid_logprobs_finish_only_the_affected_scheduler_request() -> None:
-    """A bad runner response must not bring down unrelated batch requests."""
-    bad = _Request("bad")
-    good = _Request("good")
-    scheduler = SimpleNamespace(
-        perf_metrics=None,
-        connector=None,
-        chunk_transfer_adapter=None,
-        requests={bad.request_id: bad, good.request_id: good},
-        running=[bad, good],
-        waiting=_RequestQueue(),
-        skipped_waiting=_RequestQueue(),
-        structured_output_manager=SimpleNamespace(should_advance=lambda _request: False),
-        transfer_triggered_requests=set(),
-        active_kv_transfers=set(),
-        pending_stop_after_extraction=set(),
-        waiting_for_transfer_free=set(),
-        _new_prompt_len_snapshot={},
-        finished_req_ids=set(),
-        finished_req_ids_dict=defaultdict(set),
-        kv_cache_manager=SimpleNamespace(take_events=lambda: None),
-        kv_event_publisher=SimpleNamespace(publish=lambda _events: None),
-    )
-    update_calls: list[str] = []
-
-    def update_request(request, token_ids):
-        update_calls.append(request.request_id)
-        return token_ids, False
-
-    def free_request(request):
-        scheduler.requests.pop(request.request_id, None)
-        scheduler.finished_req_ids.add(request.request_id)
-        scheduler.finished_req_ids_dict[request.client_index].add(request.request_id)
-        return None
-
-    scheduler._update_request_with_output = update_request
-    scheduler._process_kv_transfer_trigger = lambda _request, _tokens: False
-    scheduler._handle_stopped_request = lambda _request: True
-    scheduler._free_request = free_request
-    scheduler.make_spec_decoding_stats = lambda *args, **kwargs: None
-    scheduler.make_stats = lambda *args, **kwargs: None
-    scheduler._capture_omni_connector_output = lambda _output: None
-
-    scheduler_output = SimpleNamespace(
-        num_scheduled_tokens={"bad": 1, "good": 1},
-        scheduled_spec_decode_tokens={},
-        num_invalid_spec_tokens=0,
-    )
-    model_runner_output = SimpleNamespace(
-        sampled_token_ids=[[7], [8]],
-        logprobs=_logprobs(
-            [[9, 1], [8, 2]],
-            [[-0.9, -1.0], [-0.8, -1.0]],
-            cumulative_rows=[0, 1],
-        ),
-        prompt_logprobs_dict={},
-        pooler_output=None,
-        num_nans_in_logits=None,
-        kv_connector_output=None,
-        cudagraph_stats=None,
-        req_id_to_index={"bad": 0, "good": 1},
-        routed_experts=None,
-    )
-
-    outputs = OmniARScheduler.update_from_output(scheduler, scheduler_output, model_runner_output)
-    output_by_id = {output.request_id: output for output in outputs[0].outputs}
-
-    assert update_calls == ["good"]
-    assert bad.status is RequestStatus.FINISHED_ERROR
-    assert bad.stop_reason is not None and "misaligned" in bad.stop_reason
-    assert "bad" not in scheduler.requests
-    assert output_by_id["bad"].finish_reason is FinishReason.ERROR
-    assert output_by_id["bad"].new_token_ids == []
-    assert output_by_id["good"].new_token_ids == [8]
-    np.testing.assert_array_equal(output_by_id["good"].new_logprobs.logprob_token_ids[:, 0], [8])

tests/core/sched/test_omni_ar_scheduler_streaming.py

@@ -73,9 +73,6 @@ def _run_resumable_segment_stop(
 
     sched._update_request_with_output.side_effect = stop_request
     sched._handle_stopped_request.return_value = session_finished
-    # vLLM 0.26 returns (kv_xfer_params, ec_xfer_params); an unconfigured
-    # MagicMock iterates empty and fails to unpack at the call site.
-    sched._free_request.return_value = (None, None)
     sched.chunk_transfer_adapter = None
     sched.running = [session]
     sched.waiting_for_transfer_free = set()
@@ -140,23 +137,6 @@ def test_resumable_session_terminal_is_not_marked_as_segment_boundary() -> None:
     assert output.is_segment_finished is False
 
 
-def test_update_from_output_settles_in_flight_tokens() -> None:
-    """vLLM 0.26: schedule() increments num_in_flight_tokens per scheduled
-    token; update_from_output must decrement it symmetrically. If the
-    decrement is dropped the counter grows monotonically and both readers
-    (allocate_slots, _connector_finished) clamp
-    max(0, num_computed_tokens - num_in_flight_tokens) to zero forever,
-    silently freezing sliding-window block freeing.
-    """
-    session = _make_request()
-    session.status = RequestStatus.RUNNING
-    session.num_in_flight_tokens = 1  # as left by schedule() for this step
-
-    _run_resumable_segment_stop(session)
-
-    assert session.num_in_flight_tokens == 0
-
-
 def test_running_decode_step_without_inter_stage_payload_does_not_raise() -> None:
     """A decode step that neither stops nor carries an inter-stage payload.
 
@@ -262,9 +242,52 @@ def test_explicit_streaming_payload_replaces_placeholder_prompt() -> None:
         "tts_token_ids": [10, 20],
         "meta": {"replace_streaming_prompt": True},
     }
+    update.model_intermediate_buffer = {
+        "ids": {"tts": [41, 42, 99]},
+        "meta": {"turn_eos_token_id": 99},
+    }
 
     sched._update_request_as_session(session, update)
 
     assert session.prompt_token_ids == [10, 20]
     assert session.additional_information == update.additional_information
+    assert session.model_intermediate_buffer == {
+        "ids": {"tts": [41, 42, 99]},
+        "meta": {"turn_eos_token_id": 99},
+    }
+    assert session.status == RequestStatus.WAITING
+
+
+def test_model_intermediate_streaming_payload_replaces_computed_prompt() -> None:
+    sched = _make_scheduler(stage_id=1)
+    sched.chunk_transfer_adapter = SimpleNamespace(
+        receives_chunks=False,
+        segment_finished_requests=set(),
+    )
+    session = _make_request()
+    session.status = RequestStatus.WAITING_FOR_STREAMING_REQ
+    session.prompt_token_ids = [0] * 59
+    session._all_token_ids.clear()
+    session._all_token_ids.extend(session.prompt_token_ids)
+    session.num_prompt_tokens = 59
+    session.num_computed_tokens = 59
+    update = _make_update([0] * 10)
+    update.additional_information = None
+    update.model_intermediate_buffer = {
+        "ids": {"tts": list(range(8))},
+        "hidden_states": {"tts": [[0.0]] * 8},
+        "meta": {
+            "replace_streaming_prompt": True,
+            "next_stage_prompt_len": 10,
+        },
+    }
+
+    sched._update_request_as_session(session, update)
+
+    assert session.prompt_token_ids == [0] * 10
+    assert list(session._all_token_ids) == [0] * 10
+    assert session.num_prompt_tokens == 10
+    assert session.num_computed_tokens == 0
+    assert session.additional_information is None
+    assert session.model_intermediate_buffer == update.model_intermediate_buffer
     assert session.status == RequestStatus.WAITING

tests/core/sched/test_omni_generation_scheduler_update_session.py

@@ -54,6 +54,18 @@ class _SchedulerStub(OmniGenerationScheduler):
         raise AssertionError("unexpected enqueue for skipped_waiting miss")
 
 
+class _AsyncChunkStopSchedulerStub(OmniGenerationScheduler):
+    def __init__(self) -> None:
+        self.num_waiting_for_streaming_input = 0
+        self.chunk_transfer_adapter = SimpleNamespace(receives_chunks=True)
+        self.enqueued: list[Request] = []
+        self.enqueued_statuses: list[RequestStatus] = []
+
+    def _enqueue_waiting_request(self, session: Request) -> None:
+        self.enqueued.append(session)
+        self.enqueued_statuses.append(session.status)
+
+
 def _make_request(**kwargs) -> Request:
     sp = SamplingParams(max_tokens=8)
     defaults = dict(
@@ -133,6 +145,21 @@ def test_resumable_generation_stop_marks_segment_boundary() -> None:
     assert output.is_segment_finished is True
 
 
+def test_async_chunk_resumable_stop_rearms_connector_polling() -> None:
+    sched = _AsyncChunkStopSchedulerStub()
+    session = _make_request(request_id="req-async-chunk-next-segment")
+    session.status = RequestStatus.FINISHED_STOPPED
+    session.resumable = True
+
+    finished = sched._handle_stopped_request(session)
+
+    assert finished is False
+    assert sched.enqueued == [session]
+    assert sched.enqueued_statuses == [RequestStatus.WAITING]
+    assert session.status == RequestStatus.WAITING
+    assert sched.num_waiting_for_streaming_input == 0
+
+
 class TestReplaceSessionWithStreamingUpdate:
     def test_resets_tokens_and_prompt_from_update(self) -> None:
         sched = _SchedulerStub()

tests/core/sched/test_omni_scheduler_finish_requests_purge.py

@@ -76,9 +76,8 @@ def test_finish_requests_purges_finished_request_from_running(
     )
 
     def fake_finish_requests(self, request_ids, finished_status):
-        # vLLM 0.26 super() returns the finished Request objects but here
-        # leaves self.running untouched to simulate the missed removal.
-        return [finished]
+        # super() returning the abort tuple but leaving self.running untouched
+        return [(finished.request_id, 0)]
 
     monkeypatch.setattr(scheduler_mod.VLLMScheduler, "finish_requests", fake_finish_requests)
 
@@ -89,7 +88,7 @@ def test_finish_requests_purges_finished_request_from_running(
     )
 
     assert scheduler.running == []
-    assert result == [finished]
+    assert result == [(finished.request_id, 0)]
 
 
 @pytest.mark.parametrize(("scheduler_cls", "scheduler_mod"), _SCHEDULER_PARAMS)
@@ -112,7 +111,7 @@ def test_finish_requests_purges_untracked_request_from_running(
     )
 
     def fake_finish_requests(self, request_ids, finished_status):
-        return [untracked]
+        return [(untracked.request_id, 0)]
 
     monkeypatch.setattr(scheduler_mod.VLLMScheduler, "finish_requests", fake_finish_requests)
 
@@ -123,7 +122,7 @@ def test_finish_requests_purges_untracked_request_from_running(
     )
 
     assert scheduler.running == []
-    assert result == [untracked]
+    assert result == [(untracked.request_id, 0)]
 
 
 @pytest.mark.parametrize(("scheduler_cls", "scheduler_mod"), _SCHEDULER_PARAMS)
@@ -137,7 +136,7 @@ def test_finish_requests_leaves_healthy_running_intact(
     is conservative and won't aggressively drop healthy entries.
     """
     alive = _StubRequest("req-alive", RequestStatus.RUNNING)
-    leaving = _StubRequest("req-leaving", RequestStatus.FINISHED_STOPPED)
+    leaving_id = "req-leaving"
     scheduler = _make_scheduler(
         scheduler_cls,
         requests={alive.request_id: alive},
@@ -146,13 +145,13 @@ def test_finish_requests_leaves_healthy_running_intact(
     )
 
     def fake_finish_requests(self, request_ids, finished_status):
-        return [leaving]
+        return [(leaving_id, 0)]
 
     monkeypatch.setattr(scheduler_mod.VLLMScheduler, "finish_requests", fake_finish_requests)
 
     scheduler_cls.finish_requests(
         scheduler,
-        [leaving.request_id],
+        [leaving_id],
         RequestStatus.FINISHED_STOPPED,
     )
 

tests/core/sched/test_omni_scheduler_input_coordinator_cleanup.py

@@ -53,11 +53,7 @@ def test_finish_requests_cleans_input_coordinator_for_finished_ids(
     def fake_finish_requests(self, request_ids, finished_status):
         assert request_ids == ["req-a", "req-b"]
         assert finished_status == RequestStatus.FINISHED_STOPPED
-        # vLLM 0.26 returns the finished Request objects, not (req_id, client_index) tuples.
-        return [
-            SimpleNamespace(request_id="req-a", client_index=0),
-            SimpleNamespace(request_id="req-b", client_index=1),
-        ]
+        return [("req-a", 0), ("req-b", 1)]
 
     monkeypatch.setattr(scheduler_mod.VLLMScheduler, "finish_requests", fake_finish_requests)
 
@@ -67,7 +63,7 @@ def test_finish_requests_cleans_input_coordinator_for_finished_ids(
         RequestStatus.FINISHED_STOPPED,
     )
 
-    assert [(r.request_id, r.client_index) for r in result] == [("req-a", 0), ("req-b", 1)]
+    assert result == [("req-a", 0), ("req-b", 1)]
     assert coordinator.freed == ["req-a", "req-b"]
 
 
@@ -86,7 +82,7 @@ def test_ar_free_request_cleans_input_coordinator_on_normal_free() -> None:
 
     result = OmniARScheduler._free_request(scheduler, DummyRequest())
 
-    assert result == (None, None)
+    assert result is None
     assert coordinator.freed == ["req-free"]
     assert scheduler._omits_kv_transfer_cache == {}
     assert scheduler._new_prompt_len_snapshot == {}

tests/core/sched/test_omni_scheduling_coordinator.py

@@ -121,10 +121,6 @@ class TestFullPayloadCoordinatorSelection(unittest.TestCase):
             ("Qwen3TTSCode2Wav", "code2wav"),
             ("MiniCPMO45Code2Wav", "code2wav"),
             ("CosyVoice3Model", "cosyvoice3_code2wav"),
-            # audex: sync-path TTS decoder (async_chunk: false deploys) and
-            # the TTA XCodec1 stage (always sync full-payload).
-            ("AudexCode2Wav", "audex_code2wav"),
-            ("AudexXCodec1", "audex_xcodec"),
             ("IndexTTS2S2MelDecoder", "indextts2_s2mel_decoder"),
             ("DyninOmniForConditionalGeneration", "token2image"),
             ("DyninOmniForConditionalGeneration", "token2audio"),

tests/dfx/conftest.py

@@ -3,15 +3,13 @@ import os
 import re
 import subprocess
 import threading
-import time
 from datetime import datetime
 from pathlib import Path
 from typing import Any
 
 import pytest
 
-from tests.dfx.reliability.helpers import list_remote_process_pids_by_pattern, post_chat_completions_raw
-from tests.helpers.mark import hardware_marks
+from tests.helpers.mark import get_hardware_mark_list, hardware_marks
 from tests.helpers.runtime import OmniServerParams
 from tests.helpers.stage_config import modify_stage_config
 from vllm_omni.platforms import current_omni_platform
@@ -119,32 +117,6 @@ def create_unique_server_pytest_params(
     ]
 
 
-def create_benchmark_pytest_params(
-    benchmark_configs: list[dict[str, Any]],
-    server_to_benchmark_mapping: dict[str, dict],
-) -> list[Any]:
-    """Like :func:`create_benchmark_indices`, but wrap each index in ``pytest.param`` with JSON marks."""
-    marks_by_name = _marks_by_test_name(benchmark_configs)
-    params: list[Any] = []
-    seen: set[str] = set()
-    for config in benchmark_configs:
-        test_name = config["test_name"]
-        if test_name in seen:
-            continue
-        seen.add(test_name)
-        params_list = get_benchmark_params_for_server(test_name, server_to_benchmark_mapping)
-        id_suffixes = _unique_benchmark_param_id_suffixes(params_list)
-        for idx, id_suffix in enumerate(id_suffixes):
-            params.append(
-                pytest.param(
-                    (test_name, idx),
-                    marks=marks_by_name.get(test_name, []),
-                    id=f"{test_name}-{id_suffix}",
-                )
-            )
-    return params
-
-
 def create_paired_benchmark_pytest_params(
     server_entries: list[tuple[Any, str]],
     params_by_test_name: dict[str, list[Any]],
@@ -357,10 +329,6 @@ def supports_video_generation(model_name: str) -> bool:
     return any(key in lower for key in ("wan", "video", "i2v", "t2v"))
 
 
-def supports_chat_generation(model_name: str) -> bool:
-    return not supports_video_generation(model_name)
-
-
 def parse_stage_devices(stage_config_path: str) -> str:
     text = Path(stage_config_path).read_text(encoding="utf-8")
     raw_devices: list[str] = re.findall(r"^\s*devices:\s*\"?([0-9,\s]+)\"?\s*$", text, flags=re.MULTILINE)
@@ -389,48 +357,6 @@ def assert_fault_exception(exc: Exception, error_keywords: tuple[str, ...]) -> N
     assert any(key in text for key in error_keywords), f"unexpected error under fault injection: {exc}"
 
 
-def wait_chat_request_ready(host: str, port: int, model: str, timeout_sec: int = 180) -> None:
-    """Poll a minimal chat request until success."""
-    deadline = time.time() + timeout_sec
-    payload = json.dumps(
-        {
-            "model": model,
-            "messages": [{"role": "user", "content": "Say hello in one short sentence."}],
-            "stream": False,
-            "modalities": ["text"],
-        }
-    )
-    last_error: str | None = None
-    while time.time() < deadline:
-        try:
-            status, body = post_chat_completions_raw(host, port, payload)
-            if status == 200:
-                return
-            last_error = f"http={status} body={body[:200]!r}"
-        except Exception as exc:  # noqa: BLE001
-            last_error = str(exc)
-        time.sleep(2)
-    raise TimeoutError(f"runtime-teardown warmup request did not succeed within {timeout_sec}s: {last_error}")
-
-
-def assert_no_extra_worker_processes(
-    baseline_pids: set[int],
-    worker_pattern: str,
-    timeout_sec: int = 60,
-) -> None:
-    """Ensure no extra worker PIDs remain after teardown."""
-    deadline = time.time() + timeout_sec
-    while time.time() < deadline:
-        current = set(list_remote_process_pids_by_pattern(worker_pattern))
-        extra = current - baseline_pids
-        if not extra:
-            return
-        time.sleep(2)
-    current = set(list_remote_process_pids_by_pattern(worker_pattern))
-    extra = sorted(current - baseline_pids)
-    assert not extra, f"orphan worker processes remain after container teardown: {extra}"
-
-
 def create_test_parameter_mapping(configs: list[dict[str, Any]]) -> dict[str, dict]:
     mapping = {}
     for config in configs:
@@ -520,36 +446,16 @@ def _unique_benchmark_param_id_suffixes(params_list: list[dict[str, Any]]) -> li
     return unique
 
 
-def extract_mark_resource_label(mark_field: Any) -> str:
-    """Return a filename-safe hardware label from ``mark.hardware_marks.res`` values.
-
-    Example: ``{"cuda": "H100"}`` -> ``"H100"``; multiple platforms join with ``-``.
-
-    Prefer :func:`get_runtime_resource_label` for perf result filenames so labels
-    reflect the machine that actually ran the benchmark.
-    """
-    if isinstance(mark_field, list):
-        for item in mark_field:
-            if isinstance(item, dict) and "hardware_marks" in item:
-                return extract_mark_resource_label(item)
-        return "na"
-    if not isinstance(mark_field, dict):
-        return "na"
-    hw = mark_field.get("hardware_marks")
-    if not isinstance(hw, dict):
-        return "na"
-    res = hw.get("res")
-    if not isinstance(res, dict) or not res:
-        return "na"
-    labels = [_safe_filename_token(value) for value in res.values()]
-    return "-".join(labels) if labels else "na"
-
-
-_KNOWN_RUNTIME_RESOURCE_TOKENS: tuple[str, ...] = (
+# Device marketing-name tokens for ``get_runtime_resource_label`` only.
+# Longer / more specific tokens must appear before shorter substrings
+# (e.g. ``H200`` before ``H20``, ``910B4`` before ``910``).
+_RUNTIME_DEVICE_ALIASES: tuple[str, ...] = (
     "H100",
     "H800",
     "H200",
     "H20",
+    "B200",
+    "GB200",
     "L40S",
     "L40",
     "L4",
@@ -578,7 +484,7 @@ def _normalize_runtime_device_label(raw: str) -> str:
     if not raw or not str(raw).strip():
         return "na"
     upper = str(raw).upper()
-    for token in _KNOWN_RUNTIME_RESOURCE_TOKENS:
+    for token in _RUNTIME_DEVICE_ALIASES:
         if token.upper() in upper:
             return _safe_filename_token(token)
     compact = re.sub(r"[^a-zA-Z0-9]+", "", str(raw))
@@ -631,34 +537,41 @@ def resource_label_for_filename(resource_label: str | None) -> str:
     return token
 
 
-def extract_configs_resource_label(configs: list[dict[str, Any]]) -> str:
-    """Return runtime hardware label for perf result filenames."""
-    del configs
-    return get_runtime_resource_label()
+def is_hardware_nested_baseline(baseline: dict[str, Any]) -> bool:
+    """True for ``{"H100": {"metric": ...}, "A3": {...}}`` baselines.
+
+    Detection is structural + hardware-label allowlist (not metric-name hints):
+
+    - every top-level value is a non-empty dict (per-hardware metric map);
+    - every top-level key is in :func:`tests.helpers.mark.get_hardware_mark_list`
+      (``[hardware-resource]`` markers from ``pyproject.toml``).
+
+    Metric names under each hardware bucket are unconstrained.
+    Runtime device aliases (:data:`_RUNTIME_DEVICE_ALIASES`) are unrelated.
+    """
+    if not baseline:
+        return False
+    if not all(isinstance(value, dict) and bool(value) for value in baseline.values()):
+        return False
+    allowed = {label.upper() for label in get_hardware_mark_list()}
+    return all((token := str(key).strip()) and token.upper() in allowed for key in baseline.keys())
 
 
 def resolve_baseline_value(
     baseline_raw: Any,
     *,
     sweep_index: int | None,
-    max_concurrency: Any = None,
-    request_rate: Any = None,
 ) -> Any:
-    """Pick the baseline threshold for this sweep step."""
+    """Pick the baseline threshold for one metric at this sweep step.
+
+    Accepts a sweep-aligned list/tuple (indexed by ``sweep_index``) or a scalar.
+    """
     if baseline_raw is None:
-        return 100000
+        return None
     if isinstance(baseline_raw, dict):
-        if max_concurrency is not None:
-            for key in (max_concurrency, str(max_concurrency)):
-                if key in baseline_raw:
-                    return baseline_raw[key]
-        if request_rate is not None:
-            for key in (request_rate, str(request_rate)):
-                if key in baseline_raw:
-                    return baseline_raw[key]
-        raise KeyError(
-            f"baseline dict has no key for max_concurrency={max_concurrency!r} "
-            f"or request_rate={request_rate!r}; keys={list(baseline_raw.keys())!r}"
+        raise TypeError(
+            "per-metric baseline dict keyed by concurrency/request-rate is not supported; "
+            "use a sweep-aligned list or a scalar under each hardware bucket"
         )
     if isinstance(baseline_raw, (list, tuple)):
         if sweep_index is None:
@@ -669,22 +582,62 @@ def resolve_baseline_value(
     return baseline_raw
 
 
-def _baseline_thresholds_for_step(
-    baseline_data: dict[str, Any],
+def resolve_baseline_for_sweep(
+    baseline_config: dict[str, Any] | None,
     *,
     sweep_index: int | None = None,
-    max_concurrency: Any = None,
-    request_rate: Any = None,
 ) -> dict[str, Any]:
-    """Resolve baseline config to one threshold per metric for this iteration."""
-    return {
-        metric_name: resolve_baseline_value(
-            baseline_raw,
-            sweep_index=sweep_index,
-            max_concurrency=max_concurrency,
-            request_rate=request_rate,
+    """Resolve sweep-dependent baselines while keeping every hardware bucket.
+
+    Expects hardware-nested input only::
+
+        {"H100": {"throughput_qps": [a, b, c]}, "A3": {"throughput_qps": [x, y, z]}}
+
+    with ``sweep_index=2`` (e.g. concurrency 32) becomes::
+
+        {"H100": {"throughput_qps": c}, "A3": {"throughput_qps": z}}
+
+    Per-metric values may be a sweep-aligned list or a scalar.
+    """
+    if not baseline_config:
+        return {}
+    if not is_hardware_nested_baseline(baseline_config):
+        allowed = sorted(get_hardware_mark_list())
+        unknown = [
+            str(key).strip()
+            for key in baseline_config.keys()
+            if str(key).strip().upper() not in {label.upper() for label in allowed}
+        ]
+        nested_shape = all(isinstance(value, dict) and bool(value) for value in baseline_config.values())
+        if nested_shape and unknown:
+            raise ValueError(
+                "baseline top-level keys must be [hardware-resource] pytest markers from "
+                "pyproject.toml (tool.pytest.ini_options.markers). "
+                f"Unknown hardware label(s): {unknown!r}. "
+                f"Known markers: {allowed!r}. "
+                "Add a marker such as '\"A100: [hardware-resource] Tests that require A100 GPU\"' "
+                "to pyproject.toml, then retry."
+            )
+        raise ValueError(
+            "baseline must be hardware-nested with [hardware-resource] marker names as "
+            "top-level keys, e.g. "
+            '{"H100": {"throughput_qps": [...]}, "A3": {"custom_metric": 1.0}}. '
+            f"got top-level keys={list(baseline_config.keys())!r}; "
+            f"known markers={allowed!r}. "
+            "If you need a new hardware bucket, add it to pyproject.toml markers with "
+            "the [hardware-resource] tag."
         )
-        for metric_name, baseline_raw in baseline_data.items()
+
+    return {
+        hardware: {
+            metric_name: resolve_baseline_value(
+                baseline_raw,
+                sweep_index=sweep_index,
+            )
+            for metric_name, baseline_raw in metrics.items()
+        }
+        for hardware, metrics in baseline_config.items()
+        if isinstance(metrics, dict)
     }
 
 
@@ -697,17 +650,17 @@ def run_benchmark(
     *,
     baseline_config: dict[str, Any] | None = None,
     sweep_index: int | None = None,
-    request_rate: Any | None = None,
-    max_concurrency: Any | None = None,
     random_input_len: Any | None = None,
     random_output_len: Any | None = None,
     resource_label: str | None = None,
 ) -> dict[str, Any]:
     """Run one ``vllm bench serve --omni`` iteration and return parsed metrics.
 
-    After ``vllm bench`` writes the JSON, ``result["baseline"]`` holds the resolved per-metric thresholds
-    (when ``baseline_config`` is provided). If the benchmark exits without writing a result file,
-    ``result_omni_template.json`` is used as a fallback.
+    After ``vllm bench`` writes the JSON, ``result["baseline"]`` stores the
+    sweep-resolved baseline (when ``baseline_config`` is provided). Hardware-nested
+    maps keep every hardware bucket, but each metric is reduced to the value for
+    this concurrency / request-rate step. If the benchmark exits without writing a
+    result file, ``result_omni_template.json`` is used as a fallback.
     """
     current_dt = datetime.now().strftime("%Y%m%d-%H%M%S")
     ri = _safe_filename_token(random_input_len)
@@ -774,11 +727,10 @@ def run_benchmark(
             result = json.load(f)
 
     if baseline_config:
-        result["baseline"] = _baseline_thresholds_for_step(
+        # Keep every hardware bucket; resolve list metrics to this sweep step.
+        result["baseline"] = resolve_baseline_for_sweep(
             baseline_config,
             sweep_index=sweep_index,
-            request_rate=request_rate,
-            max_concurrency=max_concurrency,
         )
     else:
         result["baseline"] = {}
@@ -800,12 +752,3 @@ def pytest_addoption(parser: pytest.Parser) -> None:
         default=None,
         help=("Path to benchmark config JSON. Example: --test-config-file tests/dfx/perf/tests/test_tts.json"),
     )
-    parser.addoption(
-        "--assert-baseline",
-        action="store_true",
-        default=False,
-        help=(
-            "When set, omni/diffusion perf runners compare metrics against the baseline block in the JSON config "
-            "(default: off)."
-        ),
-    )

tests/dfx/perf/scripts/diffusion_result_template.json

@@ -21,10 +21,12 @@
       "num-input-images": 0,
       "enable-negative-prompt": false,
       "baseline": {
-        "throughput_qps": 0,
-        "latency_mean": 0,
-        "peak_memory_mb_max": 0,
-        "peak_memory_mb_mean": 0
+        "H100": {
+          "throughput_qps": 0,
+          "latency_mean": 0,
+          "peak_memory_mb_max": 0,
+          "peak_memory_mb_mean": 0
+        }
       }
     },
     "result": {

tests/dfx/perf/scripts/result_omni_template.json

@@ -46,9 +46,11 @@
   "median_audio_duration_s": 0,
   "p99_audio_duration_s": 0,
   "baseline": {
-    "mean_ttft_ms": 0,
-    "mean_audio_ttfp_ms": 0,
-    "mean_audio_rtf": 0
+    "H100": {
+      "mean_ttft_ms": 0,
+      "mean_audio_ttfp_ms": 0,
+      "mean_audio_rtf": 0
+    }
   },
   "random_input_len": 0,
   "random_output_len": 0

tests/dfx/perf/scripts/run_benchmark.py

@@ -1,6 +1,8 @@
 import json
 import os
 import threading
+from collections.abc import Callable
+from contextlib import AbstractContextManager, ExitStack, contextmanager
 from pathlib import Path
 from typing import Any
 
@@ -13,15 +15,10 @@ from tests.dfx.conftest import (
     get_runtime_resource_label,
     is_diffusion_perf_config,
     load_benchmark_configs,
-    resolve_baseline_value,
     run_benchmark,
 )
 from tests.helpers.runtime import OmniServer
 
-# Compare metrics to each test JSON ``baseline`` block only when pytest is run with ``--assert-baseline``
-# (registered in ``tests/dfx/conftest.py``; default: off). ``run_benchmark`` and ``_resolve_baseline_value`` are
-# defined in the same module.
-#
 # Optional JSON field ``mark`` is applied as pytest marks via
 # ``create_paired_omni_benchmark_pytest_params`` (e.g. ``"mark": [{"hardware_marks":
 # {"res": {"cuda": "H100"}, "num_cards": 2}}, "full_model", "omni"]``).
@@ -63,35 +60,78 @@ paired_benchmark_params = create_paired_omni_benchmark_pytest_params(BENCHMARK_C
 _omni_server_lock = threading.Lock()
 
 
+class _SingleActiveContext:
+    """Reuse one active context while its configuration key is unchanged."""
+
+    def __init__(self) -> None:
+        self._key: Any = None
+        self._stack: ExitStack | None = None
+        self._value: Any = None
+
+    def acquire(self, key: Any, factory: Callable[[], AbstractContextManager[Any]]) -> Any:
+        if self._stack is not None and key == self._key:
+            return self._value
+
+        self.close()
+        stack = ExitStack()
+        value = stack.enter_context(factory())
+        self._key = key
+        self._stack = stack
+        self._value = value
+        return value
+
+    def close(self) -> None:
+        stack = self._stack
+        self._key = None
+        self._stack = None
+        self._value = None
+        if stack is not None:
+            stack.close()
+
+
+@contextmanager
+def _start_omni_server(server_param):
+    test_name, model, stage_config_path, stage_overrides, extra_cli_args, use_omni = server_param
+
+    print(f"Starting OmniServer with test: {test_name}, model: {model}")
+
+    server_args: list[str] = []
+    if use_omni:
+        server_args += ["--stage-init-timeout", "600", "--init-timeout", "900"]
+    # --deploy-config and --stage-overrides compose at the CLI (see vllm_omni/entrypoints/utils.py):
+    # deploy-config sets the base; stage-overrides are applied on top. Both can be set.
+    if stage_config_path:
+        server_args = ["--deploy-config", stage_config_path] + server_args
+    if stage_overrides:
+        server_args = ["--stage-overrides", stage_overrides] + server_args
+    if extra_cli_args:
+        server_args = list(extra_cli_args) + server_args
+    with OmniServer(model, server_args, use_omni=use_omni) as server:
+        server.test_name = test_name
+        print("OmniServer started successfully")
+        yield server
+        print("OmniServer stopping...")
+
+    print("OmniServer stopped")
+
+
 @pytest.fixture(scope="module")
-def omni_server(request):
+def omni_server_context():
     """Start vLLM-Omni server as a subprocess with actual model weights.
-    Uses session scope so the server starts only once for the entire test session.
+    Reuse it for adjacent benchmark cases with the same server configuration.
     Multi-stage initialization can take 10-20+ minutes.
     """
     with _omni_server_lock:
-        test_name, model, stage_config_path, stage_overrides, extra_cli_args, use_omni = request.param
+        active_context = _SingleActiveContext()
+        try:
+            yield active_context
+        finally:
+            active_context.close()
 
-        print(f"Starting OmniServer with test: {test_name}, model: {model}")
 
-        server_args: list[str] = []
-        if use_omni:
-            server_args += ["--stage-init-timeout", "600", "--init-timeout", "900"]
-        # --deploy-config and --stage-overrides compose at the CLI (see vllm_omni/entrypoints/utils.py):
-        # deploy-config sets the base; stage-overrides are applied on top. Both can be set.
-        if stage_config_path:
-            server_args = ["--deploy-config", stage_config_path] + server_args
-        if stage_overrides:
-            server_args = ["--stage-overrides", stage_overrides] + server_args
-        if extra_cli_args:
-            server_args = list(extra_cli_args) + server_args
-        with OmniServer(model, server_args, use_omni=use_omni) as server:
-            server.test_name = test_name
-            print("OmniServer started successfully")
-            yield server
-            print("OmniServer stopping...")
-
-        print("OmniServer stopped")
+@pytest.fixture
+def omni_server(request, omni_server_context):
+    return omni_server_context.acquire(request.param, lambda: _start_omni_server(request.param))
 
 
 @pytest.fixture
@@ -117,36 +157,19 @@ def benchmark_params(request):
     }
 
 
-def assert_result(
-    result,
-    params,
-    num_prompt,
-    *,
-    assert_baseline: bool,
-    sweep_index: int | None = None,
-    max_concurrency: Any | None = None,
-    request_rate: Any | None = None,
-) -> None:
+def assert_result(result, params, num_prompt) -> None:
     assert result["completed"] == num_prompt, "Request failures exist"
-    if not assert_baseline:
-        return
-    baseline_data = params.get("baseline", {})
-    for metric_name, baseline_raw in baseline_data.items():
-        current_value = result[metric_name]
-        baseline_value = resolve_baseline_value(
-            baseline_raw,
-            sweep_index=sweep_index,
-            max_concurrency=max_concurrency,
-            request_rate=request_rate,
+    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")
+        assert isinstance(session_metrics, list), "Duplex session metrics are missing"
+        assert len(session_metrics) == num_prompt, (
+            f"Expected {num_prompt} duplex session metric rows, got {len(session_metrics)}"
         )
-        if "throughput" in metric_name:
-            if current_value <= baseline_value:
-                print(
-                    f"ERROR: Throughput test results were below baseline: {metric_name}: {current_value} > {baseline_value}"
-                )
-        else:
-            if current_value >= baseline_value:
-                print(f"ERROR: Test results exceeded baseline: {metric_name}: {current_value} < {baseline_value}")
+        assert all(
+            isinstance(metric, dict) and metric.get("audio_turn_count") == expected_audio_turns
+            for metric in session_metrics
+        ), f"Not every duplex session emitted {expected_audio_turns} audio turns"
 
 
 @pytest.mark.benchmark
@@ -155,7 +178,7 @@ def assert_result(
     paired_benchmark_params,
     indirect=["omni_server", "benchmark_params"],
 )
-def test_performance_benchmark(omni_server, benchmark_params, request):
+def test_performance_benchmark(omni_server, benchmark_params):
     test_name = benchmark_params["test_name"]
     params = benchmark_params["params"]
     dataset_name = params.get("dataset_name", "")
@@ -167,7 +190,6 @@ def test_performance_benchmark(omni_server, benchmark_params, request):
     print(f"Running benchmark for model: {model}")
     print(f"Benchmark parameters: {benchmark_params}")
 
-    assert_baseline = request.config.getoption("--assert-baseline", default=False)
     resource_label = get_runtime_resource_label()
 
     def to_list(value, default=None):
@@ -201,6 +223,7 @@ def test_performance_benchmark(omni_server, benchmark_params, request):
         "enabled",
         "eval_phase",
         "trust_remote_code",
+        "expected_duplex_audio_turns_per_session",
     }
 
     for key, value in params.items():
@@ -226,7 +249,7 @@ def test_performance_benchmark(omni_server, benchmark_params, request):
         break
 
     # QPS / request-rate sweep
-    for i, (qps, num_prompt) in enumerate(zip(qps_list, num_prompt_list)):
+    for sweep_index, (qps, num_prompt) in enumerate(zip(qps_list, num_prompt_list)):
         args = args + ["--request-rate", str(qps), "--num-prompts", str(num_prompt)]
         result = run_benchmark(
             args=args,
@@ -235,24 +258,15 @@ def test_performance_benchmark(omni_server, benchmark_params, request):
             dataset_name=dataset_name,
             num_prompt=num_prompt,
             baseline_config=params.get("baseline"),
-            sweep_index=i,
-            request_rate=qps,
-            max_concurrency=None,
+            sweep_index=sweep_index,
             random_input_len=params.get("random_input_len"),
             random_output_len=params.get("random_output_len"),
             resource_label=resource_label,
         )
-        assert_result(
-            result,
-            params,
-            num_prompt,
-            assert_baseline=assert_baseline,
-            sweep_index=i,
-            request_rate=qps,
-        )
+        assert_result(result, params, num_prompt)
 
     # concurrency test
-    for i, (concurrency, num_prompt) in enumerate(zip(max_concurrency_list, num_prompt_list)):
+    for sweep_index, (concurrency, num_prompt) in enumerate(zip(max_concurrency_list, num_prompt_list)):
         args = args + ["--max-concurrency", str(concurrency), "--num-prompts", str(num_prompt), "--request-rate", "inf"]
         result = run_benchmark(
             args=args,
@@ -261,18 +275,9 @@ def test_performance_benchmark(omni_server, benchmark_params, request):
             dataset_name=dataset_name,
             num_prompt=num_prompt,
             baseline_config=params.get("baseline"),
-            sweep_index=i,
-            request_rate=None,
-            max_concurrency=concurrency,
+            sweep_index=sweep_index,
             random_input_len=params.get("random_input_len"),
             random_output_len=params.get("random_output_len"),
             resource_label=resource_label,
         )
-        assert_result(
-            result,
-            params,
-            num_prompt,
-            assert_baseline=assert_baseline,
-            sweep_index=i,
-            max_concurrency=concurrency,
-        )
+        assert_result(result, params, num_prompt)

tests/dfx/perf/scripts/run_diffusion_benchmark.py

@@ -13,8 +13,6 @@ A config JSON file may be passed via --test-config-file. If omitted, every ``*.j
   pytest run_diffusion_benchmark.py -m "diffusion"
   pytest run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_qwen_image_vllm_omni.json
 
-Optional: ``--assert-baseline`` compares metrics to the ``baseline`` block in each benchmark entry (default: off).
-
 Optional JSON field ``mark`` is applied as pytest marks on that case via
 ``pytest.param`` (e.g. ``"mark": [{"hardware_marks": {"res": {"cuda": "H100"}, "num_cards": 1}}, "full_model", "diffusion"]``).
 
@@ -46,6 +44,7 @@ from tests.dfx.conftest import (
     get_runtime_resource_label,
     hardware_json_value,
     is_diffusion_perf_config,
+    resolve_baseline_for_sweep,
     resolve_pytest_marks,
     resource_label_for_filename,
 )
@@ -274,11 +273,6 @@ def _write_result_record(record: dict[str, Any]) -> Path:
     return target
 
 
-def _append_to_aggregated_file(record: dict[str, Any]) -> None:
-    """Backward-compatible wrapper; prefer :func:`_write_result_record`."""
-    _write_result_record(record)
-
-
 _server_lock = threading.Lock()
 
 # ---------------------------------------------------------------------------
@@ -678,7 +672,7 @@ def run_benchmark(
     with tempfile.NamedTemporaryFile(mode="w", suffix=".json", prefix="diffusion_bench_tmp_", delete=False) as tmp:
         tmp_result_file = Path(tmp.name)
 
-    exclude_keys = {"baseline", "dataset", "task", "name", "skip-performance-assertion"}
+    exclude_keys = {"baseline", "dataset", "task", "name"}
 
     cmd = [
         sys.executable,
@@ -772,12 +766,17 @@ def run_benchmark(
     completed = metrics.get("completed_requests", metrics.get("completed", 0))
     failed = metrics.get("failed_requests", metrics.get("failed", 0))
 
+    # Persist sweep-resolved baseline from params (already narrowed in _build_run_params).
+    baseline = params.get("baseline") or {}
+    metrics["baseline"] = baseline
+
     record: dict[str, Any] = {
         "test_name": test_name,
         "endpoint": endpoint,
         "timestamp": timestamp,
         "server_params": server_cfg.get("server_params"),
         "benchmark_params": params,
+        "baseline": baseline,
         "result": metrics,
         "log_file": str(log_file),
         "Model": model,
@@ -827,70 +826,14 @@ def run_benchmark(
 # ---------------------------------------------------------------------------
 
 
-def _resolve_baseline_value(
-    baseline_raw: Any,
-    *,
-    sweep_index: int | None,
-    max_concurrency: Any = None,
-    request_rate: Any = None,
-) -> Any:
-    """Pick the baseline threshold for this sweep step."""
-    if baseline_raw is None:
-        return 100000
-    if isinstance(baseline_raw, dict):
-        if max_concurrency is not None:
-            for key in (max_concurrency, str(max_concurrency)):
-                if key in baseline_raw:
-                    return baseline_raw[key]
-        if request_rate is not None:
-            for key in (request_rate, str(request_rate)):
-                if key in baseline_raw:
-                    return baseline_raw[key]
-        raise KeyError(
-            f"baseline dict has no key for max_concurrency={max_concurrency!r} "
-            f"or request_rate={request_rate!r}; keys={list(baseline_raw.keys())!r}"
-        )
-    if isinstance(baseline_raw, (list, tuple)):
-        if sweep_index is None:
-            raise ValueError("sweep_index is required when baseline is a list or tuple")
-        return baseline_raw[sweep_index]
-    return baseline_raw
-
-
 def assert_result(
     result: dict[str, Any],
-    params: dict[str, Any],
     num_prompts: int,
-    *,
-    sweep_index: int | None = None,
-    max_concurrency: Any = None,
-    request_rate: Any = None,
-    assert_baseline: bool = True,
 ) -> None:
-    """Assert that benchmark metrics satisfy the configured baselines."""
+    """Assert that the expected number of requests completed."""
     completed = result.get("completed_requests", result.get("completed", 0))
     assert completed == num_prompts, f"Expected {num_prompts} completed requests, got {completed}"
 
-    if not assert_baseline:
-        return
-    if params.get("skip-performance-assertion", False):
-        print("Skipping performance assertions.")
-        return
-
-    for metric, baseline_raw in params.get("baseline", {}).items():
-        current = result.get(metric)
-        assert current is not None, f"Metric '{metric}' not found in result: {list(result.keys())}"
-        threshold = _resolve_baseline_value(
-            baseline_raw,
-            sweep_index=sweep_index,
-            max_concurrency=max_concurrency,
-            request_rate=request_rate,
-        )
-        if "throughput" in metric:
-            assert current >= threshold, f"{metric}: {current:.4f} < baseline {threshold}"
-        else:
-            assert current <= threshold, f"{metric}: {current:.4f} > baseline {threshold}"
-
 
 def _default_benchmark_endpoint_for_task(task: str) -> str:
     """Return the default client-side benchmark endpoint for a diffusion task."""
@@ -932,15 +875,11 @@ def _build_run_params(
     if max_concurrency is not None:
         run_params["max-concurrency"] = max_concurrency
     if "baseline" in params:
-        run_params["baseline"] = {
-            metric: _resolve_baseline_value(
-                baseline_raw,
-                sweep_index=sweep_index,
-                max_concurrency=max_concurrency,
-                request_rate=request_rate,
-            )
-            for metric, baseline_raw in params["baseline"].items()
-        }
+        # Keep all hardware buckets; pick the metric value for this sweep step only.
+        run_params["baseline"] = resolve_baseline_for_sweep(
+            params.get("baseline"),
+            sweep_index=sweep_index,
+        )
     return run_params
 
 
@@ -963,36 +902,30 @@ def _iter_sweep_runs(params: dict[str, Any]) -> list[dict[str, Any]]:
 
     sweep_runs: list[dict[str, Any]] = []
 
-    for i, (request_rate, num_prompts) in enumerate(zip(request_rate_list, num_prompt_list)):
+    for sweep_index, (request_rate, num_prompts) in enumerate(zip(request_rate_list, num_prompt_list)):
         sweep_runs.append(
             {
                 "params": _build_run_params(
                     params,
                     request_rate=request_rate,
                     num_prompts=num_prompts,
-                    sweep_index=i,
+                    sweep_index=sweep_index,
                 ),
                 "num_prompts": num_prompts,
-                "sweep_index": i,
-                "request_rate": request_rate,
-                "max_concurrency": None,
             }
         )
 
-    for i, (max_concurrency, num_prompts) in enumerate(zip(max_concurrency_list, num_prompt_list)):
+    for sweep_index, (max_concurrency, num_prompts) in enumerate(zip(max_concurrency_list, num_prompt_list)):
         sweep_runs.append(
             {
                 "params": _build_run_params(
                     params,
                     max_concurrency=max_concurrency,
                     num_prompts=num_prompts,
-                    sweep_index=i,
                     request_rate="inf",
+                    sweep_index=sweep_index,
                 ),
                 "num_prompts": num_prompts,
-                "sweep_index": i,
-                "request_rate": None,
-                "max_concurrency": max_concurrency,
             }
         )
 
@@ -1003,11 +936,9 @@ def _iter_sweep_runs(params: dict[str, Any]) -> list[dict[str, Any]]:
                 "params": _build_run_params(
                     params,
                     num_prompts=default_num_prompts,
+                    sweep_index=0,
                 ),
                 "num_prompts": default_num_prompts,
-                "sweep_index": None,
-                "request_rate": None,
-                "max_concurrency": None,
             }
         )
 
@@ -1023,14 +954,12 @@ def _iter_sweep_runs(params: dict[str, Any]) -> list[dict[str, Any]]:
     paired_benchmark_params,
     indirect=["diffusion_server", "benchmark_params"],
 )
-def test_diffusion_performance_benchmark(diffusion_server, benchmark_params, request):
+def test_diffusion_performance_benchmark(diffusion_server, benchmark_params):
     """Run the diffusion performance benchmark and verify request completion.
 
     One server is started per unique parallel configuration (module scope).
     For each server, all benchmark parameter sets defined in the config JSON
     are executed sequentially; metrics are recorded to the aggregated result file.
-
-    Pass ``--assert-baseline`` to compare ``throughput_qps``, ``latency_mean``, etc. to the JSON ``baseline`` block.
     """
     test_name = benchmark_params["test_name"]
     params = benchmark_params["params"]
@@ -1073,14 +1002,4 @@ def test_diffusion_performance_benchmark(diffusion_server, benchmark_params, req
             print(f"\n  Aggregated results: {_aggregated_result_file_for_source(str(source))}")
         print("=" * 60)
 
-        assert_baseline = request.config.getoption("--assert-baseline", default=False)
-
-        assert_result(
-            result,
-            params,
-            sweep_run["num_prompts"],
-            sweep_index=sweep_run["sweep_index"],
-            max_concurrency=sweep_run["max_concurrency"],
-            request_rate=sweep_run["request_rate"],
-            assert_baseline=assert_baseline,
-        )
+        assert_result(result, sweep_run["num_prompts"])

tests/dfx/perf/tests/test_bagel_vllm_omni.json

@@ -40,10 +40,12 @@
         "max-concurrency": 1,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.495,
-          "latency_mean": 2.0238,
-          "peak_memory_mb_max": 29393.4286,
-          "peak_memory_mb_mean": 29392.7143
+          "H100": {
+            "throughput_qps": 0.495,
+            "latency_mean": 2.0238,
+            "peak_memory_mb_max": 29393.4286,
+            "peak_memory_mb_mean": 29392.7143
+          }
         }
       }
     ]
@@ -90,10 +92,12 @@
         "enable-negative-prompt": true,
         "num-input-images": 1,
         "baseline": {
-          "throughput_qps": 0.3451,
-          "latency_mean": 2.9031,
-          "peak_memory_mb_max": 30480.2857,
-          "peak_memory_mb_mean": 30479.2857
+          "H100": {
+            "throughput_qps": 0.3451,
+            "latency_mean": 2.9031,
+            "peak_memory_mb_max": 30480.2857,
+            "peak_memory_mb_mean": 30479.2857
+          }
         }
       }
     ]
@@ -139,10 +143,12 @@
         "max-concurrency": 1,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.5048,
-          "latency_mean": 1.987,
-          "peak_memory_mb_max": 29380.0,
-          "peak_memory_mb_mean": 29380.0
+          "H100": {
+            "throughput_qps": 0.5048,
+            "latency_mean": 1.987,
+            "peak_memory_mb_max": 29380.0,
+            "peak_memory_mb_mean": 29380.0
+          }
         }
       }
     ]
@@ -189,10 +195,12 @@
         "enable-negative-prompt": true,
         "num-input-images": 1,
         "baseline": {
-          "throughput_qps": 0.149,
-          "latency_mean": 6.7157,
-          "peak_memory_mb_max": 30574.0,
-          "peak_memory_mb_mean": 30574.0
+          "H100": {
+            "throughput_qps": 0.149,
+            "latency_mean": 6.7157,
+            "peak_memory_mb_max": 30574.0,
+            "peak_memory_mb_mean": 30574.0
+          }
         }
       }
     ]

tests/dfx/perf/tests/test_boogu_image_edit_vllm_omni.json

@@ -24,10 +24,12 @@
                 "warmup-num-inference-steps": 28,
                 "enable-negative-prompt": true,
                 "baseline": {
-                    "throughput_qps": {"1": 0.0303, "2": 0.0302, "4": 0.0302},
-                    "latency_mean": {"1": 32.6, "2": 62.2, "4": 111.7},
-                    "peak_memory_mb_max": 40500,
-                    "peak_memory_mb_mean": 40500
+                    "H100": {
+                        "throughput_qps": [0.0303, 0.0302, 0.0302],
+                        "latency_mean": [32.6, 62.2, 111.7],
+                        "peak_memory_mb_max": 40500,
+                        "peak_memory_mb_mean": 40500
+                    }
                 }
             }
         ]

tests/dfx/perf/tests/test_boogu_image_vllm_omni.json

@@ -24,10 +24,12 @@
                 "warmup-num-inference-steps": 28,
                 "enable-negative-prompt": true,
                 "baseline": {
-                    "throughput_qps": {"1": 0.0549, "2": 0.0547, "4": 0.0549},
-                    "latency_mean": {"1": 18.0, "2": 34.3, "4": 61.4},
-                    "peak_memory_mb_max": 40500,
-                    "peak_memory_mb_mean": 40500
+                    "H100": {
+                        "throughput_qps": [0.0549, 0.0547, 0.0549],
+                        "latency_mean": [18.0, 34.3, 61.4],
+                        "peak_memory_mb_max": 40500,
+                        "peak_memory_mb_mean": 40500
+                    }
                 }
             }
         ]

tests/dfx/perf/tests/test_cosmos3_vllm_omni.json

@@ -48,9 +48,11 @@
           "guardrails": false
         },
         "baseline": {
-          "throughput_qps": 1.107,
-          "latency_mean": 0.8929,
-          "peak_memory_mb_mean": 80000.0
+          "H100": {
+            "throughput_qps": 1.107,
+            "latency_mean": 0.8929,
+            "peak_memory_mb_mean": 80000.0
+          }
         }
       }
     ]
@@ -115,9 +117,11 @@
           "use_duration_template": false
         },
         "baseline": {
-          "throughput_qps": 0.027,
-          "latency_mean": 36.6824,
-          "peak_memory_mb_mean": 30155.4
+          "H100": {
+            "throughput_qps": 0.027,
+            "latency_mean": 36.6824,
+            "peak_memory_mb_mean": 30155.4
+          }
         }
       }
     ]
@@ -181,9 +185,11 @@
           "guardrails": false
         },
         "baseline": {
-          "throughput_qps": 0.027,
-          "latency_mean": 36.968,
-          "peak_memory_mb_mean": 30443.6
+          "H100": {
+            "throughput_qps": 0.027,
+            "latency_mean": 36.968,
+            "peak_memory_mb_mean": 30443.6
+          }
         }
       }
     ]
@@ -251,9 +257,11 @@
           "condition_video_keep": "first"
         },
         "baseline": {
-          "throughput_qps": 0.027,
-          "latency_mean": 37.2117,
-          "peak_memory_mb_mean": 30135.6
+          "H100": {
+            "throughput_qps": 0.027,
+            "latency_mean": 37.2117,
+            "peak_memory_mb_mean": 30135.6
+          }
         }
       }
     ]

tests/dfx/perf/tests/test_higgs_audio_v3.json

@@ -46,12 +46,14 @@
         "seed_tts_locale": "en",
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "median_audio_ttfp_ms": [
-            223.3584
-          ],
-          "median_audio_rtf": [
-            0.4393
-          ]
+          "H100": {
+            "median_audio_ttfp_ms": [
+              223.3584
+            ],
+            "median_audio_rtf": [
+              0.4393
+            ]
+          }
         }
       },
       {
@@ -70,15 +72,17 @@
         "seed_tts_locale": "en",
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "median_audio_ttfp_ms": [
-            276.575
-          ],
-          "median_audio_rtf": [
-            0.5073
-          ],
-          "audio_throughput": [
-            27.3728
-          ]
+          "H100": {
+            "median_audio_ttfp_ms": [
+              276.575
+            ],
+            "median_audio_rtf": [
+              0.5073
+            ],
+            "audio_throughput": [
+              27.3728
+            ]
+          }
         }
       }
     ]

tests/dfx/perf/tests/test_hunyuan_image3_it2i.json

@@ -140,11 +140,13 @@
         "num-prompts": 96,
         "skip-performance-assertion": true,
         "baseline": {
-          "throughput_qps": 0.1874,
-          "latency_p99": 101.852,
-          "latency_mean": 79.0385,
-          "peak_memory_mb_max": 70000,
-          "peak_memory_mb_mean": 70000
+          "H200": {
+            "throughput_qps": 0.1874,
+            "latency_p99": 101.852,
+            "latency_mean": 79.0385,
+            "peak_memory_mb_max": 70000,
+            "peak_memory_mb_mean": 70000
+          }
         }
       }
     ]

tests/dfx/perf/tests/test_hunyuan_image_tp2_cfgp2.json

@@ -36,10 +36,12 @@
         "max-concurrency": 1,
         "skip-performance-assertion": true,
         "baseline": {
-          "throughput_qps": 0.21,
-          "latency_p99": 4.7469,
-          "peak_memory_mb_max": 101100,
-          "peak_memory_mb_mean": 101100
+          "H200": {
+            "throughput_qps": 0.21,
+            "latency_p99": 4.7469,
+            "peak_memory_mb_max": 101100,
+            "peak_memory_mb_mean": 101100
+          }
         }
       }
     ]

tests/dfx/perf/tests/test_hunyuan_image_tp2_sp2.json

@@ -36,10 +36,12 @@
         "max-concurrency": 1,
         "skip-performance-assertion": true,
         "baseline": {
-          "throughput_qps": 0.20,
-          "latency_p99": 5.1025,
-          "peak_memory_mb_max": 97402,
-          "peak_memory_mb_mean": 97402
+          "H200": {
+            "throughput_qps": 0.2,
+            "latency_p99": 5.1025,
+            "peak_memory_mb_max": 97402,
+            "peak_memory_mb_mean": 97402
+          }
         }
       }
     ]

tests/dfx/perf/tests/test_hunyuan_image_tp4.json

@@ -35,11 +35,13 @@
         "max-concurrency": 1,
         "skip-performance-assertion": true,
         "baseline": {
-          "throughput_qps": 0.213,
-          "latency_mean": 4.6953,
-          "latency_p99": 4.8601,
-          "peak_memory_mb_max": 56912.0,
-          "peak_memory_mb_mean": 56912.0
+          "H200": {
+            "throughput_qps": 0.213,
+            "latency_mean": 4.6953,
+            "latency_p99": 4.8601,
+            "peak_memory_mb_max": 56912.0,
+            "peak_memory_mb_mean": 56912.0
+          }
         }
       }
     ]

tests/dfx/perf/tests/test_lingbot_video_vllm_omni.json

@@ -26,8 +26,7 @@
                 "warmup-requests": 1,
                 "warmup-concurrency": 1,
                 "warmup-num-inference-steps": 2,
-                "enable-negative-prompt": true,
-                "skip-performance-assertion": true
+                "enable-negative-prompt": true
             }
         ]
     }

tests/dfx/perf/tests/test_minicpmo_4_5.json

@@ -0,0 +1,144 @@
+[
+  {
+    "test_name": "test_minicpmo_4_5_challenge",
+    "mark": [
+      {
+        "hardware_marks": {
+          "res": {
+            "cuda": "H100",
+            "npu": "A3"
+          },
+          "num_cards": 1
+        }
+      },
+      "full_model",
+      "omni"
+    ],
+    "server_params": {
+      "model": "openbmb/MiniCPM-o-4_5",
+      "extra_cli_args": [
+        "--deploy-config",
+        "vllm_omni/deploy/minicpmo_4_5.yaml",
+        "--trust-remote-code"
+      ]
+    },
+    "benchmark_params": [
+      {
+        "dataset_name": "seed-tts",
+        "dataset_path": "zhaochenyang20/seed-tts-eval",
+        "backend": "openai-chat-omni",
+        "endpoint": "/v1/chat/completions",
+        "num_prompts": [
+          32,
+          64,
+          128
+        ],
+        "max_concurrency": [
+          1,
+          4,
+          8
+        ],
+        "no_oversample": true,
+        "disable_shuffle": true,
+        "trust_remote_code": true,
+        "seed_tts_locale": "zh",
+        "extra_body": {
+          "modalities": [
+            "text",
+            "audio"
+          ],
+          "chat_template_kwargs": {
+            "enable_thinking": false,
+            "use_tts_template": true
+          }
+        },
+        "percentile-metrics": "ttft,e2el,audio_ttfp,audio_rtf,audio_duration",
+        "baseline": {
+          "A3": {
+            "request_throughput": [
+              0.5383,
+              0.6042,
+              0.7547
+            ],
+            "mean_ttft_ms": [
+              333.2633,
+              533.8656,
+              547.3388
+            ],
+            "mean_e2el_ms": [
+              1857.2154,
+              6600.5101,
+              10450.6811
+            ],
+            "mean_audio_ttfp_ms": [
+              986.4666,
+              3411.0754,
+              3352.7515
+            ],
+            "mean_audio_rtf": [
+              0.4423,
+              1.5734,
+              2.3024
+            ]
+          }
+        }
+      }
+    ]
+  },
+  {
+    "test_name": "test_minicpmo_4_5_a2_challenge",
+    "mark": [
+      {
+        "hardware_marks": {
+          "res": {
+            "npu": "A2"
+          },
+          "num_cards": 1
+        }
+      },
+      "full_model",
+      "omni"
+    ],
+    "server_params": {
+      "model": "openbmb/MiniCPM-o-4_5",
+      "extra_cli_args": [
+        "--deploy-config",
+        "vllm_omni/deploy/minicpmo_4_5.yaml",
+        "--trust-remote-code"
+      ]
+    },
+    "benchmark_params": [
+      {
+        "dataset_name": "seed-tts",
+        "dataset_path": "zhaochenyang20/seed-tts-eval",
+        "backend": "openai-chat-omni",
+        "endpoint": "/v1/chat/completions",
+        "num_prompts": [
+          32,
+          64,
+          128
+        ],
+        "max_concurrency": [
+          1,
+          4,
+          8
+        ],
+        "no_oversample": true,
+        "disable_shuffle": true,
+        "trust_remote_code": true,
+        "seed_tts_locale": "zh",
+        "extra_body": {
+          "modalities": [
+            "text",
+            "audio"
+          ],
+          "chat_template_kwargs": {
+            "enable_thinking": false,
+            "use_tts_template": true
+          }
+        },
+        "percentile-metrics": "ttft,e2el,audio_ttfp,audio_rtf,audio_duration"
+      }
+    ]
+  }
+]

tests/dfx/perf/tests/test_minicpmo_4_5_duplex_seed_tts.json

@@ -0,0 +1,69 @@
+[
+  {
+    "test_name": "test_minicpmo_4_5_duplex_seed_tts_challenge",
+    "mark": [
+      {
+        "hardware_marks": {
+          "res": {
+            "cuda": "H100",
+            "npu": "A3"
+          },
+          "num_cards": 1
+        }
+      },
+      "full_model",
+      "omni"
+    ],
+    "server_params": {
+      "model": "openbmb/MiniCPM-o-4_5",
+      "extra_cli_args": [
+        "--deploy-config",
+        "vllm_omni/deploy/minicpmo_4_5_duplex.yaml",
+        "--trust-remote-code"
+      ]
+    },
+    "benchmark_params": [
+      {
+        "dataset_name": "seed-tts",
+        "dataset_path": "zhaochenyang20/seed-tts-eval",
+        "backend": "openai-realtime-tts",
+        "endpoint": "/v1/realtime",
+        "num_prompts": [
+          32
+        ],
+        "max_concurrency": [
+          1
+        ],
+        "no_oversample": true,
+        "disable_shuffle": true,
+        "trust_remote_code": true,
+        "seed_tts_locale": "zh",
+        "seed_tts_turns_per_session": 4,
+        "expected_duplex_audio_turns_per_session": 4,
+        "extra_body": {
+          "save_duplex_request_metrics": true
+        },
+        "percentile-metrics": "ttft,e2el,audio_ttfp,audio_rtf,audio_duration",
+        "baseline": {
+          "A3": {
+            "request_throughput": [
+              0.1594
+            ],
+            "mean_ttft_ms": [
+              274.164
+            ],
+            "mean_e2el_ms": [
+              6250.7696
+            ],
+            "mean_audio_ttfp_ms": [
+              994.422
+            ],
+            "mean_audio_rtf": [
+              0.46163
+            ]
+          }
+        }
+      }
+    ]
+  }
+]

tests/dfx/perf/tests/test_minicpmo_4_5_ready.json

@@ -0,0 +1,81 @@
+[
+  {
+    "test_name": "test_minicpmo_4_5_ready",
+    "mark": [
+      {
+        "hardware_marks": {
+          "res": {
+            "cuda": "H100",
+            "npu": "A3"
+          },
+          "num_cards": 1
+        }
+      },
+      "core_model",
+      "omni"
+    ],
+    "server_params": {
+      "model": "openbmb/MiniCPM-o-4_5",
+      "extra_cli_args": [
+        "--deploy-config",
+        "vllm_omni/deploy/minicpmo_4_5.yaml",
+        "--trust-remote-code"
+      ]
+    },
+    "benchmark_params": [
+      {
+        "dataset_name": "seed-tts",
+        "dataset_path": "zhaochenyang20/seed-tts-eval",
+        "backend": "openai-chat-omni",
+        "endpoint": "/v1/chat/completions",
+        "num_prompts": [
+          10,
+          40
+        ],
+        "max_concurrency": [
+          1,
+          4
+        ],
+        "no_oversample": true,
+        "disable_shuffle": true,
+        "trust_remote_code": true,
+        "seed_tts_locale": "zh",
+        "extra_body": {
+          "modalities": [
+            "text",
+            "audio"
+          ],
+          "chat_template_kwargs": {
+            "enable_thinking": false,
+            "use_tts_template": true
+          }
+        },
+        "percentile-metrics": "ttft,e2el,audio_ttfp,audio_rtf,audio_duration",
+        "baseline": {
+          "H100": {
+            "request_throughput": [
+              0.7803,
+              1.1008
+            ],
+            "mean_ttft_ms": [
+              166.9764,
+              307.7133
+            ],
+            "mean_e2el_ms": [
+              1361.1682,
+              3800.6332
+            ],
+            "mean_audio_ttfp_ms": [
+              635.8099,
+              1615.6591
+            ],
+            "mean_audio_rtf": [
+              0.3151,
+              0.8568
+            ]
+          }
+        }
+      }
+    ]
+  }
+]

tests/dfx/perf/tests/test_qwen3_omni_async_chunk.json

@@ -44,27 +44,29 @@
         "ignore_eos": true,
         "percentile-metrics": "ttft,tpot,itl,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "mean_ttft_ms": [
-            96.4061,
-            140.8147,
-            271.9025,
-            362.2916,
-            507.776
-          ],
-          "mean_e2el_ms": [
-            18507.4582,
-            28365.4956,
-            31907.9613,
-            48161.5829,
-            72630.8601
-          ],
-          "mean_audio_rtf": [
-            0.1175,
-            0.18,
-            0.2359,
-            0.3278,
-            0.6296
-          ]
+          "H100": {
+            "mean_ttft_ms": [
+              96.4061,
+              140.8147,
+              271.9025,
+              362.2916,
+              507.776
+            ],
+            "mean_e2el_ms": [
+              18507.4582,
+              28365.4956,
+              31907.9613,
+              48161.5829,
+              72630.8601
+            ],
+            "mean_audio_rtf": [
+              0.1175,
+              0.18,
+              0.2359,
+              0.3278,
+              0.6296
+            ]
+          }
         }
       },
       {
@@ -91,15 +93,17 @@
         },
         "percentile-metrics": "ttft,tpot,itl,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "mean_ttft_ms": [
-            354.0951
-          ],
-          "mean_e2el_ms": [
-            3952.5615
-          ],
-          "mean_audio_rtf": [
-            0.117
-          ]
+          "H100": {
+            "mean_ttft_ms": [
+              354.0951
+            ],
+            "mean_e2el_ms": [
+              3952.5615
+            ],
+            "mean_audio_rtf": [
+              0.117
+            ]
+          }
         }
       },
       {
@@ -128,15 +132,17 @@
         },
         "percentile-metrics": "ttft,tpot,itl,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "mean_ttft_ms": [
-            259.6317
-          ],
-          "mean_e2el_ms": [
-            4253.3009
-          ],
-          "mean_audio_rtf": [
-            0.1448
-          ]
+          "H100": {
+            "mean_ttft_ms": [
+              259.6317
+            ],
+            "mean_e2el_ms": [
+              4253.3009
+            ],
+            "mean_audio_rtf": [
+              0.1448
+            ]
+          }
         }
       },
       {
@@ -167,15 +173,17 @@
         },
         "percentile-metrics": "ttft,tpot,itl,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "mean_ttft_ms": [
-            457.4453
-          ],
-          "mean_e2el_ms": [
-            5912.5623
-          ],
-          "mean_audio_rtf": [
-            0.1967
-          ]
+          "H100": {
+            "mean_ttft_ms": [
+              457.4453
+            ],
+            "mean_e2el_ms": [
+              5912.5623
+            ],
+            "mean_audio_rtf": [
+              0.1967
+            ]
+          }
         }
       },
       {
@@ -206,20 +214,22 @@
         },
         "percentile-metrics": "ttft,tpot,itl,e2el",
         "baseline": {
-          "mean_ttft_ms": [
-            90.8033,
-            202.1779,
-            326.5864,
-            544.8914,
-            961.3929
-          ],
-          "mean_e2el_ms": [
-            4517.3411,
-            7857.9711,
-            9861.1687,
-            13283.3363,
-            19541.0736
-          ]
+          "H100": {
+            "mean_ttft_ms": [
+              90.8033,
+              202.1779,
+              326.5864,
+              544.8914,
+              961.3929
+            ],
+            "mean_e2el_ms": [
+              4517.3411,
+              7857.9711,
+              9861.1687,
+              13283.3363,
+              19541.0736
+            ]
+          }
         }
       },
       {
@@ -251,12 +261,14 @@
         },
         "percentile-metrics": "ttft,tpot,itl,e2el",
         "baseline": {
-          "mean_ttft_ms": [
-            318.048
-          ],
-          "mean_e2el_ms": [
-            807.6114
-          ]
+          "H100": {
+            "mean_ttft_ms": [
+              318.048
+            ],
+            "mean_e2el_ms": [
+              807.6114
+            ]
+          }
         }
       },
       {
@@ -290,12 +302,14 @@
         },
         "percentile-metrics": "ttft,tpot,itl,e2el",
         "baseline": {
-          "mean_ttft_ms": [
-            186.168
-          ],
-          "mean_e2el_ms": [
-            726.1137
-          ]
+          "H100": {
+            "mean_ttft_ms": [
+              186.168
+            ],
+            "mean_e2el_ms": [
+              726.1137
+            ]
+          }
         }
       },
       {
@@ -331,12 +345,14 @@
         },
         "percentile-metrics": "ttft,tpot,itl,e2el",
         "baseline": {
-          "mean_ttft_ms": [
-            375.5282
-          ],
-          "mean_e2el_ms": [
-            1006.1663
-          ]
+          "H100": {
+            "mean_ttft_ms": [
+              375.5282
+            ],
+            "mean_e2el_ms": [
+              1006.1663
+            ]
+          }
         }
       }
     ]

tests/dfx/perf/tests/test_qwen3_omni_multi_replicas.json

@@ -35,10 +35,12 @@
         "ignore_eos": true,
         "percentile-metrics": "ttft,tpot,itl,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-            "mean_ttft_ms": [130.5751, 278.6469, 324.7249, 911.6085],
-            "mean_tpot_ms": [7.9854, 11.1209, 16.1507, 20.7485],
-            "mean_audio_ttfp_ms": [486.486, 954.1791, 1289.6514, 2196.8275],
-            "mean_audio_rtf": [0.1896, 0.3962, 0.4098, 0.7322]
+            "H100": {
+                "mean_ttft_ms": [130.5751, 278.6469, 324.7249, 911.6085],
+                "mean_tpot_ms": [7.9854, 11.1209, 16.1507, 20.7485],
+                "mean_audio_ttfp_ms": [486.486, 954.1791, 1289.6514, 2196.8275],
+                "mean_audio_rtf": [0.1896, 0.3962, 0.4098, 0.7322]
+            }
         }
       },
       {
@@ -61,9 +63,11 @@
         },
         "percentile-metrics": "ttft,tpot,itl,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-            "mean_ttft_ms": [340.5122],
-            "mean_audio_ttfp_ms": [463.0752],
-            "mean_audio_rtf": [0.1153]
+            "H100": {
+                "mean_ttft_ms": [340.5122],
+                "mean_audio_ttfp_ms": [463.0752],
+                "mean_audio_rtf": [0.1153]
+            }
         }
       },
       {
@@ -88,9 +92,11 @@
         },
         "percentile-metrics": "ttft,tpot,itl,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-            "mean_ttft_ms": [236.926],
-            "mean_audio_ttfp_ms": [395.2182],
-            "mean_audio_rtf": [0.1259]
+            "H100": {
+                "mean_ttft_ms": [236.926],
+                "mean_audio_ttfp_ms": [395.2182],
+                "mean_audio_rtf": [0.1259]
+            }
         }
       },
       {
@@ -117,9 +123,11 @@
         },
         "percentile-metrics": "ttft,tpot,itl,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-            "mean_ttft_ms": [470.6126],
-            "mean_audio_ttfp_ms": [665.5474],
-            "mean_audio_rtf": [0.166]
+            "H100": {
+                "mean_ttft_ms": [470.6126],
+                "mean_audio_ttfp_ms": [665.5474],
+                "mean_audio_rtf": [0.166]
+            }
         }
       }
     ]

tests/dfx/perf/tests/test_qwen3_omni_no_async_chunk.json

@@ -44,27 +44,29 @@
         "ignore_eos": true,
         "percentile-metrics": "ttft,tpot,itl,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "mean_ttft_ms": [
-            115.8348,
-            173.1823,
-            446.672,
-            464.5985,
-            3420.0634
-          ],
-          "mean_e2el_ms": [
-            25070.9523,
-            31992.0646,
-            37619.7669,
-            50393.8257,
-            136619.2743
-          ],
-          "mean_audio_rtf": [
-            0.169,
-            0.2111,
-            0.2475,
-            0.3551,
-            0.9266
-          ]
+          "H100": {
+            "mean_ttft_ms": [
+              115.8348,
+              173.1823,
+              446.672,
+              464.5985,
+              3420.0634
+            ],
+            "mean_e2el_ms": [
+              25070.9523,
+              31992.0646,
+              37619.7669,
+              50393.8257,
+              136619.2743
+            ],
+            "mean_audio_rtf": [
+              0.169,
+              0.2111,
+              0.2475,
+              0.3551,
+              0.9266
+            ]
+          }
         }
       },
       {
@@ -91,15 +93,17 @@
         },
         "percentile-metrics": "ttft,tpot,itl,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "mean_ttft_ms": [
-            343.1377
-          ],
-          "mean_e2el_ms": [
-            4811.1123
-          ],
-          "mean_audio_rtf": [
-            0.1463
-          ]
+          "H100": {
+            "mean_ttft_ms": [
+              343.1377
+            ],
+            "mean_e2el_ms": [
+              4811.1123
+            ],
+            "mean_audio_rtf": [
+              0.1463
+            ]
+          }
         }
       },
       {
@@ -128,15 +132,17 @@
         },
         "percentile-metrics": "ttft,tpot,itl,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "mean_ttft_ms": [
-            245.5446
-          ],
-          "mean_e2el_ms": [
-            5019.1256
-          ],
-          "mean_audio_rtf": [
-            0.1713
-          ]
+          "H100": {
+            "mean_ttft_ms": [
+              245.5446
+            ],
+            "mean_e2el_ms": [
+              5019.1256
+            ],
+            "mean_audio_rtf": [
+              0.1713
+            ]
+          }
         }
       },
       {
@@ -167,15 +173,17 @@
         },
         "percentile-metrics": "ttft,tpot,itl,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "mean_ttft_ms": [
-            506.1407
-          ],
-          "mean_e2el_ms": [
-            6814.9656
-          ],
-          "mean_audio_rtf": [
-            0.2213
-          ]
+          "H100": {
+            "mean_ttft_ms": [
+              506.1407
+            ],
+            "mean_e2el_ms": [
+              6814.9656
+            ],
+            "mean_audio_rtf": [
+              0.2213
+            ]
+          }
         }
       }
     ]

tests/dfx/perf/tests/test_qwen3_omni_vllm_text.json

@@ -51,20 +51,22 @@
         },
         "percentile-metrics": "ttft,tpot,itl,e2el",
         "baseline": {
-          "mean_ttft_ms": [
-            69.8081,
-            173.0651,
-            271.8108,
-            413.4079,
-            497.2546
-          ],
-          "mean_e2el_ms": [
-            4720.8225,
-            7937.2532,
-            9962.3566,
-            13724.5479,
-            18324.091
-          ]
+          "H100": {
+            "mean_ttft_ms": [
+              69.8081,
+              173.0651,
+              271.8108,
+              413.4079,
+              497.2546
+            ],
+            "mean_e2el_ms": [
+              4720.8225,
+              7937.2532,
+              9962.3566,
+              13724.5479,
+              18324.091
+            ]
+          }
         }
       },
       {
@@ -97,12 +99,14 @@
         },
         "percentile-metrics": "ttft,tpot,itl,e2el",
         "baseline": {
-          "mean_ttft_ms": [
-            25.9477
-          ],
-          "mean_e2el_ms": [
-            517.3155
-          ]
+          "H100": {
+            "mean_ttft_ms": [
+              25.9477
+            ],
+            "mean_e2el_ms": [
+              517.3155
+            ]
+          }
         }
       },
       {
@@ -137,12 +141,14 @@
         },
         "percentile-metrics": "ttft,tpot,itl,e2el",
         "baseline": {
-          "mean_ttft_ms": [
-            26.7056
-          ],
-          "mean_e2el_ms": [
-            551.308
-          ]
+          "H100": {
+            "mean_ttft_ms": [
+              26.7056
+            ],
+            "mean_e2el_ms": [
+              551.308
+            ]
+          }
         }
       },
       {
@@ -179,12 +185,14 @@
         },
         "percentile-metrics": "ttft,tpot,itl,e2el",
         "baseline": {
-          "mean_ttft_ms": [
-            27.6906
-          ],
-          "mean_e2el_ms": [
-            596.5284
-          ]
+          "H100": {
+            "mean_ttft_ms": [
+              27.6906
+            ],
+            "mean_e2el_ms": [
+              596.5284
+            ]
+          }
         }
       }
     ]

tests/dfx/perf/tests/test_qwen_image_edit_2511_vllm_omni.json

@@ -34,10 +34,12 @@
         "num-input-images": 2,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.0721,
-          "latency_mean": 13.8398,
-          "peak_memory_mb_max": 57610.0,
-          "peak_memory_mb_mean": 57610.0
+          "H100": {
+            "throughput_qps": 0.0721,
+            "latency_mean": 13.8398,
+            "peak_memory_mb_max": 57610.0,
+            "peak_memory_mb_mean": 57610.0
+          }
         }
       },
       {
@@ -52,10 +54,12 @@
         "num-input-images": 2,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.0186,
-          "latency_mean": 53.6942,
-          "peak_memory_mb_max": 67790.0,
-          "peak_memory_mb_mean": 67790.0
+          "H100": {
+            "throughput_qps": 0.0186,
+            "latency_mean": 53.6942,
+            "peak_memory_mb_max": 67790.0,
+            "peak_memory_mb_mean": 67790.0
+          }
         }
       }
     ]
@@ -100,10 +104,12 @@
         "num-input-images": 2,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.0538,
-          "latency_mean": 18.5591,
-          "peak_memory_mb_max": 56449.4286,
-          "peak_memory_mb_mean": 56449.4286
+          "H100": {
+            "throughput_qps": 0.0538,
+            "latency_mean": 18.5591,
+            "peak_memory_mb_max": 56449.4286,
+            "peak_memory_mb_mean": 56449.4286
+          }
         }
       }
     ]
@@ -157,10 +163,12 @@
         "num-input-images": 2,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.2776,
-          "latency_mean": 3.5938,
-          "peak_memory_mb_max": 58113.7143,
-          "peak_memory_mb_mean": 58075.1429
+          "H100": {
+            "throughput_qps": 0.2776,
+            "latency_mean": 3.5938,
+            "peak_memory_mb_max": 58113.7143,
+            "peak_memory_mb_mean": 58075.1429
+          }
         }
       },
       {
@@ -175,10 +183,12 @@
         "num-input-images": 2,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.1051,
-          "latency_mean": 9.4918,
-          "peak_memory_mb_max": 68107.4286,
-          "peak_memory_mb_mean": 68107.4286
+          "H100": {
+            "throughput_qps": 0.1051,
+            "latency_mean": 9.4918,
+            "peak_memory_mb_max": 68107.4286,
+            "peak_memory_mb_mean": 68107.4286
+          }
         }
       }
     ]

tests/dfx/perf/tests/test_qwen_image_edit_vllm_omni.json

@@ -33,10 +33,12 @@
         "max-concurrency": 1,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.1129,
-          "latency_mean": 8.8352,
-          "peak_memory_mb_max": 58572.0,
-          "peak_memory_mb_mean": 58572.0
+          "H100": {
+            "throughput_qps": 0.1129,
+            "latency_mean": 8.8352,
+            "peak_memory_mb_max": 58572.0,
+            "peak_memory_mb_mean": 58572.0
+          }
         }
       },
       {
@@ -50,10 +52,12 @@
         "max-concurrency": 1,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.024,
-          "latency_mean": 41.5686,
-          "peak_memory_mb_max": 66368.0,
-          "peak_memory_mb_mean": 66368.0
+          "H100": {
+            "throughput_qps": 0.024,
+            "latency_mean": 41.5686,
+            "peak_memory_mb_max": 66368.0,
+            "peak_memory_mb_mean": 66368.0
+          }
         }
       }
     ]
@@ -96,10 +100,12 @@
         "max-concurrency": 1,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.0672,
-          "latency_mean": 14.8517,
-          "peak_memory_mb_max": 57076.0,
-          "peak_memory_mb_mean": 57076.0
+          "H100": {
+            "throughput_qps": 0.0672,
+            "latency_mean": 14.8517,
+            "peak_memory_mb_max": 57076.0,
+            "peak_memory_mb_mean": 57076.0
+          }
         }
       }
     ]
@@ -152,10 +158,12 @@
         "max-concurrency": 1,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.2879,
-          "latency_mean": 3.4701,
-          "peak_memory_mb_max": 58758.0,
-          "peak_memory_mb_mean": 58758.0
+          "H100": {
+            "throughput_qps": 0.2879,
+            "latency_mean": 3.4701,
+            "peak_memory_mb_max": 58758.0,
+            "peak_memory_mb_mean": 58758.0
+          }
         }
       },
       {
@@ -169,10 +177,12 @@
         "max-concurrency": 1,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.1113,
-          "latency_mean": 8.9656,
-          "peak_memory_mb_max": 67424.0,
-          "peak_memory_mb_mean": 67424.0
+          "H100": {
+            "throughput_qps": 0.1113,
+            "latency_mean": 8.9656,
+            "peak_memory_mb_max": 67424.0,
+            "peak_memory_mb_mean": 67424.0
+          }
         }
       }
     ]

tests/dfx/perf/tests/test_qwen_image_layered_vllm_omni.json

@@ -33,10 +33,12 @@
         "max-concurrency": 1,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.0703,
-          "latency_mean": 14.208,
-          "peak_memory_mb_max": 63225.1429,
-          "peak_memory_mb_mean": 63225.1429
+          "H100": {
+            "throughput_qps": 0.0703,
+            "latency_mean": 14.208,
+            "peak_memory_mb_max": 63225.1429,
+            "peak_memory_mb_mean": 63225.1429
+          }
         }
       },
       {
@@ -50,10 +52,12 @@
         "max-concurrency": 1,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.0417,
-          "latency_mean": 23.914,
-          "peak_memory_mb_max": 63225.1429,
-          "peak_memory_mb_mean": 63225.1429
+          "H100": {
+            "throughput_qps": 0.0417,
+            "latency_mean": 23.914,
+            "peak_memory_mb_max": 63225.1429,
+            "peak_memory_mb_mean": 63225.1429
+          }
         }
       }
     ]

tests/dfx/perf/tests/test_qwen_image_vllm_omni.json

@@ -33,9 +33,11 @@
         "max-concurrency": 1,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.4153,
-          "latency_mean": 2.4065,
-          "peak_memory_mb_mean": 55121.4286
+          "H100": {
+            "throughput_qps": 0.4153,
+            "latency_mean": 2.4065,
+            "peak_memory_mb_mean": 55121.4286
+          }
         }
       },
       {
@@ -49,9 +51,11 @@
         "max-concurrency": 1,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.0423,
-          "latency_mean": 23.5991,
-          "peak_memory_mb_mean": 65118.0
+          "H100": {
+            "throughput_qps": 0.0423,
+            "latency_mean": 23.5991,
+            "peak_memory_mb_mean": 65118.0
+          }
         }
       }
     ]
@@ -91,9 +95,11 @@
         "max-concurrency": 1,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.4175,
-          "latency_mean": 2.3953,
-          "peak_memory_mb_mean": 55088.5714
+          "H100": {
+            "throughput_qps": 0.4175,
+            "latency_mean": 2.3953,
+            "peak_memory_mb_mean": 55088.5714
+          }
         }
       },
       {
@@ -107,9 +113,11 @@
         "max-concurrency": 1,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.0423,
-          "latency_mean": 23.5768,
-          "peak_memory_mb_mean": 66012.2857
+          "H100": {
+            "throughput_qps": 0.0423,
+            "latency_mean": 23.5768,
+            "peak_memory_mb_mean": 66012.2857
+          }
         }
       }
     ]
@@ -152,9 +160,11 @@
         "max-concurrency": 1,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.1214,
-          "latency_mean": 8.2211,
-          "peak_memory_mb_mean": 54501.1429
+          "H100": {
+            "throughput_qps": 0.1214,
+            "latency_mean": 8.2211,
+            "peak_memory_mb_mean": 54501.1429
+          }
         }
       }
     ]
@@ -207,9 +217,11 @@
         "max-concurrency": 1,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.559,
-          "latency_mean": 1.794,
-          "peak_memory_mb_mean": 55290.5714
+          "H100": {
+            "throughput_qps": 0.559,
+            "latency_mean": 1.794,
+            "peak_memory_mb_mean": 55290.5714
+          }
         }
       },
       {
@@ -223,9 +235,11 @@
         "max-concurrency": 1,
         "enable-negative-prompt": true,
         "baseline": {
-          "throughput_qps": 0.1918,
-          "latency_mean": 5.2022,
-          "peak_memory_mb_mean": 64838.8571
+          "H100": {
+            "throughput_qps": 0.1918,
+            "latency_mean": 5.2022,
+            "peak_memory_mb_mean": 64838.8571
+          }
         }
       }
     ]
@@ -269,9 +283,11 @@
         "warmup-num-inference-steps": 20,
         "enable-negative-prompt": true,
         "baseline": {
-            "throughput_qps": [0.4125, 0.6714, 0.6674, 0.8159],
-            "latency_mean": [2.4215, 2.992, 5.9293, 9.656],
-            "peak_memory_mb_mean": [55147.4286, 55148.0, 55150.4643, 55152.3821]
+            "H100": {
+                "throughput_qps": [0.4125, 0.6714, 0.6674, 0.8159],
+                "latency_mean": [2.4215, 2.992, 5.9293, 9.656],
+                "peak_memory_mb_mean": [55147.4286, 55148.0, 55150.4643, 55152.3821]
+            }
         }
       }
     ]

tests/dfx/perf/tests/test_runner_metadata.py

@@ -1,6 +1,8 @@
 """Tests for DFX runner metadata field exclusion."""
 
 import json
+from contextlib import contextmanager
+from types import SimpleNamespace
 
 import pytest
 
@@ -15,7 +17,7 @@ def test_task_excluded_from_cli_args():
         "backend": "openai-audio-speech",
         "endpoint": "/v1/audio/speech",
         "percentile-metrics": "audio_rtf,audio_ttfp",
-        "baseline": {"mean_audio_rtf": [0.5]},
+        "baseline": {"H100": {"mean_audio_rtf": [0.5]}},
     }
     exclude_keys = {
         "request_rate",
@@ -121,35 +123,6 @@ def test_resolve_pytest_marks_rejects_legacy_object_format():
         )
 
 
-def test_extract_mark_resource_label():
-    from tests.dfx.conftest import extract_mark_resource_label
-
-    assert extract_mark_resource_label(None) == "na"
-    assert (
-        extract_mark_resource_label(
-            [
-                {"hardware_marks": {"res": {"cuda": "H100"}, "num_cards": 1}},
-                "full_model",
-            ]
-        )
-        == "H100"
-    )
-    assert (
-        extract_mark_resource_label(
-            [
-                {
-                    "hardware_marks": {
-                        "res": {"cuda": "H100", "rocm": "MI325"},
-                        "num_cards": 2,
-                    }
-                },
-                "diffusion",
-            ]
-        )
-        == "H100-MI325"
-    )
-
-
 def test_resource_label_for_filename():
     from tests.dfx.conftest import resource_label_for_filename
 
@@ -167,16 +140,14 @@ def test_hardware_json_value():
     assert hardware_json_value(None) == ""
 
 
-def test_extract_configs_resource_label(monkeypatch):
-    from tests.dfx.conftest import extract_configs_resource_label, get_runtime_resource_label
+def test_get_runtime_resource_label(monkeypatch):
+    from tests.dfx.conftest import get_runtime_resource_label
 
     monkeypatch.setattr(
         "tests.dfx.conftest._read_runtime_device_name",
         lambda *, device_id=0: "NVIDIA H100 80GB HBM3",
     )
-    get_runtime_resource_label(refresh=True)
-    assert extract_configs_resource_label([]) == "H100"
-    get_runtime_resource_label(refresh=True)
+    assert get_runtime_resource_label(refresh=True) == "H100"
     monkeypatch.setattr(
         "tests.dfx.conftest._read_runtime_device_name",
         lambda *, device_id=0: "Ascend910B2",
@@ -254,8 +225,9 @@ def test_benchmark_param_id_suffix_from_task_eval_phase():
     ]
 
 
-def test_create_paired_omni_benchmark_pytest_params(tmp_path):
+def test_paired_omni_benchmark_reuses_server_and_preserves_case_metadata(tmp_path, monkeypatch):
     from tests.dfx.conftest import create_paired_omni_benchmark_pytest_params
+    from tests.dfx.perf.scripts import run_benchmark
 
     configs = [
         {
@@ -285,3 +257,231 @@ def test_create_paired_omni_benchmark_pytest_params(tmp_path):
     assert bench_row == ("test_tts", 0)
     assert any(m.name == "tts" for m in by_id["test_tts-p0"].marks)
     assert not any(m.name == "tts" for m in by_id["test_omni-p0"].marks)
+
+    omni_p0_server = by_id["test_omni-p0"].values[0]
+    omni_p1_server = by_id["test_omni-p1"].values[0]
+    tts_server = by_id["test_tts-p0"].values[0]
+    assert omni_p0_server == omni_p1_server
+    assert omni_p0_server != tts_server
+
+    events = []
+
+    @contextmanager
+    def fake_start(server_param):
+        events.append(("start", server_param))
+        yield object()
+        events.append(("stop", server_param))
+
+    monkeypatch.setattr(run_benchmark, "_start_omni_server", fake_start)
+    active_context = run_benchmark._SingleActiveContext()
+    try:
+        first = run_benchmark.omni_server.__wrapped__(
+            SimpleNamespace(param=omni_p0_server),
+            active_context,
+        )
+        second = run_benchmark.omni_server.__wrapped__(
+            SimpleNamespace(param=omni_p1_server),
+            active_context,
+        )
+        assert first is second
+        assert events == [("start", omni_p0_server)]
+
+        third = run_benchmark.omni_server.__wrapped__(
+            SimpleNamespace(param=tts_server),
+            active_context,
+        )
+        assert third is not first
+        assert events == [
+            ("start", omni_p0_server),
+            ("stop", omni_p0_server),
+            ("start", tts_server),
+        ]
+    finally:
+        active_context.close()
+
+    assert events == [
+        ("start", omni_p0_server),
+        ("stop", omni_p0_server),
+        ("start", tts_server),
+        ("stop", tts_server),
+    ]
+
+
+def test_is_hardware_nested_baseline():
+    from tests.dfx.conftest import (
+        _RUNTIME_DEVICE_ALIASES,
+        is_hardware_nested_baseline,
+    )
+    from tests.helpers.mark import get_hardware_mark_list
+
+    hardware_marks = get_hardware_mark_list()
+    assert hardware_marks
+    assert {"H100", "L4", "A3", "MI325", "B200"} <= hardware_marks
+    assert "cuda" not in hardware_marks
+    assert "full_model" not in hardware_marks
+
+    # Runtime aliases are a full static list (independent of baseline allowlist).
+    assert "H100" in _RUNTIME_DEVICE_ALIASES
+    assert "A100" in _RUNTIME_DEVICE_ALIASES
+
+    assert is_hardware_nested_baseline(
+        {
+            "H100": {"mean_ttft_ms": [1.0, 2.0], "mean_e2el_ms": [10.0, 20.0]},
+            "A3": {"mean_ttft_ms": [0.9, 1.8], "mean_e2el_ms": [9.0, 18.0]},
+        }
+    )
+    # Custom metric names under a known hardware label are allowed.
+    assert is_hardware_nested_baseline({"H100": {"custom_stage_ms": 12.0, "foo_bar": [1.0, 2.0]}})
+    assert is_hardware_nested_baseline({"A3": {"request_throughput": 0.5}})
+    # Flat metric maps / concurrency-keyed maps are not hardware-nested.
+    assert not is_hardware_nested_baseline({"mean_ttft_ms": [1.0, 2.0], "mean_e2el_ms": [10.0, 20.0]})
+    assert not is_hardware_nested_baseline({"mean_ttft_ms": {"1": 1.0, "32": 2.0}})
+    assert not is_hardware_nested_baseline({})
+    # Alias-only labels (not [hardware-resource] markers) are rejected for baselines.
+    assert not is_hardware_nested_baseline({"A100": {"throughput_qps": 1.0}})
+    # Unknown top-level hardware label is rejected.
+    assert not is_hardware_nested_baseline({"UnknownGPU": {"throughput_qps": 1.0}})
+    # Empty per-hardware metric map is rejected.
+    assert not is_hardware_nested_baseline({"H100": {}})
+
+
+def test_resolve_baseline_for_sweep_keeps_all_hardware_for_one_concurrency():
+    from tests.dfx.conftest import resolve_baseline_for_sweep
+
+    baseline = {
+        "H100": {
+            "mean_ttft_ms": [96.4, 140.8, 271.9, 362.3, 507.8],
+            "mean_e2el_ms": [18507.0, 28365.0, 31907.0, 48161.0, 72630.0],
+        },
+        "A3": {
+            "mean_ttft_ms": [90.0, 130.0, 250.0, 340.0, 480.0],
+            "mean_e2el_ms": [17000.0, 26000.0, 30000.0, 45000.0, 70000.0],
+        },
+    }
+    # max_concurrency=[1,4,8,16,32] -> index 4 is concurrency 32
+    got = resolve_baseline_for_sweep(baseline, sweep_index=4)
+    assert got == {
+        "H100": {"mean_ttft_ms": 507.8, "mean_e2el_ms": 72630.0},
+        "A3": {"mean_ttft_ms": 480.0, "mean_e2el_ms": 70000.0},
+    }
+    # First sweep step keeps both hardware buckets too.
+    got0 = resolve_baseline_for_sweep(baseline, sweep_index=0)
+    assert set(got0) == {"H100", "A3"}
+    assert got0["H100"]["mean_ttft_ms"] == 96.4
+    assert got0["A3"]["mean_ttft_ms"] == 90.0
+
+
+def test_resolve_baseline_for_sweep_rejects_flat_baseline():
+    from tests.dfx.conftest import resolve_baseline_for_sweep
+
+    with pytest.raises(ValueError, match=r"hardware-nested.*pyproject\.toml"):
+        resolve_baseline_for_sweep(
+            {"throughput_qps": [0.4, 0.6], "latency_mean": [1.0, 2.0]},
+            sweep_index=1,
+        )
+
+
+def test_resolve_baseline_for_sweep_rejects_unknown_hardware_label():
+    from tests.dfx.conftest import resolve_baseline_for_sweep
+
+    with pytest.raises(ValueError, match=r"Unknown hardware label\(s\): \['A100'\].*pyproject\.toml"):
+        resolve_baseline_for_sweep(
+            {"A100": {"throughput_qps": 1.0}},
+            sweep_index=0,
+        )
+
+
+def test_resolve_baseline_for_sweep_supports_list_and_scalar_under_hardware():
+    from tests.dfx.conftest import resolve_baseline_for_sweep
+
+    # Sweep-aligned lists under each hardware bucket (canonical form).
+    listed = {
+        "H100": {"throughput_qps": [0.4, 0.6, 0.8], "latency_mean": [1.0, 2.0, 3.0]},
+        "A3": {"throughput_qps": [0.5, 0.7, 0.9], "latency_mean": [0.9, 1.8, 2.7]},
+    }
+    assert resolve_baseline_for_sweep(listed, sweep_index=1) == {
+        "H100": {"throughput_qps": 0.6, "latency_mean": 2.0},
+        "A3": {"throughput_qps": 0.7, "latency_mean": 1.8},
+    }
+
+    # Scalars under hardware stay as-is (single-concurrency cases).
+    scalar = {"H100": {"throughput_qps": 0.5}}
+    assert resolve_baseline_for_sweep(scalar, sweep_index=0) == {"H100": {"throughput_qps": 0.5}}
+    # Custom metric names are preserved.
+    custom = {"H100": {"custom_stage_ms": [10.0, 20.0]}}
+    assert resolve_baseline_for_sweep(custom, sweep_index=1) == {"H100": {"custom_stage_ms": 20.0}}
+    assert resolve_baseline_for_sweep(None) == {}
+    assert resolve_baseline_for_sweep({}) == {}
+
+
+def test_resolve_baseline_value_errors():
+    from tests.dfx.conftest import resolve_baseline_value
+
+    with pytest.raises(ValueError, match="sweep_index"):
+        resolve_baseline_value([1.0, 2.0], sweep_index=None)
+    with pytest.raises(IndexError):
+        resolve_baseline_value([1.0], sweep_index=1)
+    with pytest.raises(TypeError, match="not supported"):
+        resolve_baseline_value({"1": 0.4}, sweep_index=0)
+
+
+def test_diffusion_build_run_params_resolves_baseline_per_sweep(tmp_path, monkeypatch):
+    """``_build_run_params`` / ``_iter_sweep_runs`` narrow list baselines per concurrency."""
+    import importlib
+    import sys
+
+    cfg = tmp_path / "mini_diffusion_perf.json"
+    cfg.write_text(
+        json.dumps(
+            [
+                {
+                    "test_name": "test_mini",
+                    "server_type": "vllm-omni",
+                    "mark": [
+                        {"hardware_marks": {"res": {"cuda": "H100"}, "num_cards": 1}},
+                        "diffusion",
+                        "full_model",
+                    ],
+                    "server_params": {"model": "m/mini"},
+                    "benchmark_params": [{"name": "p0", "num-prompts": 1, "max-concurrency": 1}],
+                }
+            ]
+        ),
+        encoding="utf-8",
+    )
+    monkeypatch.setattr(sys, "argv", ["pytest", "--test-config-file", str(cfg)])
+    sys.modules.pop("tests.dfx.perf.scripts.run_diffusion_benchmark", None)
+    from tests.dfx.perf.scripts import run_diffusion_benchmark as rdb
+
+    importlib.reload(rdb)
+
+    params = {
+        "name": "c_sweep",
+        "dataset": "random",
+        "task": "t2i",
+        "num-prompts": [8, 16, 32],
+        "max-concurrency": [1, 8, 32],
+        "baseline": {
+            "H100": {"throughput_qps": [0.1, 0.2, 0.3], "latency_mean": [10.0, 20.0, 30.0]},
+            "A3": {"throughput_qps": [0.11, 0.22, 0.33], "latency_mean": [9.0, 19.0, 29.0]},
+        },
+    }
+    run32 = rdb._build_run_params(
+        params,
+        num_prompts=32,
+        max_concurrency=32,
+        request_rate="inf",
+        sweep_index=2,
+    )
+    assert run32["max-concurrency"] == 32
+    assert run32["baseline"] == {
+        "H100": {"throughput_qps": 0.3, "latency_mean": 30.0},
+        "A3": {"throughput_qps": 0.33, "latency_mean": 29.0},
+    }
+
+    sweeps = rdb._iter_sweep_runs(params)
+    assert len(sweeps) == 3
+    assert sweeps[2]["params"]["max-concurrency"] == 32
+    assert sweeps[2]["params"]["baseline"]["H100"]["throughput_qps"] == 0.3
+    assert "A3" in sweeps[2]["params"]["baseline"]
+    assert not isinstance(sweeps[2]["params"]["baseline"]["H100"]["throughput_qps"], list)

tests/dfx/perf/tests/test_tts.json

@@ -39,12 +39,14 @@
         "seed_tts_locale": "en",
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "median_audio_ttfp_ms": [
-            141.2791
-          ],
-          "median_audio_rtf": [
-            0.1591
-          ]
+          "H100": {
+            "median_audio_ttfp_ms": [
+              141.2791
+            ],
+            "median_audio_rtf": [
+              0.1591
+            ]
+          }
         }
       },
       {
@@ -67,21 +69,23 @@
         "seed_tts_locale": "en",
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "median_audio_ttfp_ms": [
-            226.747,
-            647.1272,
-            4777.2714
-          ],
-          "median_audio_rtf": [
-            0.3313,
-            0.3642,
-            1.2282
-          ],
-          "audio_throughput": [
-            24.2296,
-            40.3552,
-            43.2572
-          ]
+          "H100": {
+            "median_audio_ttfp_ms": [
+              226.747,
+              647.1272,
+              4777.2714
+            ],
+            "median_audio_rtf": [
+              0.3313,
+              0.3642,
+              1.2282
+            ],
+            "audio_throughput": [
+              24.2296,
+              40.3552,
+              43.2572
+            ]
+          }
         }
       },
       {
@@ -102,9 +106,11 @@
         "seed_tts_wer_eval": true,
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "mean_audio_rtf": [
-            0.45
-          ]
+          "H100": {
+            "mean_audio_rtf": [
+              0.45
+            ]
+          }
         }
       }
     ]
@@ -153,12 +159,14 @@
         },
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "median_audio_ttfp_ms": [
-            46.8434
-          ],
-          "median_audio_rtf": [
-            0.1367
-          ]
+          "H100": {
+            "median_audio_ttfp_ms": [
+              46.8434
+            ],
+            "median_audio_rtf": [
+              0.1367
+            ]
+          }
         }
       },
       {
@@ -186,21 +194,23 @@
         },
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "median_audio_ttfp_ms": [
-            75.3341,
-            713.7727,
-            5941.7179
-          ],
-          "median_audio_rtf": [
-            0.1859,
-            0.3148,
-            1.1662
-          ],
-          "audio_throughput": [
-            38.3858,
-            55.3523,
-            68.7032
-          ]
+          "H100": {
+            "median_audio_ttfp_ms": [
+              75.3341,
+              713.7727,
+              5941.7179
+            ],
+            "median_audio_rtf": [
+              0.1859,
+              0.3148,
+              1.1662
+            ],
+            "audio_throughput": [
+              38.3858,
+              55.3523,
+              68.7032
+            ]
+          }
         }
       },
       {
@@ -226,9 +236,11 @@
         "seed_tts_wer_eval": true,
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "mean_audio_rtf": [
-            0.35
-          ]
+          "H100": {
+            "mean_audio_rtf": [
+              0.35
+            ]
+          }
         }
       },
       {
@@ -252,15 +264,17 @@
         },
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "median_audio_ttfp_ms": [
-            56.746
-          ],
-          "median_audio_rtf": [
-            0.1571
-          ],
-          "audio_throughput": [
-            11.1226
-          ]
+          "H100": {
+            "median_audio_ttfp_ms": [
+              56.746
+            ],
+            "median_audio_rtf": [
+              0.1571
+            ],
+            "audio_throughput": [
+              11.1226
+            ]
+          }
         }
       },
       {
@@ -283,12 +297,14 @@
         },
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "median_audio_ttfp_ms": [
-            46.3479
-          ],
-          "median_audio_rtf": [
-            0.1352
-          ]
+          "H100": {
+            "median_audio_ttfp_ms": [
+              46.3479
+            ],
+            "median_audio_rtf": [
+              0.1352
+            ]
+          }
         }
       },
       {
@@ -315,21 +331,23 @@
         },
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "median_audio_ttfp_ms": [
-            75.5967,
-            587.5877,
-            4960.3726
-          ],
-          "median_audio_rtf": [
-            0.1871,
-            0.3143,
-            1.1703
-          ],
-          "audio_throughput": [
-            38.9378,
-            55.1846,
-            71.5708
-          ]
+          "H100": {
+            "median_audio_ttfp_ms": [
+              75.5967,
+              587.5877,
+              4960.3726
+            ],
+            "median_audio_rtf": [
+              0.1871,
+              0.3143,
+              1.1703
+            ],
+            "audio_throughput": [
+              38.9378,
+              55.1846,
+              71.5708
+            ]
+          }
         }
       },
       {
@@ -352,15 +370,17 @@
         },
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "median_audio_ttfp_ms": [
-            55.5029
-          ],
-          "median_audio_rtf": [
-            0.1508
-          ],
-          "audio_throughput": [
-            9.4885
-          ]
+          "H100": {
+            "median_audio_ttfp_ms": [
+              55.5029
+            ],
+            "median_audio_rtf": [
+              0.1508
+            ],
+            "audio_throughput": [
+              9.4885
+            ]
+          }
         }
       }
     ]
@@ -402,12 +422,14 @@
         },
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "median_audio_ttfp_ms": [
-            82.1886
-          ],
-          "median_audio_rtf": [
-            0.1241
-          ]
+          "H100": {
+            "median_audio_ttfp_ms": [
+              82.1886
+            ],
+            "median_audio_rtf": [
+              0.1241
+            ]
+          }
         }
       },
       {
@@ -429,15 +451,17 @@
         },
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "median_audio_ttfp_ms": [
-            160.9168
-          ],
-          "median_audio_rtf": [
-            0.1907
-          ],
-          "audio_throughput": [
-            39.2332
-          ]
+          "H100": {
+            "median_audio_ttfp_ms": [
+              160.9168
+            ],
+            "median_audio_rtf": [
+              0.1907
+            ],
+            "audio_throughput": [
+              39.2332
+            ]
+          }
         }
       },
       {
@@ -459,15 +483,17 @@
         },
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "median_audio_ttfp_ms": [
-            100.1265
-          ],
-          "median_audio_rtf": [
-            0.1437
-          ],
-          "audio_throughput": [
-            11.7605
-          ]
+          "H100": {
+            "median_audio_ttfp_ms": [
+              100.1265
+            ],
+            "median_audio_rtf": [
+              0.1437
+            ],
+            "audio_throughput": [
+              11.7605
+            ]
+          }
         }
       }
     ]

tests/dfx/perf/tests/test_voxcpm2.json

@@ -37,12 +37,14 @@
         "seed_tts_locale": "en",
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "median_audio_ttfp_ms": [
-            137.645
-          ],
-          "median_audio_rtf": [
-            0.0497
-          ]
+          "H100": {
+            "median_audio_ttfp_ms": [
+              137.645
+            ],
+            "median_audio_rtf": [
+              0.0497
+            ]
+          }
         }
       },
       {
@@ -61,15 +63,17 @@
         "seed_tts_locale": "en",
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "median_audio_ttfp_ms": [
-            475.3357
-          ],
-          "median_audio_rtf": [
-            0.1243
-          ],
-          "audio_throughput": [
-            36.6358
-          ]
+          "H100": {
+            "median_audio_ttfp_ms": [
+              475.3357
+            ],
+            "median_audio_rtf": [
+              0.1243
+            ],
+            "audio_throughput": [
+              36.6358
+            ]
+          }
         }
       },
       {
@@ -90,9 +94,11 @@
         "seed_tts_wer_eval": true,
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "mean_audio_rtf": [
-            0.45
-          ]
+          "H100": {
+            "mean_audio_rtf": [
+              0.45
+            ]
+          }
         }
       },
       {
@@ -114,12 +120,14 @@
         },
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "median_audio_ttfp_ms": [
-            79.6927
-          ],
-          "median_audio_rtf": [
-            0.0423
-          ]
+          "H100": {
+            "median_audio_ttfp_ms": [
+              79.6927
+            ],
+            "median_audio_rtf": [
+              0.0423
+            ]
+          }
         }
       },
       {
@@ -141,15 +149,17 @@
         },
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "median_audio_ttfp_ms": [
-            262.6212
-          ],
-          "median_audio_rtf": [
-            0.0753
-          ],
-          "audio_throughput": [
-            103.5843
-          ]
+          "H100": {
+            "median_audio_ttfp_ms": [
+              262.6212
+            ],
+            "median_audio_rtf": [
+              0.0753
+            ],
+            "audio_throughput": [
+              103.5843
+            ]
+          }
         }
       },
       {
@@ -171,15 +181,17 @@
         },
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "median_audio_ttfp_ms": [
-            80.1095
-          ],
-          "median_audio_rtf": [
-            0.0479
-          ],
-          "audio_throughput": [
-            17.2918
-          ]
+          "H100": {
+            "median_audio_ttfp_ms": [
+              80.1095
+            ],
+            "median_audio_rtf": [
+              0.0479
+            ],
+            "audio_throughput": [
+              17.2918
+            ]
+          }
         }
       },
       {
@@ -203,9 +215,11 @@
         "seed_tts_wer_eval": true,
         "percentile-metrics": "ttft,e2el,audio_rtf,audio_ttfp,audio_duration",
         "baseline": {
-          "mean_audio_rtf": [
-            0.35
-          ]
+          "H100": {
+            "mean_audio_rtf": [
+              0.35
+            ]
+          }
         }
       }
     ]

tests/dfx/perf/tests/test_wan22_i2v_vllm_omni.json

@@ -42,9 +42,11 @@
           }
         ],
         "baseline": {
-          "throughput_qps": 0.0434,
-          "latency_mean": 23.0654,
-          "peak_memory_mb_mean": 76360.0
+          "H100": {
+            "throughput_qps": 0.0434,
+            "latency_mean": 23.0654,
+            "peak_memory_mb_mean": 76360.0
+          }
         }
       }
     ]
@@ -97,9 +99,11 @@
           }
         ],
         "baseline": {
-          "throughput_qps": 0.0606,
-          "latency_mean": 16.4928,
-          "peak_memory_mb_mean": 47164.8
+          "H100": {
+            "throughput_qps": 0.0606,
+            "latency_mean": 16.4928,
+            "peak_memory_mb_mean": 47164.8
+          }
         }
       },
       {
@@ -122,9 +126,11 @@
           }
         ],
         "baseline": {
-          "throughput_qps": 0.0124,
-          "latency_mean": 80.7784,
-          "peak_memory_mb_mean": 55374.3429
+          "H100": {
+            "throughput_qps": 0.0124,
+            "latency_mean": 80.7784,
+            "peak_memory_mb_mean": 55374.3429
+          }
         }
       }
     ]

tests/dfx/reliability/invalid_param_test/test_invalid_audio_speech.py

@@ -1,6 +1,6 @@
 # SPDX-License-Identifier: Apache-2.0
 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project
-"""HTTP + WebSocket validation for Qwen3-TTS Base, Higgs Audio V3 and Audex TTA ``/v1/audio/speech`` (and related routes)."""
+"""HTTP + WebSocket validation for Qwen3-TTS Base and Higgs Audio V3 ``/v1/audio/speech`` (and related routes)."""
 
 from __future__ import annotations
 
@@ -70,19 +70,6 @@ _HIGGS_AUDIO_V3_SPEECH = [
     ),
 ]
 
-_AUDEX_TTA_CAPTION = "Heavy rain falling on a tin roof."
-_AUDEX_TTA_SPEECH = [
-    pytest.param(
-        OmniServerParams(
-            model="nvidia/Nemotron-Labs-Audex-2B",
-            stage_config_path=get_deploy_config_path("audex_tta.yaml"),
-            server_args=_SPEECH_SERVER_ARGS,
-        ),
-        id="audex_tta",
-        marks=_L4_SPEECH_HW,
-    ),
-]
-
 
 def _apply_batch_overrides(body: dict[str, object], *, loc: str, overrides: dict[str, object]) -> None:
     if loc == "batch":
@@ -338,37 +325,6 @@ def test_speech_higgs_invalid_field_values(
     )
 
 
-# Audex TTA (``audex_tta`` deployment): the serving layer builds the caption
-# prompt and injects the RVQ phase contract + default guidance server-side, so
-# guidance knobs and speech-only fields must be rejected before generation.
-@pytest.mark.parametrize(
-    "overrides, err_message",
-    [
-        pytest.param({"extra_params": {"cfg_scale": 99}}, "cfg_scale", id="cfg_scale_out_of_range"),
-        pytest.param({"extra_params": {"cfg_scale": "big"}}, "cfg_scale", id="cfg_scale_wrong_type"),
-        pytest.param({"extra_params": {"tta_rvq": {}}}, "tta_rvq", id="tta_rvq_not_overridable"),
-        pytest.param({"input": "   "}, "caption", id="caption_whitespace_only"),
-        pytest.param({"voice": "alloy"}, "voice", id="voice_not_supported"),
-        pytest.param(
-            {"ref_audio": "https://example.com/ref.wav"},
-            "reference audio",
-            id="ref_audio_not_supported",
-        ),
-    ],
-)
-@pytest.mark.parametrize("omni_server", _AUDEX_TTA_SPEECH, indirect=True)
-def test_speech_audex_tta_invalid_field_values(
-    omni_server: OmniServer,
-    openai_client: OpenAIClientHandler,
-    overrides: dict[str, object],
-    err_message: str | tuple[str, ...],
-) -> None:
-    body = {"model": omni_server.model, "input": _AUDEX_TTA_CAPTION, **overrides}
-    openai_client.send_audio_speech_http_request(
-        {"json": body, "timeout": 120, "err_code": 400, "err_message": err_message}
-    )
-
-
 # ─────────────────────────────────────────────────────────────────────────────
 # POST /v1/audio/speech/batch (JSON)
 # ─────────────────────────────────────────────────────────────────────────────