[MiniMax-M3] Retune decode indexer launch grid for occupancy (#130)

Signed-off-by: Zijing Liu <liuzijing2014@gmail.com>

.buildkite/hardware_tests/intel_xpu_ci/test-intel.yaml

@@ -21,10 +21,6 @@ steps:
     timeout_in_minutes: 30
     optional: true
     device: intel_gpu
-    agent_tags:
-      label: production
-      gpu: 2+
-      mem: 24+
     no_plugin: true
     env:
       REGISTRY: "public.ecr.aws/q9t5s3a7"
@@ -42,10 +38,6 @@ steps:
     timeout_in_minutes: 30
     optional: true
     device: intel_gpu
-    agent_tags:
-      label: production
-      gpu: 1+
-      mem: 16+
     no_plugin: true
     env:
       REGISTRY: "public.ecr.aws/q9t5s3a7"
@@ -63,10 +55,6 @@ steps:
     timeout_in_minutes: 30
     optional: true
     device: intel_gpu
-    agent_tags:
-      label: production
-      gpu: 1+
-      mem: 16+
     no_plugin: true
     env:
       REGISTRY: "public.ecr.aws/q9t5s3a7"

.buildkite/intel_jobs/basic_correctness.yaml

@@ -5,10 +5,6 @@ steps:
 - label: XPU Sleep Mode
   timeout_in_minutes: 30
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 1+
-    mem: 16+
   no_plugin: true
   working_dir: "."
   env:

.buildkite/intel_jobs/engine_intel.yaml

@@ -5,10 +5,6 @@ steps:
 - label: Engine (1 GPU)
   timeout_in_minutes: 30
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 1+
-    mem: 16+
   no_plugin: true
   working_dir: "."
   env:

.buildkite/intel_jobs/expert_parallelism_intel.yaml

@@ -6,10 +6,6 @@ steps:
   key: eplb-algorithm
   timeout_in_minutes: 45
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 1+
-    mem: 16+
   no_plugin: true
   working_dir: "."
   env:

.buildkite/intel_jobs/kernels_intel.yaml

@@ -5,10 +5,6 @@ steps:
 - label: vLLM IR Tests
   timeout_in_minutes: 30
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 1+
-    mem: 16+
   no_plugin: true
   working_dir: "."
   env:

.buildkite/intel_jobs/lora_intel.yaml

@@ -5,10 +5,6 @@ steps:
 - label: LoRA Runtime + Utils
   timeout_in_minutes: 45
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 1+
-    mem: 24+
   no_plugin: true
   working_dir: "."
   env:
@@ -38,10 +34,6 @@ steps:
 - label: LoRA Fused/MoE Kernels
   timeout_in_minutes: 45
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 1+
-    mem: 16+
   no_plugin: true
   working_dir: "."
   env:
@@ -62,10 +54,6 @@ steps:
 - label: LoRA Punica Kernels
   timeout_in_minutes: 45
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 1+
-    mem: 16+
   no_plugin: true
   working_dir: "."
   env:
@@ -86,10 +74,6 @@ steps:
 - label: LoRA Punica FP8/XPU Ops
   timeout_in_minutes: 45
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 1+
-    mem: 16+
   no_plugin: true
   working_dir: "."
   env:
@@ -110,10 +94,6 @@ steps:
 - label: LoRA Models
   timeout_in_minutes: 45
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 2+
-    mem: 24+
   no_plugin: true
   working_dir: "."
   env:
@@ -137,10 +117,6 @@ steps:
 - label: LoRA Multimodal
   timeout_in_minutes: 45
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 1+
-    mem: 16+
   no_plugin: true
   working_dir: "."
   env:

.buildkite/intel_jobs/misc_intel.yaml

@@ -5,10 +5,6 @@ steps:
 - label: V1 Core + KV + Metrics
   timeout_in_minutes: 30
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 1+
-    mem: 16+
   no_plugin: true
   working_dir: "."
   env:
@@ -35,10 +31,6 @@ steps:
 - label: V1 Sample + Logits
   timeout_in_minutes: 30
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 1+
-    mem: 16+
   no_plugin: true
   working_dir: "."
   env:
@@ -79,10 +71,6 @@ steps:
 - label: XPU CPU Offload
   timeout_in_minutes: 60
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 1+
-    mem: 16+
   no_plugin: true
   working_dir: "."
   env:
@@ -107,10 +95,6 @@ steps:
   key: regression
   timeout_in_minutes: 30
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 1+
-    mem: 16+
   no_plugin: true
   working_dir: "."
   env:
@@ -142,10 +126,6 @@ steps:
   timeout_in_minutes: 30
   num_devices: 2
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 2+
-    mem: 16+
   no_plugin: true
   working_dir: "."
   env:
@@ -177,10 +157,6 @@ steps:
   key: async-engine-inputs-utils-worker
   timeout_in_minutes: 30
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 1+
-    mem: 24+
   no_plugin: true
   working_dir: "."
   env:

.buildkite/intel_jobs/model_runner_v2_intel.yaml

@@ -5,10 +5,6 @@ steps:
 - label: Model Runner V2 Core Tests (Intel)
   timeout_in_minutes: 45
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 2+
-    mem: 16+
   no_plugin: true
   working_dir: "."
   env:
@@ -34,10 +30,6 @@ steps:
 - label: Model Runner V2 Examples (Intel)
   timeout_in_minutes: 45
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 1+
-    mem: 24+
   no_plugin: true
   working_dir: "."
   env:

.buildkite/intel_jobs/models_multimodal_intel.yaml

@@ -6,10 +6,6 @@ steps:
   key: multi-modal-models-standard-1-qwen2
   timeout_in_minutes: 45
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 1+
-    mem: 16+
   no_plugin: true
   working_dir: "."
   env:
@@ -31,10 +27,6 @@ steps:
   key: multi-modal-models-standard-2-qwen3-gemma
   timeout_in_minutes: 45
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 1+
-    mem: 16+
   no_plugin: true
   working_dir: "."
   env:
@@ -55,10 +47,6 @@ steps:
   key: multi-modal-models-standard-3-llava-qwen2-vl
   timeout_in_minutes: 45
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 1+
-    mem: 24+
   no_plugin: true
   working_dir: "."
   env:
@@ -80,10 +68,6 @@ steps:
   key: multi-modal-models-standard-4-other-whisper
   timeout_in_minutes: 45
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 1+
-    mem: 16+
   no_plugin: true
   working_dir: "."
   env:
@@ -104,10 +88,6 @@ steps:
   key: multi-modal-processor
   timeout_in_minutes: 45
   device: intel_gpu
-  agent_tags:
-    label: production
-    gpu: 1+
-    mem: 16+
   no_plugin: true
   working_dir: "."
   env:

.buildkite/intel_jobs/test-intel.yaml

@@ -19,10 +19,6 @@ steps:
       - image-build-xpu
     timeout_in_minutes: 30
     device: intel_gpu
-    agent_tags:
-      label: production
-      gpu: 2+
-      mem: 24+
     no_plugin: true
     env:
       REGISTRY: "public.ecr.aws/q9t5s3a7"
@@ -53,10 +49,6 @@ steps:
       - image-build-xpu
     timeout_in_minutes: 30
     device: intel_gpu
-    agent_tags:
-      label: production
-      gpu: 1+
-      mem: 16+
     no_plugin: true
     env:
       REGISTRY: "public.ecr.aws/q9t5s3a7"
@@ -82,10 +74,6 @@ steps:
       - image-build-xpu
     timeout_in_minutes: 30
     device: intel_gpu
-    agent_tags:
-      label: production
-      gpu: 1+
-      mem: 16+
     no_plugin: true
     env:
       REGISTRY: "public.ecr.aws/q9t5s3a7"
@@ -105,10 +93,6 @@ steps:
       - image-build-xpu
     timeout_in_minutes: 30
     device: intel_gpu
-    agent_tags:
-      label: production
-      gpu: 1+
-      mem: 16+
     no_plugin: true
     env:
       REGISTRY: "public.ecr.aws/q9t5s3a7"

.buildkite/scripts/hardware_ci/run-intel-ci-test.sh

@@ -34,7 +34,7 @@ case "${test_suite}" in
     pytest -v -s v1/worker --ignore=v1/worker/test_gpu_model_runner.py --ignore=v1/worker/test_worker_memory_snapshot.py
     pytest -v -s v1/structured_output
     pytest -v -s v1/test_serial_utils.py
-    pytest -v -s v1/spec_decode --ignore=v1/spec_decode/test_max_len.py --ignore=v1/spec_decode/test_speculators_eagle3.py --ignore=v1/spec_decode/test_acceptance_length.py --ignore=v1/spec_decode/test_speculators_correctness.py
+    pytest -v -s v1/spec_decode --ignore=v1/spec_decode/test_max_len.py --ignore=v1/spec_decode/test_speculators_eagle3.py --ignore=v1/spec_decode/test_acceptance_length.py
     pytest -v -s v1/kv_connector/unit --ignore=v1/kv_connector/unit/test_multi_connector.py --ignore=v1/kv_connector/unit/test_example_connector.py --ignore=v1/kv_connector/unit/test_lmcache_integration.py --ignore=v1/kv_connector/unit/test_hf3fs_client.py --ignore=v1/kv_connector/unit/test_hf3fs_connector.py --ignore=v1/kv_connector/unit/test_hf3fs_metadata_server.py --ignore=v1/kv_connector/unit/test_offloading_connector.py
     ;;
   server)

.buildkite/scripts/install-kv-connectors.sh

@@ -4,11 +4,6 @@
 
 set -euo pipefail
 
-if python3 -c "import torch; raise SystemExit(0 if torch.version.hip is not None else 1)"; then
-    uv pip install --system -r /vllm-workspace/requirements/kv_connectors_rocm.txt
-    exit 0
-fi
-
 REQUIREMENTS_FILE="${KV_CONNECTORS_REQUIREMENTS:-/vllm-workspace/requirements/kv_connectors.txt}"
 
 uv pip install --system -r "${REQUIREMENTS_FILE}"

.buildkite/test-amd.yaml

@@ -1594,10 +1594,9 @@ steps:
 #----------------------------------------------------------  mi300 · kernels  ----------------------------------------------------------#
 
 - label: Kernels Attention Test %N # TBD
-  timeout_in_minutes: 55
+  timeout_in_minutes: 180
   mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]
   agent_pool: mi300_1
-  optional: true
   parallelism: 2
   working_dir: "/vllm-workspace/tests"
   source_file_dependencies:
@@ -1628,11 +1627,10 @@ steps:
   - pytest -v -s kernels/core --ignore=kernels/core/test_minimax_reduce_rms.py  kernels/test_concat_mla_q.py kernels/test_top_k_per_row.py
 
 - label: Kernels MoE Test %N # TBD
-  timeout_in_minutes: 50
+  timeout_in_minutes: 180
   mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]
   agent_pool: mi300_1
-  optional: true
-  parallelism: 5
+  parallelism: 4
   working_dir: "/vllm-workspace/tests"
   source_file_dependencies:
   - csrc/quantization/cutlass_w8a8/moe/
@@ -2122,10 +2120,9 @@ steps:
   - pytest -v -s v1/e2e/spec_decode -k "draft_model or no_sync or batch_inference"
 
 - label: Spec Decode Eagle # TBD
-  timeout_in_minutes: 45
+  timeout_in_minutes: 180
   mirror_hardwares: [amdexperimental, amdproduction, amdgfx942nightly, amdmi300]
   agent_pool: mi300_1
-  optional: true
   working_dir: "/vllm-workspace/tests"
   source_file_dependencies:
   - vllm/v1/spec_decode/
@@ -3043,7 +3040,7 @@ steps:
 #----------------------------------------------------------  mi355 · kernels  ----------------------------------------------------------#
 
 - label: Kernels (B200-MI355) # TBD
-  timeout_in_minutes: 15
+  timeout_in_minutes: 180
   mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
   agent_pool: mi355_1
   working_dir: "/vllm-workspace/"
@@ -3067,10 +3064,11 @@ steps:
   - pytest -v -s tests/kernels/attention/test_attention_selector.py
 
 - label: Kernels Attention Test %N # TBD
-  timeout_in_minutes: 60
+  timeout_in_minutes: 180
   mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
   agent_pool: mi355_1
   parallelism: 2
+  optional: true
   working_dir: "/vllm-workspace/tests"
   source_file_dependencies:
   - csrc/attention/
@@ -3084,10 +3082,10 @@ steps:
   - pytest -v -s kernels/attention --shard-id=$$BUILDKITE_PARALLEL_JOB --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT
 
 - label: Kernels MoE Test %N # TBD
-  timeout_in_minutes: 50
+  timeout_in_minutes: 180
   mirror_hardwares: [amdexperimental, amdproduction, amdgfx950nightly, amdmi355]
   agent_pool: mi355_1
-  parallelism: 5
+  parallelism: 4
   working_dir: "/vllm-workspace/tests"
   source_file_dependencies:
   - csrc/quantization/cutlass_w8a8/moe/

.buildkite/test_areas/basic_correctness.yaml

@@ -12,6 +12,7 @@ steps:
   - tests/basic_correctness/test_cpu_offload
   - tests/basic_correctness/test_mem.py
   commands:
-  - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s basic_correctness/test_mem.py
-  - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s basic_correctness/test_basic_correctness.py
-  - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s basic_correctness/test_cpu_offload.py
+  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+  - pytest -v -s basic_correctness/test_mem.py
+  - pytest -v -s basic_correctness/test_basic_correctness.py
+  - pytest -v -s basic_correctness/test_cpu_offload.py

.buildkite/test_areas/compile.yaml

@@ -14,7 +14,8 @@ steps:
   - vllm/v1/cudagraph_dispatcher.py
   - tests/compile/correctness_e2e/test_sequence_parallel.py
   commands:
-  - VLLM_TEST_CLEAN_GPU_MEMORY=1 pytest -v -s tests/compile/correctness_e2e/test_sequence_parallel.py
+  - export VLLM_TEST_CLEAN_GPU_MEMORY=1
+  - pytest -v -s tests/compile/correctness_e2e/test_sequence_parallel.py
 
 - label: Sequence Parallel Correctness Tests (2xH100)
   key: sequence-parallel-correctness-tests-2xh100
@@ -24,7 +25,8 @@ steps:
   optional: true
   num_devices: 2
   commands:
-  - VLLM_TEST_CLEAN_GPU_MEMORY=1 pytest -v -s tests/compile/correctness_e2e/test_sequence_parallel.py
+  - export VLLM_TEST_CLEAN_GPU_MEMORY=1
+  - pytest -v -s tests/compile/correctness_e2e/test_sequence_parallel.py
 
 - label: AsyncTP Correctness Tests (2xH100)
   key: asynctp-correctness-tests-2xh100
@@ -34,7 +36,8 @@ steps:
   optional: true
   num_devices: 2
   commands:
-  - VLLM_TEST_CLEAN_GPU_MEMORY=1 pytest -v -s tests/compile/correctness_e2e/test_async_tp.py
+  - export VLLM_TEST_CLEAN_GPU_MEMORY=1
+  - pytest -v -s tests/compile/correctness_e2e/test_async_tp.py
 
 - label: AsyncTP Correctness Tests (B200)
   key: asynctp-correctness-tests-b200
@@ -44,7 +47,8 @@ steps:
   optional: true
   num_devices: 2
   commands:
-  - VLLM_TEST_CLEAN_GPU_MEMORY=1 pytest -v -s tests/compile/correctness_e2e/test_async_tp.py
+  - export VLLM_TEST_CLEAN_GPU_MEMORY=1
+  - pytest -v -s tests/compile/correctness_e2e/test_async_tp.py
 
 - label: Distributed Compile Unit Tests (2xH100)
   key: distributed-compile-unit-tests-2xh100
@@ -57,7 +61,8 @@ steps:
   - vllm/model_executor/layers
   - tests/compile/passes/distributed/
   commands:
-  - VLLM_TEST_CLEAN_GPU_MEMORY=1 pytest -s -v tests/compile/passes/distributed
+  - export VLLM_TEST_CLEAN_GPU_MEMORY=1
+  - pytest -s -v tests/compile/passes/distributed
 
 - label: Fusion and Compile Unit Tests (2xB200)
   key: fusion-and-compile-unit-tests-2xb200

.buildkite/test_areas/distributed.yaml

@@ -32,10 +32,11 @@ steps:
   - tests/entrypoints/openai/test_multi_api_servers.py
   commands:
   # https://github.com/NVIDIA/nccl/issues/1838
-  - NCCL_CUMEM_HOST_ENABLE=0 TP_SIZE=1 DP_SIZE=2 pytest -v -s v1/distributed/test_async_llm_dp.py
-  - NCCL_CUMEM_HOST_ENABLE=0 TP_SIZE=1 DP_SIZE=2 pytest -v -s v1/distributed/test_eagle_dp.py
-  - NCCL_CUMEM_HOST_ENABLE=0 TP_SIZE=1 DP_SIZE=2 pytest -v -s v1/distributed/test_external_lb_dp.py
-  - NCCL_CUMEM_HOST_ENABLE=0 DP_SIZE=2 pytest -v -s entrypoints/openai/test_multi_api_servers.py
+  - export NCCL_CUMEM_HOST_ENABLE=0
+  - TP_SIZE=1 DP_SIZE=2 pytest -v -s v1/distributed/test_async_llm_dp.py
+  - TP_SIZE=1 DP_SIZE=2 pytest -v -s v1/distributed/test_eagle_dp.py
+  - TP_SIZE=1 DP_SIZE=2 pytest -v -s v1/distributed/test_external_lb_dp.py
+  - DP_SIZE=2 pytest -v -s entrypoints/openai/test_multi_api_servers.py
 
 - label: Distributed Compile + RPC Tests (2 GPUs)
   key: distributed-compile-rpc-tests-2-gpus
@@ -55,9 +56,10 @@ steps:
   - tests/entrypoints/llm/test_collective_rpc.py
   commands:
   # https://github.com/NVIDIA/nccl/issues/1838
-  - NCCL_CUMEM_HOST_ENABLE=0 pytest -v -s entrypoints/llm/test_collective_rpc.py
-  - NCCL_CUMEM_HOST_ENABLE=0 pytest -v -s ./compile/fullgraph/test_basic_correctness.py
-  - NCCL_CUMEM_HOST_ENABLE=0 pytest -v -s ./compile/test_wrapper.py
+  - export NCCL_CUMEM_HOST_ENABLE=0
+  - pytest -v -s entrypoints/llm/test_collective_rpc.py
+  - pytest -v -s ./compile/fullgraph/test_basic_correctness.py
+  - pytest -v -s ./compile/test_wrapper.py
 
 - label: Distributed Torchrun + Shutdown Tests (2 GPUs)
   key: distributed-torchrun-shutdown-tests-2-gpus
@@ -76,10 +78,11 @@ steps:
   - tests/v1/worker/test_worker_memory_snapshot.py
   commands:
   # https://github.com/NVIDIA/nccl/issues/1838
-  - NCCL_CUMEM_HOST_ENABLE=0 VLLM_TEST_SAME_HOST=1 torchrun --nproc-per-node=4 distributed/test_same_node.py | grep 'Same node test passed'
-  - NCCL_CUMEM_HOST_ENABLE=0 VLLM_TEST_SAME_HOST=1 VLLM_TEST_WITH_DEFAULT_DEVICE_SET=1 torchrun --nproc-per-node=4 distributed/test_same_node.py | grep 'Same node test passed'
-  - NCCL_CUMEM_HOST_ENABLE=0 CUDA_VISIBLE_DEVICES=0,1 pytest -v -s v1/shutdown
-  - NCCL_CUMEM_HOST_ENABLE=0 pytest -v -s v1/worker/test_worker_memory_snapshot.py
+  - export NCCL_CUMEM_HOST_ENABLE=0
+  - VLLM_TEST_SAME_HOST=1 torchrun --nproc-per-node=4 distributed/test_same_node.py | grep 'Same node test passed'
+  - VLLM_TEST_SAME_HOST=1 VLLM_TEST_WITH_DEFAULT_DEVICE_SET=1 torchrun --nproc-per-node=4 distributed/test_same_node.py | grep 'Same node test passed'
+  - CUDA_VISIBLE_DEVICES=0,1 pytest -v -s v1/shutdown
+  - pytest -v -s v1/worker/test_worker_memory_snapshot.py
 
 - label: Distributed Torchrun + Examples (4 GPUs)
   key: distributed-torchrun-examples-4-gpus
@@ -94,23 +97,24 @@ steps:
   - tests/examples/features/data_parallel/data_parallel_offline.py
   commands:
   # https://github.com/NVIDIA/nccl/issues/1838
+  - export NCCL_CUMEM_HOST_ENABLE=0
   # test with torchrun tp=2 and external_dp=2
-  - NCCL_CUMEM_HOST_ENABLE=0 torchrun --nproc-per-node=4 tests/distributed/test_torchrun_example.py
+  - torchrun --nproc-per-node=4 tests/distributed/test_torchrun_example.py
   # test with torchrun tp=2 and pp=2
-  - NCCL_CUMEM_HOST_ENABLE=0 PP_SIZE=2 torchrun --nproc-per-node=4 tests/distributed/test_torchrun_example.py
+  - PP_SIZE=2 torchrun --nproc-per-node=4 tests/distributed/test_torchrun_example.py
   # test with torchrun tp=4 and dp=1
-  - NCCL_CUMEM_HOST_ENABLE=0 TP_SIZE=4 torchrun --nproc-per-node=4 tests/distributed/test_torchrun_example_moe.py
+  - TP_SIZE=4 torchrun --nproc-per-node=4 tests/distributed/test_torchrun_example_moe.py
   # test with torchrun tp=2, pp=2 and dp=1
-  - NCCL_CUMEM_HOST_ENABLE=0 PP_SIZE=2 TP_SIZE=2 torchrun --nproc-per-node=4 tests/distributed/test_torchrun_example_moe.py
+  - PP_SIZE=2 TP_SIZE=2 torchrun --nproc-per-node=4 tests/distributed/test_torchrun_example_moe.py
   # test with torchrun tp=1 and dp=4 with ep
-  - NCCL_CUMEM_HOST_ENABLE=0 DP_SIZE=4 ENABLE_EP=1 torchrun --nproc-per-node=4 tests/distributed/test_torchrun_example_moe.py
+  - DP_SIZE=4 ENABLE_EP=1 torchrun --nproc-per-node=4 tests/distributed/test_torchrun_example_moe.py
   # test with torchrun tp=2 and dp=2 with ep
-  - NCCL_CUMEM_HOST_ENABLE=0 TP_SIZE=2 DP_SIZE=2 ENABLE_EP=1 torchrun --nproc-per-node=4 tests/distributed/test_torchrun_example_moe.py
+  - TP_SIZE=2 DP_SIZE=2 ENABLE_EP=1 torchrun --nproc-per-node=4 tests/distributed/test_torchrun_example_moe.py
   # test with internal dp
-  - NCCL_CUMEM_HOST_ENABLE=0 python3 examples/features/data_parallel/data_parallel_offline.py --enforce-eager
+  - python3 examples/features/data_parallel/data_parallel_offline.py --enforce-eager
   # rlhf examples
-  - NCCL_CUMEM_HOST_ENABLE=0 VLLM_ALLOW_INSECURE_SERIALIZATION=1 python3 examples/rl/rlhf_nccl.py
-  - NCCL_CUMEM_HOST_ENABLE=0 VLLM_ALLOW_INSECURE_SERIALIZATION=1 python3 examples/rl/rlhf_ipc.py
+  - VLLM_ALLOW_INSECURE_SERIALIZATION=1 python3 examples/rl/rlhf_nccl.py
+  - VLLM_ALLOW_INSECURE_SERIALIZATION=1 python3 examples/rl/rlhf_ipc.py
 
 - label: Distributed DP Tests (4 GPUs)
   key: distributed-dp-tests-4-gpus
@@ -124,13 +128,14 @@ steps:
   - tests/distributed/test_utils
   commands:
   # https://github.com/NVIDIA/nccl/issues/1838
-  - NCCL_CUMEM_HOST_ENABLE=0 TP_SIZE=2 DP_SIZE=2 pytest -v -s v1/distributed/test_async_llm_dp.py
-  - NCCL_CUMEM_HOST_ENABLE=0 TP_SIZE=2 DP_SIZE=2 pytest -v -s v1/distributed/test_eagle_dp.py
-  - NCCL_CUMEM_HOST_ENABLE=0 TP_SIZE=2 DP_SIZE=2 pytest -v -s v1/distributed/test_external_lb_dp.py
-  - NCCL_CUMEM_HOST_ENABLE=0 TP_SIZE=1 DP_SIZE=4 pytest -v -s v1/distributed/test_internal_lb_dp.py
-  - NCCL_CUMEM_HOST_ENABLE=0 TP_SIZE=1 DP_SIZE=4 pytest -v -s v1/distributed/test_hybrid_lb_dp.py
-  - NCCL_CUMEM_HOST_ENABLE=0 pytest -v -s v1/engine/test_engine_core_client.py::test_kv_cache_events_dp
-  - NCCL_CUMEM_HOST_ENABLE=0 pytest -v -s distributed/test_utils.py
+  - export NCCL_CUMEM_HOST_ENABLE=0
+  - TP_SIZE=2 DP_SIZE=2 pytest -v -s v1/distributed/test_async_llm_dp.py
+  - TP_SIZE=2 DP_SIZE=2 pytest -v -s v1/distributed/test_eagle_dp.py
+  - TP_SIZE=2 DP_SIZE=2 pytest -v -s v1/distributed/test_external_lb_dp.py
+  - TP_SIZE=1 DP_SIZE=4 pytest -v -s v1/distributed/test_internal_lb_dp.py
+  - TP_SIZE=1 DP_SIZE=4 pytest -v -s v1/distributed/test_hybrid_lb_dp.py
+  - pytest -v -s v1/engine/test_engine_core_client.py::test_kv_cache_events_dp
+  - pytest -v -s distributed/test_utils.py
 
 - label: Distributed Compile + Comm (4 GPUs)
   key: distributed-compile-comm-4-gpus
@@ -146,12 +151,13 @@ steps:
   - tests/distributed/test_multiproc_executor.py
   commands:
   # https://github.com/NVIDIA/nccl/issues/1838
-  - NCCL_CUMEM_HOST_ENABLE=0 pytest -v -s compile/fullgraph/test_basic_correctness.py
-  - NCCL_CUMEM_HOST_ENABLE=0 pytest -v -s distributed/test_pynccl.py
-  - NCCL_CUMEM_HOST_ENABLE=0 pytest -v -s distributed/test_events.py
-  - NCCL_CUMEM_HOST_ENABLE=0 pytest -v -s distributed/test_symm_mem_allreduce.py
+  - export NCCL_CUMEM_HOST_ENABLE=0
+  - pytest -v -s compile/fullgraph/test_basic_correctness.py
+  - pytest -v -s distributed/test_pynccl.py
+  - pytest -v -s distributed/test_events.py
+  - pytest -v -s distributed/test_symm_mem_allreduce.py
   # test multi-node TP with multiproc executor (simulated on single node)
-  - NCCL_CUMEM_HOST_ENABLE=0 pytest -v -s distributed/test_multiproc_executor.py::test_multiproc_executor_multi_node
+  - pytest -v -s distributed/test_multiproc_executor.py::test_multiproc_executor_multi_node
 
 - label: Distributed Tests (8 GPUs)(H100)
   key: distributed-tests-8-gpus-h100
@@ -170,8 +176,9 @@ steps:
 
   commands:
   # https://github.com/NVIDIA/nccl/issues/1838
+  - export NCCL_CUMEM_HOST_ENABLE=0
   # test with torchrun tp=2 and dp=4 with ep
-  - NCCL_CUMEM_HOST_ENABLE=0 torchrun --nproc-per-node=8 ../examples/features/torchrun/torchrun_dp_example_offline.py --tp-size=2 --pp-size=1 --dp-size=4 --enable-ep
+  - torchrun --nproc-per-node=8 ../examples/features/torchrun/torchrun_dp_example_offline.py --tp-size=2 --pp-size=1 --dp-size=4 --enable-ep
 
 - label: Distributed Tests (4 GPUs)(A100)
   key: distributed-tests-4-gpus-a100
@@ -264,7 +271,9 @@ steps:
   - tests/distributed/test_pipeline_parallel.py
   - tests/basic_correctness/test_basic_correctness.py
   commands:
-  - VLLM_USE_RAY_V2_EXECUTOR_BACKEND=1 NCCL_CUMEM_HOST_ENABLE=0 pytest -v -s distributed/test_ray_v2_executor.py
-  - VLLM_USE_RAY_V2_EXECUTOR_BACKEND=1 NCCL_CUMEM_HOST_ENABLE=0 pytest -v -s distributed/test_ray_v2_executor_e2e.py
-  - VLLM_USE_RAY_V2_EXECUTOR_BACKEND=1 NCCL_CUMEM_HOST_ENABLE=0 pytest -v -s distributed/test_pipeline_parallel.py -k "ray"
-  - VLLM_USE_RAY_V2_EXECUTOR_BACKEND=1 NCCL_CUMEM_HOST_ENABLE=0 TARGET_TEST_SUITE=L4 pytest -v -s basic_correctness/test_basic_correctness.py -k "ray"
+  - export VLLM_USE_RAY_V2_EXECUTOR_BACKEND=1
+  - export NCCL_CUMEM_HOST_ENABLE=0
+  - pytest -v -s distributed/test_ray_v2_executor.py
+  - pytest -v -s distributed/test_ray_v2_executor_e2e.py
+  - pytest -v -s distributed/test_pipeline_parallel.py -k "ray"
+  - TARGET_TEST_SUITE=L4 pytest -v -s basic_correctness/test_basic_correctness.py -k "ray"

.buildkite/test_areas/entrypoints.yaml

@@ -22,9 +22,10 @@ steps:
   - vllm/
   - tests/entrypoints/llm
   commands:
-  - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/llm --ignore=entrypoints/llm/test_generate.py --ignore=entrypoints/llm/test_collective_rpc.py --ignore=entrypoints/llm/offline_mode
-  - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/llm/test_generate.py # it needs a clean process
-  - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/llm/offline_mode # Needs to avoid interference with other tests
+  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+  - pytest -v -s entrypoints/llm --ignore=entrypoints/llm/test_generate.py --ignore=entrypoints/llm/test_collective_rpc.py --ignore=entrypoints/llm/offline_mode
+  - pytest -v -s entrypoints/llm/test_generate.py # it needs a clean process
+  - pytest -v -s entrypoints/llm/offline_mode # Needs to avoid interference with other tests
   mirror:
     amd:
       device: mi325_1
@@ -40,8 +41,9 @@ steps:
   - vllm/
   - tests/entrypoints/serve
   commands:
-  - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/serve --ignore=entrypoints/serve/dev/rpc
-  - VLLM_WORKER_MULTIPROC_METHOD=spawn PYTHONPATH=/vllm-workspace pytest -v -s entrypoints/serve/dev/rpc
+  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+  - pytest -v -s entrypoints/serve --ignore=entrypoints/serve/dev/rpc
+  - PYTHONPATH=/vllm-workspace pytest -v -s entrypoints/serve/dev/rpc
   mirror:
     amd:
       device: mi325_1
@@ -57,7 +59,8 @@ steps:
   - tests/entrypoints/openai
   - tests/entrypoints/test_chat_utils
   commands:
-  - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/openai --ignore=entrypoints/openai/completion --ignore=entrypoints/openai/chat_completion --ignore=entrypoints/openai/responses --ignore=entrypoints/openai/correctness
+  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+  - pytest -v -s entrypoints/openai --ignore=entrypoints/openai/completion --ignore=entrypoints/openai/chat_completion --ignore=entrypoints/openai/responses --ignore=entrypoints/openai/correctness
   mirror:
     amd:
       device: mi325_1
@@ -74,8 +77,9 @@ steps:
   - tests/entrypoints/openai
   - tests/entrypoints/test_chat_utils
   commands:
-  - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/openai/chat_completion
-  - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/openai/completion --ignore=entrypoints/openai/completion/test_tensorizer_entrypoint.py
+  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+  - pytest -v -s entrypoints/openai/chat_completion
+  - pytest -v -s entrypoints/openai/completion --ignore=entrypoints/openai/completion/test_tensorizer_entrypoint.py
   mirror:
     amd:
       device: mi325_1
@@ -124,7 +128,8 @@ steps:
   - vllm/
   - tests/entrypoints/speech_to_text
   commands:
-  - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/speech_to_text
+  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+  - pytest -v -s entrypoints/speech_to_text
 
 - label: Entrypoints Integration (Multimodal)
   device: h200_35gb
@@ -135,7 +140,8 @@ steps:
   - vllm/
   - tests/entrypoints/multimodal
   commands:
-  - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/multimodal
+  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+  - pytest -v -s entrypoints/multimodal
 
 - label: Entrypoints Integration (Pooling)
   key: entrypoints-integration-pooling
@@ -145,7 +151,8 @@ steps:
   - vllm/
   - tests/entrypoints/pooling
   commands:
-  - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/pooling
+  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+  - pytest -v -s entrypoints/pooling
 
 - label: OpenAI API Correctness
   key: openai-api-correctness

.buildkite/test_areas/kernels.yaml

@@ -74,20 +74,6 @@ steps:
   commands:
     - pytest -v -s kernels/attention --shard-id=$$BUILDKITE_PARALLEL_JOB --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT
   parallelism: 2
-  mirror:
-    amd:
-      device: mi325_1
-      timeout_in_minutes: 55
-      depends_on:
-      - image-build-amd
-      source_file_dependencies:
-      - csrc/attention/
-      - vllm/v1/attention
-      - vllm/model_executor/layers/attention
-      - tests/kernels/attention
-      - vllm/_aiter_ops.py
-      - vllm/envs.py
-      - vllm/platforms/rocm.py
 
 - label: Kernels Attention DiffKV Test (H100)
   key: kernels-attention-diffkv-test-h100
@@ -142,22 +128,6 @@ steps:
     - pytest -v -s kernels/moe --ignore=kernels/moe/test_modular_oai_triton_moe.py --shard-id=$$BUILDKITE_PARALLEL_JOB --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT
     - pytest -v -s kernels/moe/test_modular_oai_triton_moe.py --shard-id=$$BUILDKITE_PARALLEL_JOB --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT
   parallelism: 5
-  mirror:
-    amd:
-      device: mi325_1
-      timeout_in_minutes: 50
-      source_file_dependencies:
-      - csrc/quantization/cutlass_w8a8/moe/
-      - csrc/moe/
-      - tests/kernels/moe
-      - vllm/model_executor/layers/fused_moe/
-      - vllm/distributed/device_communicators/
-      - vllm/envs.py
-      - vllm/config
-      - vllm/_aiter_ops.py
-      - vllm/platforms/rocm.py
-      depends_on:
-      - image-build-amd
 
 - label: Kernels Mamba Test
   key: kernels-mamba-test

.buildkite/test_areas/lm_eval.yaml

@@ -50,7 +50,8 @@ steps:
   - csrc/
   - vllm/model_executor/layers/quantization
   commands:
-    - VLLM_USE_DEEP_GEMM=0 pytest -s -v test_lm_eval_correctness.py --config-list-file=configs/models-large-hopper.txt --tp-size=4  # Triton is faster than DeepGEMM for H100
+    - export VLLM_USE_DEEP_GEMM=0  # We found Triton is faster than DeepGEMM for H100
+    - pytest -s -v test_lm_eval_correctness.py --config-list-file=configs/models-large-hopper.txt --tp-size=4
 
 - label: LM Eval Small Models (B200)
   key: lm-eval-small-models-b200
@@ -107,7 +108,8 @@ steps:
       depends_on:
       - image-build-amd
       commands:
-      - VLLM_WORKER_MULTIPROC_METHOD=spawn PYTORCH_ROCM_ARCH=gfx942 pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=configs/models-mi3xx.txt  # Limit Quark compilation to save time
+      - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+      - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=configs/models-mi3xx.txt
 
 - label: MoE Refactor Integration Test (H100 - TEMPORARY)
   key: moe-refactor-integration-test-h100-temporary

.buildkite/test_areas/lora.yaml

@@ -36,14 +36,14 @@ steps:
   commands:
     # FIXIT: find out which code initialize cuda before running the test
     # before the fix, we need to use spawn to test it
-    #
+    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
     # Alot of these tests are on the edge of OOMing
-    #
+    - export PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True
     # There is some Tensor Parallelism related processing logic in LoRA that
     # requires multi-GPU testing for validation.
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True pytest -v -s -x lora/test_chatglm3_tp.py
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True pytest -v -s -x lora/test_llama_tp.py
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True pytest -v -s -x lora/test_qwen3_with_multi_loras.py
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True pytest -v -s -x lora/test_olmoe_tp.py
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True pytest -v -s -x lora/test_gptoss_tp.py
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True pytest -v -s -x lora/test_qwen35_densemodel_lora.py
+    - pytest -v -s -x lora/test_chatglm3_tp.py
+    - pytest -v -s -x lora/test_llama_tp.py
+    - pytest -v -s -x lora/test_qwen3_with_multi_loras.py
+    - pytest -v -s -x lora/test_olmoe_tp.py
+    - pytest -v -s -x lora/test_gptoss_tp.py
+    - pytest -v -s -x lora/test_qwen35_densemodel_lora.py

.buildkite/test_areas/misc.yaml

@@ -18,8 +18,9 @@ steps:
     - vllm/v1/
     - tests/v1/spec_decode
   commands:
+    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
     # TODO: create another `optional` test group for slow tests
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s -m 'not slow_test' v1/spec_decode
+    - pytest -v -s -m 'not slow_test' v1/spec_decode
   mirror:
     amd:
       device: mi300_1
@@ -49,11 +50,12 @@ steps:
     - tests/v1/test_request.py
     - tests/v1/test_outputs.py
   commands:
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s v1/sample
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s v1/logits_processors
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s v1/test_oracle.py
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s v1/test_request.py
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s v1/test_outputs.py
+    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+    - pytest -v -s v1/sample
+    - pytest -v -s v1/logits_processors
+    - pytest -v -s v1/test_oracle.py
+    - pytest -v -s v1/test_request.py
+    - pytest -v -s v1/test_outputs.py
   mirror:
     amd:
       device: mi325_1
@@ -91,23 +93,18 @@ steps:
     - tests/entrypoints/openai/correctness/test_lmeval.py
   commands:
     - bash /vllm-workspace/.buildkite/scripts/install-kv-connectors.sh
+    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
     # split the test to avoid interference
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s -m 'not cpu_test' v1/core
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s v1/executor
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s v1/kv_offload
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s v1/simple_kv_offload
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s v1/worker
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s -m 'not cpu_test' v1/kv_connector/unit
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s -m 'not cpu_test' v1/metrics
+    - pytest -v -s -m 'not cpu_test' v1/core
+    - pytest -v -s v1/executor
+    - pytest -v -s v1/kv_offload
+    - pytest -v -s v1/simple_kv_offload
+    - pytest -v -s v1/worker
+    - pytest -v -s -m 'not cpu_test' v1/kv_connector/unit
+    - pytest -v -s -m 'not cpu_test' v1/metrics
     # Integration test for streaming correctness (requires special branch).
     - pip install -U git+https://github.com/robertgshaw2-redhat/lm-evaluation-harness.git@streaming-api
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/openai/correctness/test_lmeval.py::test_lm_eval_accuracy_v1_engine
-  mirror:
-    amd:
-      device: mi325_1
-      timeout_in_minutes: 60
-      depends_on:
-      - image-build-amd
+    - pytest -v -s entrypoints/openai/correctness/test_lmeval.py::test_lm_eval_accuracy_v1_engine
 
 - label: V1 Others (CPU)
   key: v1-others-cpu
@@ -150,7 +147,8 @@ steps:
     - vllm/distributed/kv_transfer/kv_connector/v1/example_hidden_states_connector.py
     - tests/v1/kv_connector/extract_hidden_states_integration
   commands:
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s v1/kv_connector/extract_hidden_states_integration
+    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+    - pytest -v -s v1/kv_connector/extract_hidden_states_integration
 
 - label: Extract Hidden States Integration (2 GPUs)
   key: extract-hidden-states-integration-2-gpus
@@ -163,7 +161,8 @@ steps:
     - vllm/distributed/kv_transfer/kv_connector/v1/example_hidden_states_connector.py
     - tests/v1/kv_connector/extract_hidden_states_integration
   commands:
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s -m 'distributed' v1/kv_connector/extract_hidden_states_integration
+    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+    - pytest -v -s -m 'distributed' v1/kv_connector/extract_hidden_states_integration
 
 - label: Regression
   key: regression
@@ -355,9 +354,10 @@ steps:
     - vllm/model_executor/layers
     - tests/v1/determinism/
   commands:
+    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
     - pip install pytest-timeout pytest-forked
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s v1/determinism/test_batch_invariance.py
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn VLLM_TEST_MODEL=deepseek-ai/DeepSeek-V2-Lite-Chat pytest -v -s v1/determinism/test_batch_invariance.py::test_v1_generation_is_deterministic_across_batch_sizes_with_needle[TRITON_MLA]
+    - pytest -v -s v1/determinism/test_batch_invariance.py
+    - VLLM_TEST_MODEL=deepseek-ai/DeepSeek-V2-Lite-Chat pytest -v -s v1/determinism/test_batch_invariance.py::test_v1_generation_is_deterministic_across_batch_sizes_with_needle[TRITON_MLA]
 
 - label: Batch Invariance (H100)
   key: batch-invariance-h100
@@ -368,11 +368,12 @@ steps:
     - vllm/model_executor/layers
     - tests/v1/determinism/
   commands:
+    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
     - pip install pytest-timeout pytest-forked
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s v1/determinism/test_batch_invariance.py
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s v1/determinism/test_rms_norm_batch_invariant.py
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn VLLM_TEST_MODEL=deepseek-ai/DeepSeek-V2-Lite-Chat pytest -v -s v1/determinism/test_batch_invariance.py::test_v1_generation_is_deterministic_across_batch_sizes_with_needle[TRITON_MLA]
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn VLLM_TEST_MODEL=Qwen/Qwen3-30B-A3B-Thinking-2507-FP8 pytest -v -s v1/determinism/test_batch_invariance.py::test_v1_generation_is_deterministic_across_batch_sizes_with_needle[FLASH_ATTN]
+    - pytest -v -s v1/determinism/test_batch_invariance.py
+    - pytest -v -s v1/determinism/test_rms_norm_batch_invariant.py
+    - VLLM_TEST_MODEL=deepseek-ai/DeepSeek-V2-Lite-Chat pytest -v -s v1/determinism/test_batch_invariance.py::test_v1_generation_is_deterministic_across_batch_sizes_with_needle[TRITON_MLA]
+    - VLLM_TEST_MODEL=Qwen/Qwen3-30B-A3B-Thinking-2507-FP8 pytest -v -s v1/determinism/test_batch_invariance.py::test_v1_generation_is_deterministic_across_batch_sizes_with_needle[FLASH_ATTN]
 
 - label: Batch Invariance (B200)
   key: batch-invariance-b200
@@ -383,13 +384,14 @@ steps:
     - vllm/model_executor/layers
     - tests/v1/determinism/
   commands:
+    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
     - pip install pytest-timeout pytest-forked
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s v1/determinism/test_batch_invariance.py
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s v1/determinism/test_rms_norm_batch_invariant.py
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn VLLM_TEST_MODEL=deepseek-ai/DeepSeek-V2-Lite-Chat pytest -v -s v1/determinism/test_batch_invariance.py::test_v1_generation_is_deterministic_across_batch_sizes_with_needle[TRITON_MLA]
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn VLLM_TEST_MODEL=Qwen/Qwen3-30B-A3B-Thinking-2507-FP8 pytest -v -s v1/determinism/test_batch_invariance.py::test_v1_generation_is_deterministic_across_batch_sizes_with_needle[FLASH_ATTN]
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s v1/determinism/test_nvfp4_batch_invariant.py
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s v1/determinism/test_nvfp4_batch_invariant_scaled_mm.py
+    - pytest -v -s v1/determinism/test_batch_invariance.py
+    - pytest -v -s v1/determinism/test_rms_norm_batch_invariant.py
+    - VLLM_TEST_MODEL=deepseek-ai/DeepSeek-V2-Lite-Chat pytest -v -s v1/determinism/test_batch_invariance.py::test_v1_generation_is_deterministic_across_batch_sizes_with_needle[TRITON_MLA]
+    - VLLM_TEST_MODEL=Qwen/Qwen3-30B-A3B-Thinking-2507-FP8 pytest -v -s v1/determinism/test_batch_invariance.py::test_v1_generation_is_deterministic_across_batch_sizes_with_needle[FLASH_ATTN]
+    - pytest -v -s v1/determinism/test_nvfp4_batch_invariant.py
+    - pytest -v -s v1/determinism/test_nvfp4_batch_invariant_scaled_mm.py
   
 - label: Acceptance Length Test (Large Models) # optional
   device: h200_35gb
@@ -404,4 +406,5 @@ steps:
   - vllm/model_executor/models/mlp_speculator.py
   - tests/v1/spec_decode/test_acceptance_length.py
   commands:
-    - VLLM_ALLOW_INSECURE_SERIALIZATION=1 pytest -v -s v1/spec_decode/test_acceptance_length.py -m slow_test
+    - export VLLM_ALLOW_INSECURE_SERIALIZATION=1
+    - pytest -v -s v1/spec_decode/test_acceptance_length.py -m slow_test

.buildkite/test_areas/model_executor.yaml

@@ -13,12 +13,13 @@ steps:
   - tests/entrypoints/openai/completion/test_tensorizer_entrypoint.py
   commands:
     - apt-get update && apt-get install -y curl libsodium23
+    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+    # Dump tracebacks of all threads if a test hangs, so a wedged GPU/CUDA
+    # init surfaces a stack instead of silently stalling.
+    - export PYTHONFAULTHANDLER=1
     # Per-test watchdog: a single hung test (e.g. stuck during engine/CUDA
     # init) fails fast with a traceback instead of running until the global
     # build timeout. The `thread` method also handles hangs inside C/CUDA
     # calls that the signal method cannot interrupt.
-    #
-    # Env vars are inlined because CONTINUE_ON_FAILURE wraps each command
-    # in a subshell, so a standalone `export` would be lost.
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn PYTHONFAULTHANDLER=1 pytest -v -s model_executor -m '(not slow_test)' --timeout=900 --timeout-method=thread
-    - VLLM_WORKER_MULTIPROC_METHOD=spawn PYTHONFAULTHANDLER=1 pytest -v -s entrypoints/openai/completion/test_tensorizer_entrypoint.py --timeout=900 --timeout-method=thread
+    - pytest -v -s model_executor -m '(not slow_test)' --timeout=900 --timeout-method=thread
+    - pytest -v -s entrypoints/openai/completion/test_tensorizer_entrypoint.py --timeout=900 --timeout-method=thread

.buildkite/test_areas/model_runner_v2.yaml

@@ -16,14 +16,15 @@ steps:
   - tests/entrypoints/llm/test_struct_output_generate.py
   commands:
   - set -x
-  - VLLM_USE_V2_MODEL_RUNNER=1 pytest -v -s v1/engine/test_llm_engine.py -k "not test_engine_metrics"
+  - export VLLM_USE_V2_MODEL_RUNNER=1
+  - pytest -v -s v1/engine/test_llm_engine.py -k "not test_engine_metrics"
   # This requires eager until we sort out CG correctness issues.
   # TODO: remove ENFORCE_EAGER here after https://github.com/vllm-project/vllm/pull/32936 is merged.
-  - VLLM_USE_V2_MODEL_RUNNER=1 ENFORCE_EAGER=1 pytest -v -s v1/e2e/general/test_async_scheduling.py -k "not ngram"
-  - VLLM_USE_V2_MODEL_RUNNER=1 pytest -v -s v1/e2e/general/test_context_length.py
-  - VLLM_USE_V2_MODEL_RUNNER=1 pytest -v -s v1/e2e/general/test_min_tokens.py
+  - ENFORCE_EAGER=1 pytest -v -s v1/e2e/general/test_async_scheduling.py -k "not ngram"
+  - pytest -v -s v1/e2e/general/test_context_length.py
+  - pytest -v -s v1/e2e/general/test_min_tokens.py
   # Temporary hack filter to exclude ngram spec decoding based tests.
-  - VLLM_USE_V2_MODEL_RUNNER=1 pytest -v -s entrypoints/llm/test_struct_output_generate.py -k "xgrammar and not speculative_config6 and not speculative_config7 and not speculative_config8 and not speculative_config0"
+  - pytest -v -s entrypoints/llm/test_struct_output_generate.py -k "xgrammar and not speculative_config6 and not speculative_config7 and not speculative_config8 and not speculative_config0"
 
 - label: Model Runner V2 Examples
   device: h200_35gb
@@ -41,25 +42,26 @@ steps:
     - examples/features/tensorize_vllm_model.py
   commands:
     - set -x
+    - export VLLM_USE_V2_MODEL_RUNNER=1
     - pip install tensorizer # for tensorizer test
-    - VLLM_USE_V2_MODEL_RUNNER=1 python3 basic/offline_inference/chat.py # for basic
-    - VLLM_USE_V2_MODEL_RUNNER=1 python3 basic/offline_inference/generate.py --model facebook/opt-125m
-    #- VLLM_USE_V2_MODEL_RUNNER=1 python3 basic/offline_inference/generate.py --model meta-llama/Llama-2-13b-chat-hf --cpu-offload-gb 10  # TODO
-    #- VLLM_USE_V2_MODEL_RUNNER=1 python3 basic/offline_inference/embed.py   # TODO
+    - python3 basic/offline_inference/chat.py # for basic
+    - python3 basic/offline_inference/generate.py --model facebook/opt-125m
+    #- python3 basic/offline_inference/generate.py --model meta-llama/Llama-2-13b-chat-hf --cpu-offload-gb 10  # TODO
+    #- python3 basic/offline_inference/embed.py   # TODO
     # for multi-modal models
-    - VLLM_USE_V2_MODEL_RUNNER=1 python3 generate/multimodal/audio_language_offline.py --seed 0
-    - VLLM_USE_V2_MODEL_RUNNER=1 python3 generate/multimodal/vision_language_offline.py --seed 0
-    - VLLM_USE_V2_MODEL_RUNNER=1 python3 generate/multimodal/vision_language_multi_image_offline.py --seed 0
-    - VLLM_USE_V2_MODEL_RUNNER=1 python3 generate/multimodal/encoder_decoder_multimodal_offline.py --model-type whisper --seed 0
+    - python3 generate/multimodal/audio_language_offline.py --seed 0
+    - python3 generate/multimodal/vision_language_offline.py --seed 0
+    - python3 generate/multimodal/vision_language_multi_image_offline.py --seed 0
+    - python3 generate/multimodal/encoder_decoder_multimodal_offline.py --model-type whisper --seed 0
     # for pooling models
-    - VLLM_USE_V2_MODEL_RUNNER=1 python3 pooling/embed/vision_embedding_offline.py --seed 0
+    - python3 pooling/embed/vision_embedding_offline.py --seed 0
     # for features demo
-    - VLLM_USE_V2_MODEL_RUNNER=1 python3 features/automatic_prefix_caching/prefix_caching_offline.py
-    - VLLM_USE_V2_MODEL_RUNNER=1 python3 deployment/llm_engine_example.py
-    - VLLM_USE_V2_MODEL_RUNNER=1 python3 features/tensorize_vllm_model.py --model facebook/opt-125m serialize --serialized-directory /tmp/ --suffix v1 && VLLM_USE_V2_MODEL_RUNNER=1 python3 features/tensorize_vllm_model.py --model facebook/opt-125m deserialize --path-to-tensors /tmp/vllm/facebook/opt-125m/v1/model.tensors
-    - VLLM_USE_V2_MODEL_RUNNER=1 python3 features/speculative_decoding/spec_decode_offline.py --test --method eagle --num_spec_tokens 3 --dataset-name hf --dataset-path philschmid/mt-bench --num-prompts 80 --temp 0 --top-p 1.0 --top-k -1 --tp 1 --enable-chunked-prefill --max-model-len 2048
+    - python3 features/automatic_prefix_caching/prefix_caching_offline.py
+    - python3 deployment/llm_engine_example.py
+    - python3 features/tensorize_vllm_model.py --model facebook/opt-125m serialize --serialized-directory /tmp/ --suffix v1 && python3 features/tensorize_vllm_model.py --model facebook/opt-125m deserialize --path-to-tensors /tmp/vllm/facebook/opt-125m/v1/model.tensors
+    - python3 features/speculative_decoding/spec_decode_offline.py --test --method eagle --num_spec_tokens 3 --dataset-name hf --dataset-path philschmid/mt-bench --num-prompts 80 --temp 0 --top-p 1.0 --top-k -1 --tp 1 --enable-chunked-prefill --max-model-len 2048
     # https://github.com/vllm-project/vllm/pull/26682 uses slightly more memory in PyTorch 2.9+ causing this test to OOM in 1xL4 GPU
-    - VLLM_USE_V2_MODEL_RUNNER=1 python3 features/speculative_decoding/spec_decode_offline.py --test --method eagle3 --num_spec_tokens 3 --dataset-name hf --dataset-path philschmid/mt-bench --num-prompts 80 --temp 0 --top-p 1.0 --top-k -1 --tp 1 --enable-chunked-prefill --max-model-len 1536
+    - python3 features/speculative_decoding/spec_decode_offline.py --test --method eagle3 --num_spec_tokens 3 --dataset-name hf --dataset-path philschmid/mt-bench --num-prompts 80 --temp 0 --top-p 1.0 --top-k -1 --tp 1 --enable-chunked-prefill --max-model-len 1536
 
 - label: Model Runner V2 Distributed (2 GPUs)
   key: model-runner-v2-distributed-2-gpus
@@ -74,11 +76,13 @@ steps:
     - tests/v1/distributed/test_eagle_dp.py
   commands:
     - set -x
+    - export VLLM_USE_V2_MODEL_RUNNER=1
     # The "and not True" here is a hacky way to exclude the prompt_embeds cases which aren't yet supported.
-    - VLLM_USE_V2_MODEL_RUNNER=1 TARGET_TEST_SUITE=L4 pytest -v -s basic_correctness/test_basic_correctness.py -m 'distributed(num_gpus=2)' -k "not ray and not True"
+    - TARGET_TEST_SUITE=L4 pytest -v -s basic_correctness/test_basic_correctness.py -m 'distributed(num_gpus=2)' -k "not ray and not True"
     # https://github.com/NVIDIA/nccl/issues/1838
-    - VLLM_USE_V2_MODEL_RUNNER=1 NCCL_CUMEM_HOST_ENABLE=0 TP_SIZE=1 DP_SIZE=2 pytest -v -s v1/distributed/test_async_llm_dp.py -k "not ray"
-    - VLLM_USE_V2_MODEL_RUNNER=1 NCCL_CUMEM_HOST_ENABLE=0 TP_SIZE=1 DP_SIZE=2 pytest -v -s v1/distributed/test_eagle_dp.py
+    - export NCCL_CUMEM_HOST_ENABLE=0
+    - TP_SIZE=1 DP_SIZE=2 pytest -v -s v1/distributed/test_async_llm_dp.py -k "not ray"
+    - TP_SIZE=1 DP_SIZE=2 pytest -v -s v1/distributed/test_eagle_dp.py
 
 - label: Model Runner V2 Pipeline Parallelism (4 GPUs)
   key: model-runner-v2-pipeline-parallelism-4-gpus
@@ -93,9 +97,10 @@ steps:
     - tests/v1/distributed/test_pp_dp_v2.py
   commands:
     - set -x
-    - VLLM_USE_V2_MODEL_RUNNER=1 pytest -v -s distributed/test_pipeline_parallel.py -k "not ray and not Jamba"
-    - VLLM_USE_V2_MODEL_RUNNER=1 pytest -v -s distributed/test_pp_cudagraph.py -k "not ray"
-    - VLLM_USE_V2_MODEL_RUNNER=1 pytest -v -s v1/distributed/test_pp_dp_v2.py
+    - export VLLM_USE_V2_MODEL_RUNNER=1
+    - pytest -v -s distributed/test_pipeline_parallel.py -k "not ray and not Jamba"
+    - pytest -v -s distributed/test_pp_cudagraph.py -k "not ray"
+    - pytest -v -s v1/distributed/test_pp_dp_v2.py
 
 - label: Model Runner V2 Spec Decode
   device: h200_35gb
@@ -110,7 +115,8 @@ steps:
   - tests/v1/e2e/spec_decode/test_spec_decode.py
   commands:
   - set -x
-  - VLLM_USE_V2_MODEL_RUNNER=1 pytest -v -s v1/spec_decode/test_max_len.py -k "eagle or mtp"
-  - VLLM_USE_V2_MODEL_RUNNER=1 pytest -v -s v1/spec_decode/test_rejection_sampler_utils.py
-  - VLLM_USE_V2_MODEL_RUNNER=1 pytest -v -s v1/spec_decode/test_synthetic_rejection_sampler_utils.py
-  - VLLM_USE_V2_MODEL_RUNNER=1 pytest -v -s v1/e2e/spec_decode/test_spec_decode.py -k "eagle or mtp"
+  - export VLLM_USE_V2_MODEL_RUNNER=1
+  - pytest -v -s v1/spec_decode/test_max_len.py -k "eagle or mtp"
+  - pytest -v -s v1/spec_decode/test_rejection_sampler_utils.py
+  - pytest -v -s v1/spec_decode/test_synthetic_rejection_sampler_utils.py
+  - pytest -v -s v1/e2e/spec_decode/test_spec_decode.py -k "eagle or mtp"

.buildkite/test_areas/pytorch.yaml

@@ -107,12 +107,6 @@ steps:
     - tests/compile/passes
   commands:
   - pytest -s -v compile/passes --ignore compile/passes/distributed
-  mirror:
-    amd:
-      device: mi300_1
-      timeout_in_minutes: 180
-      depends_on:
-      - image-build-amd
 
 - label: PyTorch Fullgraph Smoke Test
   key: pytorch-fullgraph-smoke-test

.buildkite/test_areas/rust_frontend.yaml

@@ -23,15 +23,17 @@ steps:
   # - tests/entrypoints/openai/test_uds.py
   - tests/v1/sample/test_logprobs_e2e.py
   commands:
-  - VLLM_USE_RUST_FRONTEND=1 VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s benchmarks/test_serve_cli.py -k "not insecure and not (test_bench_serve and not test_bench_serve_chat)"
-  - VLLM_USE_RUST_FRONTEND=1 VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/openai/chat_completion/test_chat_completion.py
-  # - VLLM_USE_RUST_FRONTEND=1 VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/openai/chat_completion/test_chat_logit_bias_validation.py -k "not invalid"
+  - export VLLM_USE_RUST_FRONTEND=1
+  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+  - pytest -v -s benchmarks/test_serve_cli.py -k "not insecure and not (test_bench_serve and not test_bench_serve_chat)"
+  - pytest -v -s entrypoints/openai/chat_completion/test_chat_completion.py
+  # - pytest -v -s entrypoints/openai/chat_completion/test_chat_logit_bias_validation.py -k "not invalid"
 
-  # - VLLM_USE_RUST_FRONTEND=1 VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/openai/completion/test_prompt_validation.py -k "not prompt_embeds"
-  - VLLM_USE_RUST_FRONTEND=1 VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/openai/completion/test_shutdown.py -k "not engine_failure and not test_abort_timeout_exits_quickly"
-  # - VLLM_USE_RUST_FRONTEND=1 VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/openai/test_return_token_ids.py
-  # - VLLM_USE_RUST_FRONTEND=1 VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/openai/test_uds.py
-  - VLLM_USE_RUST_FRONTEND=1 VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s v1/sample/test_logprobs_e2e.py -k "test_prompt_logprobs_e2e_server"
+  # - pytest -v -s entrypoints/openai/completion/test_prompt_validation.py -k "not prompt_embeds"
+  - pytest -v -s entrypoints/openai/completion/test_shutdown.py -k "not engine_failure and not test_abort_timeout_exits_quickly"
+  # - pytest -v -s entrypoints/openai/test_return_token_ids.py
+  # - pytest -v -s entrypoints/openai/test_uds.py
+  - pytest -v -s v1/sample/test_logprobs_e2e.py -k "test_prompt_logprobs_e2e_server"
 
 - label: Rust Frontend Serve/Admin Coverage
   timeout_in_minutes: 60
@@ -49,11 +51,13 @@ steps:
   - tests/entrypoints/serve/instrumentator/test_metrics.py
   # - tests/entrypoints/serve/dev/test_sleep.py
   commands:
-  # - VLLM_USE_RUST_FRONTEND=1 VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/serve/dev/rpc/test_collective_rpc.py
-  - VLLM_USE_RUST_FRONTEND=1 VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/serve/instrumentator/test_basic.py -k "not show_version and not server_load"
-  - VLLM_USE_RUST_FRONTEND=1 VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/serve/disagg/test_serving_tokens.py -k "not stream and not lora and not test_generate_logprobs and not stop_string_workflow"
-  - VLLM_USE_RUST_FRONTEND=1 VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/serve/instrumentator/test_metrics.py -k "text and not show and not run_batch and not test_metrics_counts and not test_metrics_exist"
-  # - VLLM_USE_RUST_FRONTEND=1 VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s entrypoints/serve/dev/test_sleep.py
+  - export VLLM_USE_RUST_FRONTEND=1
+  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+  # - pytest -v -s entrypoints/serve/dev/rpc/test_collective_rpc.py
+  - pytest -v -s entrypoints/serve/instrumentator/test_basic.py -k "not show_version and not server_load"
+  - pytest -v -s entrypoints/serve/disagg/test_serving_tokens.py -k "not stream and not lora and not test_generate_logprobs and not stop_string_workflow"
+  - pytest -v -s entrypoints/serve/instrumentator/test_metrics.py -k "text and not show and not run_batch and not test_metrics_counts and not test_metrics_exist"
+  # - pytest -v -s entrypoints/serve/dev/test_sleep.py
 
 - label: Rust Frontend Core Correctness
   timeout_in_minutes: 30
@@ -65,7 +69,9 @@ steps:
   - tests/utils.py
   - tests/entrypoints/openai/correctness/test_lmeval.py
   commands:
-  - VLLM_USE_RUST_FRONTEND=1 VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -s entrypoints/openai/correctness/test_lmeval.py::test_lm_eval_accuracy_v1_engine
+  - export VLLM_USE_RUST_FRONTEND=1
+  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+  - pytest -s entrypoints/openai/correctness/test_lmeval.py::test_lm_eval_accuracy_v1_engine
 
 - label: Rust Frontend Tool Use
   timeout_in_minutes: 60
@@ -77,7 +83,9 @@ steps:
   - tests/utils.py
   - tests/tool_use/
   commands:
-  - VLLM_USE_RUST_FRONTEND=1 VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -v -s tool_use --ignore=tool_use/mistral --models llama3.2 -k "not test_response_format_with_tool_choice_required and not test_parallel_tool_calls_false and not test_tool_call_and_choice"
+  - export VLLM_USE_RUST_FRONTEND=1
+  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+  - pytest -v -s tool_use --ignore=tool_use/mistral --models llama3.2 -k "not test_response_format_with_tool_choice_required and not test_parallel_tool_calls_false and not test_tool_call_and_choice"
 
 - label: Rust Frontend Distributed
   timeout_in_minutes: 30
@@ -95,6 +103,9 @@ steps:
   - tests/v1/distributed/test_hybrid_lb_dp.py
   - tests/v1/distributed/test_internal_lb_dp.py
   commands:
-  - VLLM_USE_RUST_FRONTEND=1 VLLM_WORKER_MULTIPROC_METHOD=spawn NCCL_CUMEM_HOST_ENABLE=0 TP_SIZE=1 DP_SIZE=4 pytest -v -s v1/distributed/test_internal_lb_dp.py -k "not 4 and not server_info"
-  - VLLM_USE_RUST_FRONTEND=1 VLLM_WORKER_MULTIPROC_METHOD=spawn NCCL_CUMEM_HOST_ENABLE=0 TP_SIZE=1 DP_SIZE=2 pytest -v -s v1/distributed/test_external_lb_dp.py -k "not 4 and not server_info"
-  - VLLM_USE_RUST_FRONTEND=1 VLLM_WORKER_MULTIPROC_METHOD=spawn NCCL_CUMEM_HOST_ENABLE=0 TP_SIZE=1 DP_SIZE=4 pytest -v -s v1/distributed/test_hybrid_lb_dp.py -k "not 4 and not server_info"
+  - export VLLM_USE_RUST_FRONTEND=1
+  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+  - export NCCL_CUMEM_HOST_ENABLE=0
+  - TP_SIZE=1 DP_SIZE=4 pytest -v -s v1/distributed/test_internal_lb_dp.py -k "not 4 and not server_info"
+  - TP_SIZE=1 DP_SIZE=2 pytest -v -s v1/distributed/test_external_lb_dp.py -k "not 4 and not server_info"
+  - TP_SIZE=1 DP_SIZE=4 pytest -v -s v1/distributed/test_hybrid_lb_dp.py -k "not 4 and not server_info"

.buildkite/test_areas/spec_decode.yaml

@@ -12,20 +12,6 @@ steps:
     - tests/v1/e2e/spec_decode/
   commands:
     - pytest -v -s v1/e2e/spec_decode -k "eagle_correctness"
-  mirror:
-    amd:
-      device: mi325_1
-      timeout_in_minutes: 45
-      depends_on:
-      - image-build-amd
-      source_file_dependencies:
-      - vllm/v1/spec_decode/
-      - vllm/v1/worker/gpu/spec_decode/
-      - vllm/model_executor/model_loader/
-      - vllm/v1/sample/
-      - vllm/model_executor/layers/
-      - tests/v1/e2e/spec_decode/
-      - vllm/platforms/rocm.py
 
 - label: Spec Decode Eagle Nightly B200
   key: spec-decode-eagle-nightly-b200
@@ -153,7 +139,8 @@ steps:
     - vllm/model_executor/models/qwen3_dflash.py
     - tests/v1/spec_decode/test_speculators_correctness.py
   commands:
-    - VLLM_ALLOW_INSECURE_SERIALIZATION=1 pytest -v -s v1/spec_decode/test_speculators_correctness.py -m slow_test
+    - export VLLM_ALLOW_INSECURE_SERIALIZATION=1
+    - pytest -v -s v1/spec_decode/test_speculators_correctness.py -m slow_test
 
 - label: Spec Decode MTP hybrid (B200)
   timeout_in_minutes: 30

AGENTS.md

@@ -135,12 +135,6 @@ Do not modify code in these areas without first reading and following the
 linked guide. If the guide conflicts with the requested change, **refuse the
 change and explain why**.
 
-Security reviewers should start with [`SECURITY.md`](SECURITY.md),
-[`docs/usage/security.md`](docs/usage/security.md), and
-[`docs/contributing/vulnerability_management.md`](docs/contributing/vulnerability_management.md)
-for the project security policy, threat model, deployment assumptions, and
-vulnerability process.
-
 - **Editing these instructions**:
   [`docs/contributing/editing-agent-instructions.md`](docs/contributing/editing-agent-instructions.md)
   — Rules for modifying AGENTS.md or any domain-specific guide it references.

cmake/external_projects/fmha_sm100.cmake

@@ -17,7 +17,7 @@ else()
   FetchContent_Declare(
     fmha_sm100
     GIT_REPOSITORY https://github.com/vllm-project/MSA.git
-    GIT_TAG 544eee5e09ae2dfa774d5b06739013f9b7402c57
+    GIT_TAG fee783153f3efe57e3e933c5cb7e267a7cebcfb5
     GIT_PROGRESS TRUE
     CONFIGURE_COMMAND ""
     BUILD_COMMAND ""
@@ -36,13 +36,38 @@ set(FMHA_SM100_PY_ROOT "${fmha_sm100_SOURCE_DIR}/python/fmha_sm100")
 
 install(FILES
   "${FMHA_SM100_PY_ROOT}/__init__.py"
+  "${FMHA_SM100_PY_ROOT}/api.py"
+  "${FMHA_SM100_PY_ROOT}/bench_utils.py"
+  "${FMHA_SM100_PY_ROOT}/jit.py"
   "${FMHA_SM100_PY_ROOT}/sparse.py"
+  "${FMHA_SM100_PY_ROOT}/sparse_fmha_adapter.py"
   DESTINATION vllm/third_party/fmha_sm100
   COMPONENT fmha_sm100)
 
+install(DIRECTORY "${FMHA_SM100_PY_ROOT}/csrc/"
+  DESTINATION vllm/third_party/fmha_sm100/csrc
+  COMPONENT fmha_sm100
+  PATTERN "__pycache__" EXCLUDE
+  PATTERN "*.pyc" EXCLUDE
+  PATTERN ".git*" EXCLUDE)
+
 install(DIRECTORY "${FMHA_SM100_PY_ROOT}/cute/"
   DESTINATION vllm/third_party/fmha_sm100/cute
   COMPONENT fmha_sm100
   PATTERN "__pycache__" EXCLUDE
   PATTERN "*.pyc" EXCLUDE
   PATTERN ".git*" EXCLUDE)
+
+install(DIRECTORY "${FMHA_SM100_PY_ROOT}/cutlass/include/"
+  DESTINATION vllm/third_party/fmha_sm100/cutlass/include
+  COMPONENT fmha_sm100
+  PATTERN "__pycache__" EXCLUDE
+  PATTERN "*.pyc" EXCLUDE
+  PATTERN ".git*" EXCLUDE)
+
+install(DIRECTORY "${FMHA_SM100_PY_ROOT}/cutlass/tools/util/include/"
+  DESTINATION vllm/third_party/fmha_sm100/cutlass/tools/util/include
+  COMPONENT fmha_sm100
+  PATTERN "__pycache__" EXCLUDE
+  PATTERN "*.pyc" EXCLUDE
+  PATTERN ".git*" EXCLUDE)

csrc/cpu/cpu_wna16.cpp

@@ -4,9 +4,6 @@
 #ifdef CPU_CAPABILITY_AMXBF16
   #include "cpu/micro_gemm/cpu_micro_gemm_amx.hpp"
 #endif
-#if defined(__riscv_v)
-  #include "cpu/micro_gemm/cpu_micro_gemm_rvv.hpp"
-#endif
 #include "cpu/micro_gemm/cpu_micro_gemm_vec.hpp"
 
 #define VLLM_DISPATCH_CASE_16B_TYPES(...)                 \
@@ -322,8 +319,6 @@ void cpu_gemm_wna16(
       return ISA::AMX;
     } else if (isa_hint == "vec") {
       return ISA::VEC;
-    } else if (isa_hint == "rvv") {
-      return ISA::RVV;
     } else {
       TORCH_CHECK(false, "unsupported isa hint: " + isa_hint);
     }
@@ -402,40 +397,6 @@ void cpu_gemm_wna16(
             pack_factor);
         return;
       }
-    } else if (isa == ISA::RVV) {
-      using gemm_t = cpu_micro_gemm::MicroGemm<ISA::RVV, scalar_t>;
-      if (has_zp) {
-        using dequantizer_t = Dequantizer4b<scalar_t, ISA::RVV, true, false>;
-        cpu_gemm_wna16_impl<scalar_t, dequantizer_t, gemm_t>(
-            input.data_ptr<scalar_t>(), q_weight.data_ptr<int32_t>(),
-            output.data_ptr<scalar_t>(), scales.data_ptr<scalar_t>(), zeros_ptr,
-            g_idx_ptr, bias.has_value() ? bias->data_ptr<scalar_t>() : nullptr,
-            a_m_size, b_n_size, a_k_size, a_m_stride, output_m_stride,
-            scales_group_stride, zeros_group_stride, group_num, group_size,
-            pack_factor);
-        return;
-      }
-      if (use_desc_act) {
-        using dequantizer_t = Dequantizer4b<scalar_t, ISA::RVV, false, true>;
-        cpu_gemm_wna16_impl<scalar_t, dequantizer_t, gemm_t>(
-            input.data_ptr<scalar_t>(), q_weight.data_ptr<int32_t>(),
-            output.data_ptr<scalar_t>(), scales.data_ptr<scalar_t>(), zeros_ptr,
-            g_idx_ptr, bias.has_value() ? bias->data_ptr<scalar_t>() : nullptr,
-            a_m_size, b_n_size, a_k_size, a_m_stride, output_m_stride,
-            scales_group_stride, zeros_group_stride, group_num, group_size,
-            pack_factor);
-        return;
-      } else {
-        using dequantizer_t = Dequantizer4b<scalar_t, ISA::RVV, false, false>;
-        cpu_gemm_wna16_impl<scalar_t, dequantizer_t, gemm_t>(
-            input.data_ptr<scalar_t>(), q_weight.data_ptr<int32_t>(),
-            output.data_ptr<scalar_t>(), scales.data_ptr<scalar_t>(), zeros_ptr,
-            g_idx_ptr, bias.has_value() ? bias->data_ptr<scalar_t>() : nullptr,
-            a_m_size, b_n_size, a_k_size, a_m_stride, output_m_stride,
-            scales_group_stride, zeros_group_stride, group_num, group_size,
-            pack_factor);
-        return;
-      }
     }
   });
 }

csrc/cpu/micro_gemm/cpu_micro_gemm_rvv.hpp

@@ -1,228 +0,0 @@
-#ifndef CPU_MICRO_GEMM_RVV_HPP
-#define CPU_MICRO_GEMM_RVV_HPP
-
-#include "cpu/micro_gemm/cpu_micro_gemm_impl.hpp"
-
-#if defined(__riscv_v)
-
-namespace cpu_micro_gemm {
-namespace {
-
-constexpr int32_t RVV_MGEMM_N8 = 8;
-constexpr int32_t RVV_MGEMM_B_GROUP_STRIDE = 16;
-
-template <typename scalar_t>
-FORCE_INLINE fixed_fp32x8_t load_row8_b_as_f32(const scalar_t* ptr);
-
-template <>
-FORCE_INLINE fixed_fp32x8_t load_row8_b_as_f32<float>(const float* ptr) {
-  return RVVI(__riscv_vle32_v_f32, LMUL_256)(ptr, RVV_MGEMM_N8);
-}
-
-template <>
-FORCE_INLINE fixed_fp32x8_t
-load_row8_b_as_f32<c10::Half>(const c10::Half* ptr) {
-  #if defined(__riscv_zvfh)
-  fixed_fp16x8_t vec = RVVI(__riscv_vle16_v_f16, LMUL_128)(
-      reinterpret_cast<const _Float16*>(ptr), RVV_MGEMM_N8);
-  return RVVI(__riscv_vfwcvt_f_f_v_f32, LMUL_256)(vec, RVV_MGEMM_N8);
-  #else
-  alignas(32) float values[RVV_MGEMM_N8];
-  for (int32_t i = 0; i < RVV_MGEMM_N8; ++i) {
-    values[i] = static_cast<float>(ptr[i]);
-  }
-  return RVVI(__riscv_vle32_v_f32, LMUL_256)(values, RVV_MGEMM_N8);
-  #endif
-}
-
-template <>
-FORCE_INLINE fixed_fp32x8_t
-load_row8_b_as_f32<c10::BFloat16>(const c10::BFloat16* ptr) {
-  #if defined(__riscv_zvfbfmin)
-  fixed_u16x8_t raw = RVVI(__riscv_vle16_v_u16, LMUL_128)(
-      reinterpret_cast<const uint16_t*>(ptr), RVV_MGEMM_N8);
-  fixed_bf16x8_t vec =
-      RVVI4(__riscv_vreinterpret_v_u16, LMUL_128, _bf16, LMUL_128)(raw);
-  return RVVI(__riscv_vfwcvtbf16_f_f_v_f32, LMUL_256)(vec, RVV_MGEMM_N8);
-  #else
-  fixed_u16x8_t raw = RVVI(__riscv_vle16_v_u16, LMUL_128)(
-      reinterpret_cast<const uint16_t*>(ptr), RVV_MGEMM_N8);
-  auto wide = RVVI(__riscv_vzext_vf2_u32, LMUL_256)(raw, RVV_MGEMM_N8);
-  auto shifted = RVVI(__riscv_vsll_vx_u32, LMUL_256)(wide, 16, RVV_MGEMM_N8);
-  return RVVI4(__riscv_vreinterpret_v_u32, LMUL_256, _f32, LMUL_256)(shifted);
-  #endif
-}
-
-// Mx8 RVV kernel. B points at one 8-channel half of a 16-channel packed group,
-// with rows separated by RVV_MGEMM_B_GROUP_STRIDE scalar elements.
-template <int32_t M, typename scalar_t>
-FORCE_INLINE void gemm_micro_rvv_fma_mx8_ku4(const scalar_t* __restrict__ a_ptr,
-                                             const scalar_t* __restrict__ b_ptr,
-                                             float* __restrict__ c_ptr,
-                                             const int64_t lda,
-                                             const int64_t ldc, const int32_t k,
-                                             const bool accum_c) {
-  static_assert(0 < M && M <= 8);
-
-  #define RVV_ROWS_APPLY(OP) OP(0) OP(1) OP(2) OP(3) OP(4) OP(5) OP(6) OP(7)
-  #define RVV_IF_M(i) if constexpr (M > (i))
-
-  #define RVV_DECL_A(i) const scalar_t* __restrict__ a##i = a_ptr + (i) * lda;
-  RVV_ROWS_APPLY(RVV_DECL_A)
-  #undef RVV_DECL_A
-
-  #define RVV_DECL_ACC(i) fixed_fp32x8_t acc##i;
-  RVV_ROWS_APPLY(RVV_DECL_ACC)
-  #undef RVV_DECL_ACC
-
-  #define RVV_INIT_ACC(i)                                                  \
-    RVV_IF_M(i) {                                                          \
-      if (accum_c) {                                                       \
-        acc##i = RVVI(__riscv_vle32_v_f32, LMUL_256)(c_ptr + (i) * ldc,    \
-                                                     RVV_MGEMM_N8);        \
-      } else {                                                             \
-        acc##i = RVVI(__riscv_vfmv_v_f_f32, LMUL_256)(0.0f, RVV_MGEMM_N8); \
-      }                                                                    \
-    }
-  RVV_ROWS_APPLY(RVV_INIT_ACC)
-  #undef RVV_INIT_ACC
-
-  int32_t k_idx = 0;
-  for (; k_idx + 3 < k; k_idx += 4) {
-  #define RVV_FMA_ROW(i, K_OFFSET)                                     \
-    RVV_IF_M(i) {                                                      \
-      acc##i = RVVI(__riscv_vfmacc_vf_f32, LMUL_256)(                  \
-          acc##i, static_cast<float>(*(a##i + k_idx + (K_OFFSET))), b, \
-          RVV_MGEMM_N8);                                               \
-    }
-
-  #define RVV_STEP_K(K_OFFSET)                                      \
-    {                                                               \
-      fixed_fp32x8_t b = load_row8_b_as_f32<scalar_t>(              \
-          b_ptr + (k_idx + (K_OFFSET)) * RVV_MGEMM_B_GROUP_STRIDE); \
-      RVV_FMA_ROW(0, K_OFFSET)                                      \
-      RVV_FMA_ROW(1, K_OFFSET)                                      \
-      RVV_FMA_ROW(2, K_OFFSET)                                      \
-      RVV_FMA_ROW(3, K_OFFSET)                                      \
-      RVV_FMA_ROW(4, K_OFFSET)                                      \
-      RVV_FMA_ROW(5, K_OFFSET)                                      \
-      RVV_FMA_ROW(6, K_OFFSET)                                      \
-      RVV_FMA_ROW(7, K_OFFSET)                                      \
-    }
-
-    RVV_STEP_K(0)
-    RVV_STEP_K(1)
-    RVV_STEP_K(2)
-    RVV_STEP_K(3)
-  #undef RVV_STEP_K
-  #undef RVV_FMA_ROW
-  }
-
-  for (; k_idx < k; ++k_idx) {
-    fixed_fp32x8_t b =
-        load_row8_b_as_f32<scalar_t>(b_ptr + k_idx * RVV_MGEMM_B_GROUP_STRIDE);
-  #define RVV_TAIL_ROW(i)                                                \
-    RVV_IF_M(i) {                                                        \
-      acc##i = RVVI(__riscv_vfmacc_vf_f32, LMUL_256)(                    \
-          acc##i, static_cast<float>(*(a##i + k_idx)), b, RVV_MGEMM_N8); \
-    }
-    RVV_ROWS_APPLY(RVV_TAIL_ROW)
-  #undef RVV_TAIL_ROW
-  }
-
-  #define RVV_STORE_ROW(i)                                           \
-    RVV_IF_M(i) {                                                    \
-      RVVI(__riscv_vse32_v_f32, LMUL_256)(c_ptr + (i) * ldc, acc##i, \
-                                          RVV_MGEMM_N8);             \
-    }
-  RVV_ROWS_APPLY(RVV_STORE_ROW)
-  #undef RVV_STORE_ROW
-
-  #undef RVV_ROWS_APPLY
-  #undef RVV_IF_M
-}
-
-template <int32_t M, typename scalar_t>
-FORCE_INLINE void gemm_micro_rvv_mx32_ku4(DEFINE_CPU_MICRO_GEMM_PARAMS) {
-  static_assert(0 < M && M <= 8);
-  scalar_t* __restrict__ curr_b_0 = b_ptr;
-  scalar_t* __restrict__ curr_b_1 = b_ptr + b_n_group_stride;
-
-  gemm_micro_rvv_fma_mx8_ku4<M>(a_ptr, curr_b_0, c_ptr, lda, ldc, k, accum_c);
-  gemm_micro_rvv_fma_mx8_ku4<M>(a_ptr, curr_b_0 + RVV_MGEMM_N8,
-                                c_ptr + RVV_MGEMM_N8, lda, ldc, k, accum_c);
-  gemm_micro_rvv_fma_mx8_ku4<M>(a_ptr, curr_b_1, c_ptr + 16, lda, ldc, k,
-                                accum_c);
-  gemm_micro_rvv_fma_mx8_ku4<M>(a_ptr, curr_b_1 + RVV_MGEMM_N8, c_ptr + 24, lda,
-                                ldc, k, accum_c);
-}
-
-class TileGemmRVV {
- public:
-  template <typename scalar_t>
-  FORCE_INLINE static void gemm(DEFINE_CPU_MICRO_GEMM_PARAMS) {
-    switch (m) {
-      case 1:
-        gemm_micro_rvv_mx32_ku4<1>(CPU_MICRO_GEMM_PARAMS);
-        break;
-      case 2:
-        gemm_micro_rvv_mx32_ku4<2>(CPU_MICRO_GEMM_PARAMS);
-        break;
-      case 3:
-        gemm_micro_rvv_mx32_ku4<3>(CPU_MICRO_GEMM_PARAMS);
-        break;
-      case 4:
-        gemm_micro_rvv_mx32_ku4<4>(CPU_MICRO_GEMM_PARAMS);
-        break;
-      case 5:
-        gemm_micro_rvv_mx32_ku4<5>(CPU_MICRO_GEMM_PARAMS);
-        break;
-      case 6:
-        gemm_micro_rvv_mx32_ku4<6>(CPU_MICRO_GEMM_PARAMS);
-        break;
-      case 7:
-        gemm_micro_rvv_mx32_ku4<7>(CPU_MICRO_GEMM_PARAMS);
-        break;
-      case 8:
-        gemm_micro_rvv_mx32_ku4<8>(CPU_MICRO_GEMM_PARAMS);
-        break;
-    }
-  }
-};
-
-}  // namespace
-
-template <typename scalar_t>
-class MicroGemm<cpu_utils::ISA::RVV, scalar_t> {
- public:
-  static constexpr int32_t MaxMSize = 8;
-  static constexpr int32_t NSize = 32;
-
- public:
-  void gemm(DEFINE_CPU_MICRO_GEMM_PARAMS) {
-    TileGemmRVV::gemm<scalar_t>(CPU_MICRO_GEMM_PARAMS);
-  }
-
-  static void pack_weight(const scalar_t* __restrict__ weight,
-                          scalar_t* __restrict__ packed_weight,
-                          const int32_t output_size, const int32_t input_size) {
-    TORCH_CHECK_EQ(output_size % 16, 0);
-    for (int32_t o_idx = 0; o_idx < output_size; ++o_idx) {
-      const scalar_t* __restrict__ curr_weight = weight + o_idx * input_size;
-      scalar_t* __restrict__ curr_packed_weight =
-          packed_weight + (o_idx / 16) * (16 * input_size) + o_idx % 16;
-      for (int32_t i_idx = 0; i_idx < input_size; ++i_idx) {
-        *curr_packed_weight = *curr_weight;
-
-        curr_packed_weight += 16;
-        ++curr_weight;
-      }
-    }
-  }
-};
-
-}  // namespace cpu_micro_gemm
-
-#endif  // defined(__riscv_v)
-
-#endif  // CPU_MICRO_GEMM_RVV_HPP

csrc/cpu/utils.hpp

@@ -8,15 +8,13 @@
 #include "cpu/cpu_types.hpp"
 
 namespace cpu_utils {
-enum class ISA { AMX, VEC, RVV };
+enum class ISA { AMX, VEC };
 
 inline ISA get_isa(const std::string& isa) {
   if (isa == "amx") {
     return ISA::AMX;
   } else if (isa == "vec") {
     return ISA::VEC;
-  } else if (isa == "rvv") {
-    return ISA::RVV;
   } else {
     TORCH_CHECK(false, "Invalid isa type: " + isa);
   }

csrc/libtorch_stable/attention/mla/sm100_cutlass_mla_kernel.cu

@@ -136,12 +136,8 @@ typename T::Fmha::Arguments args_from_options(
   StrideQ stride_Q_pe = cute::make_tuple(
       static_cast<int64_t>(q_pe.stride(1)), _1{}, static_cast<int64_t>(q_pe.stride(0)));
 
-  // Read the token and page strides from the cache tensor instead of assuming
-  // packed pages, so strided views (e.g. per-layer views into a cross-layer
-  // block-major cache) are addressed correctly.
   StrideK stride_C = cute::make_tuple(
-      static_cast<int64_t>(kv_c_and_k_pe_cache.stride(1)), _1{},
-      static_cast<int64_t>(kv_c_and_k_pe_cache.stride(0)));
+      static_cast<int64_t>(0 + D_latent + D_rope), _1{}, static_cast<int64_t>(page_size * (D_latent + D_rope)));
   StrideLSE stride_PT = cute::make_stride(_1{}, page_count_per_seq);
   StrideLSE stride_LSE = cute::make_tuple(_1{}, 0 + H);
   StrideO stride_O = cute::make_tuple(static_cast<int64_t>(0 + D_latent), _1{}, static_cast<int64_t>(0 + H * D_latent));

csrc/libtorch_stable/cache_kernels.cu

@@ -549,7 +549,7 @@ __global__ void indexer_k_quant_and_cache_kernel(
     const int head_dim,                        // dimension of each head
     const int quant_block_size,                // quantization block size
     const int cache_block_size,                // cache block size
-    const int64_t cache_block_stride,  // stride for each block in kv_cache
+    const int cache_stride,  // stride for each token in kv_cache
 
     const bool use_ue8m0  // use ue8m0 scale format
 ) {
@@ -590,15 +590,16 @@ __global__ void indexer_k_quant_and_cache_kernel(
     scale = exp2f(ceilf(log2f(scale)));
   }
 
-  const int64_t dst_offset =
-      block_idx * cache_block_stride + block_offset * head_dim + head_dim_idx;
+  const int64_t dst_offset = block_idx * cache_block_size * cache_stride +
+                             block_offset * head_dim + head_dim_idx;
   for (int i = 0; i < VEC_SIZE; i++) {
     kv_cache[dst_offset + i] =
         fp8::scaled_convert<cache_t, scalar_t, kv_dt>(k_val_ptr[i], scale);
   }
   if (threadIdx.x == 0) {
     const int64_t dst_scale_idx =
-        block_idx * cache_block_stride + cache_block_size * head_dim +
+        block_idx * cache_block_size * cache_stride +
+        cache_block_size * head_dim +
         (block_offset * head_dim + head_dim_idx) * 4 / quant_block_size;
     reinterpret_cast<float*>(kv_cache)[dst_scale_idx / 4] = scale;
   }
@@ -1451,7 +1452,7 @@ void cp_gather_and_upconvert_fp8_kv_cache(
           reinterpret_cast<KV_T*>(k.data_ptr()),                              \
           reinterpret_cast<CACHE_T*>(kv_cache.data_ptr()),                    \
           slot_mapping.const_data_ptr<int64_t>(), head_dim, quant_block_size, \
-          cache_block_size, cache_block_stride, use_ue8m0);
+          cache_block_size, cache_stride, use_ue8m0);
 
 void indexer_k_quant_and_cache(
     torch::stable::Tensor& k,         // [num_tokens, head_dim]
@@ -1462,7 +1463,7 @@ void indexer_k_quant_and_cache(
   int num_tokens = k.size(0);
   int head_dim = k.size(1);
   int cache_block_size = kv_cache.size(1);
-  int64_t cache_block_stride = kv_cache.stride(0);
+  int cache_stride = kv_cache.size(2);
   bool use_ue8m0 = scale_fmt == "ue8m0";
 
   STD_TORCH_CHECK(k.device() == kv_cache.device(),

csrc/libtorch_stable/fused_minimax_m3_qknorm_rope_kv_insert_kernel.cu

@@ -67,6 +67,13 @@
   #include "../quantization/w8a8/fp8/nvidia/quant_utils.cuh"
 #endif
 
+// Direct float -> E4M3 FP8 conversion for the indexer Q / index-K outputs.
+#ifndef USE_ROCM
+  #include <cuda_fp8.h>
+#else
+  #include <hip/hip_fp8.h>
+#endif
+
 #ifndef FINAL_MASK
   #ifdef USE_ROCM
     #define FINAL_MASK 0xffffffffffffffffULL
@@ -75,6 +82,19 @@
   #endif
 #endif
 
+#ifdef USE_ROCM
+// ROCm-compatible direct float -> E4M3 FP8 conversion (mirrors the DeepSeek V4
+// fused kernel).
+__device__ __forceinline__ uint8_t rocm_cvt_float_to_fp8_e4m3(float val) {
+  #if defined(HIP_FP8_TYPE_OCP)
+  __hip_fp8_e4m3 fp8_val(val);
+  #else
+  __hip_fp8_e4m3_fnuz fp8_val(val);
+  #endif
+  return reinterpret_cast<uint8_t&>(fp8_val);
+}
+#endif
+
 namespace vllm {
 namespace minimax_m3_fused_ops {
 
@@ -193,6 +213,8 @@ __device__ __forceinline__ void storeElems(
   *reinterpret_cast<uint2*>(dst) = v;
 }
 
+// Main K/V cache store. kAuto = unquantized (cache_t == scalar_t); fp8 cache
+// dtypes use the scaled-convert path with identity scale.
 template <typename scalar_t, typename cache_t, Fp8KVCacheDataType kv_dt>
 __device__ __forceinline__ void storeCacheElems(
     cache_t* __restrict__ dst, float const (&elems)[kElemsPerLane]) {
@@ -208,6 +230,32 @@ __device__ __forceinline__ void storeCacheElems(
   }
 }
 
+// Store 4 fp32 registers -> 4 contiguous E4M3 FP8 bytes (direct cast,
+// saturating to ±448). Used for the fp8 indexer-Q / index-K outputs; no scale
+// (RMSNorm outputs are O(1) and the score path only needs relative block
+// ordering).
+__device__ __forceinline__ void storeElemsFp8(
+    uint8_t* __restrict__ dst, float const (&elems)[kElemsPerLane]) {
+  constexpr float kFp8Max = 448.0f;
+#ifndef USE_ROCM
+  __nv_fp8x2_storage_t out2[kElemsPerLane / 2];
+  #pragma unroll
+  for (int i = 0; i < kElemsPerLane / 2; i++) {
+    float2 vv = make_float2(elems[2 * i], elems[2 * i + 1]);
+    vv.x = fminf(fmaxf(vv.x, -kFp8Max), kFp8Max);
+    vv.y = fminf(fmaxf(vv.y, -kFp8Max), kFp8Max);
+    out2[i] = __nv_cvt_float2_to_fp8x2(vv, __NV_SATFINITE, __NV_E4M3);
+  }
+  *reinterpret_cast<uint32_t*>(dst) = *reinterpret_cast<uint32_t const*>(out2);
+#else
+  #pragma unroll
+  for (int i = 0; i < kElemsPerLane; i++) {
+    float vv = fminf(fmaxf(elems[i], -kFp8Max), kFp8Max);
+    dst[i] = rocm_cvt_float_to_fp8_e4m3(vv);
+  }
+#endif
+}
+
 // ────────────────────────────────────────────────────────────────────────────
 // Kernel
 // ────────────────────────────────────────────────────────────────────────────
@@ -224,12 +272,14 @@ __device__ __forceinline__ void storeCacheElems(
 //     V : nkv  only if kInsertKV        (V-cache insert; no warps in dense)
 //     IQ: niq  only if kIsSparse        (norm+RoPE)
 //     IK: 1    only if kIsSparse        (norm+RoPE; +index-cache insert)
+// cache_t/kv_dt: main attention KV-cache dtype (auto/fp8). out_idx_t/kFp8Idx:
+// indexer index-K cache + index-Q output dtype (scalar_t or e4m3 byte).
 template <typename scalar_t, typename cache_t, Fp8KVCacheDataType kv_dt,
-          bool kIsSparse, bool kInsertKV>
+          typename out_idx_t, bool kIsSparse, bool kInsertKV, bool kFp8Idx>
 __global__ void fusedMiniMaxM3QNormRopeKVInsertKernel(
     scalar_t* __restrict__ qkv,  // [N, qkv_row] in/out (packs index if sparse)
-    scalar_t* __restrict__ q_out,        // [N, nq*128] contiguous, or nullptr
-    scalar_t* __restrict__ index_q_out,  // [N, niq*128] contiguous, or nullptr
+    scalar_t* __restrict__ q_out,         // [N, nq*128] contiguous, or nullptr
+    out_idx_t* __restrict__ index_q_out,  // [N, niq*128]; scalar_t or e4m3 byte
     scalar_t const* __restrict__ q_norm_w,
     scalar_t const* __restrict__ k_norm_w,
     scalar_t const* __restrict__ iq_norm_w,
@@ -238,8 +288,8 @@ __global__ void fusedMiniMaxM3QNormRopeKVInsertKernel(
     int64_t const* __restrict__ positions,       // [N] i64
     int64_t const* __restrict__ slot_mapping,    // main K/V slots or nullptr
     int64_t const* __restrict__ index_slot_mapping,  // index K slots/nullptr
-    cache_t* __restrict__ kv_cache,      // [nb,2,bs,nkv,128] or nullptr
-    scalar_t* __restrict__ index_cache,  // [nb*bs, 128] or nullptr
+    cache_t* __restrict__ kv_cache,       // [nb,2,bs,nkv,128] or nullptr
+    out_idx_t* __restrict__ index_cache,  // [nb*bs, 128]; scalar_t or e4m3 byte
     float const eps, int const rotary_dim, int const num_tokens, int const nq,
     int const nkv, int const niq, int const block_size,
     // kv_cache strides (in elements) for logical shape [nb, 2, bs, nkv, 128].
@@ -334,9 +384,12 @@ __global__ void fusedMiniMaxM3QNormRopeKVInsertKernel(
       store_ptr = q_out + static_cast<int64_t>(tokenIdx) * nq * kHeadDim +
                   slot * kHeadDim;
     } else if (isIQ && index_q_out != nullptr) {
-      store_ptr = index_q_out +
-                  static_cast<int64_t>(tokenIdx) * niq * kHeadDim +
-                  (slot - iq_begin) * kHeadDim;
+      // bf16 index_q_out: gather here. fp8: written by the explicit fp8 store.
+      if constexpr (!kFp8Idx) {
+        store_ptr = index_q_out +
+                    static_cast<int64_t>(tokenIdx) * niq * kHeadDim +
+                    (slot - iq_begin) * kHeadDim;
+      }
     }
 
     // PDL: wait for the predecessor kernel (the qkv-projection GEMM that
@@ -356,7 +409,19 @@ __global__ void fusedMiniMaxM3QNormRopeKVInsertKernel(
       scalar_t const* cos_ptr = cos_sin_cache + pos * rotary_dim;
       normAndRope<scalar_t>(elems, laneId, eps, norm_w, do_rope, rotary_dim,
                             cos_ptr, /*apply_norm=*/norm_w != nullptr);
-      storeElems<scalar_t>(store_ptr + dim_base, elems);
+      if constexpr (kFp8Idx) {
+        // index_q is e4m3 bytes; Q/K (and in-place index_k) stay scalar_t.
+        if (isIQ && index_q_out != nullptr) {
+          storeElemsFp8(index_q_out +
+                            static_cast<int64_t>(tokenIdx) * niq * kHeadDim +
+                            (slot - iq_begin) * kHeadDim + dim_base,
+                        elems);
+        } else {
+          storeElems<scalar_t>(store_ptr + dim_base, elems);
+        }
+      } else {
+        storeElems<scalar_t>(store_ptr + dim_base, elems);
+      }
     }
 
     // ── Cache inserts (sparse serving only). ───────────────────────────────
@@ -367,8 +432,11 @@ __global__ void fusedMiniMaxM3QNormRopeKVInsertKernel(
                              : (isIK ? index_slot_mapping[tokenIdx] : -1);
       if (sm >= 0) {  // skip padded / unscheduled tokens
         if (isIK) {
-          scalar_t* dst = index_cache + sm * kHeadDim + dim_base;
-          storeElems<scalar_t>(dst, elems);
+          if constexpr (kFp8Idx) {
+            storeElemsFp8(index_cache + sm * kHeadDim + dim_base, elems);
+          } else {
+            storeElems<scalar_t>(index_cache + sm * kHeadDim + dim_base, elems);
+          }
         } else if (isK || isV) {
           // kv_cache logical shape [num_blocks, 2, block_size, nkv, head_dim].
           // Paging is logical (block = sm/block_size, token = sm%block_size);
@@ -398,19 +466,19 @@ __global__ void fusedMiniMaxM3QNormRopeKVInsertKernel(
 // Launch wrapper
 // ────────────────────────────────────────────────────────────────────────────
 template <typename scalar_t, typename cache_t, Fp8KVCacheDataType kv_dt>
-void launchFusedMiniMaxM3(scalar_t* qkv, scalar_t* q_out, scalar_t* index_q_out,
-                          scalar_t const* q_norm_w, scalar_t const* k_norm_w,
-                          scalar_t const* iq_norm_w, scalar_t const* ik_norm_w,
-                          scalar_t const* cos_sin_cache,
-                          int64_t const* positions, int64_t const* slot_mapping,
-                          int64_t const* index_slot_mapping, cache_t* kv_cache,
-                          scalar_t* index_cache, float const eps,
-                          int const rotary_dim, int const num_tokens,
-                          int const nq, int const nkv, int const niq,
-                          int const block_size, int64_t const kv_s_block,
-                          int64_t const kv_s_kv, int64_t const kv_s_token,
-                          int64_t const kv_s_head, bool const has_index,
-                          bool const insert_kv, cudaStream_t stream) {
+void launchFusedMiniMaxM3(
+    scalar_t* qkv, scalar_t* q_out, void* index_q_out, scalar_t const* q_norm_w,
+    scalar_t const* k_norm_w, scalar_t const* iq_norm_w,
+    scalar_t const* ik_norm_w, scalar_t const* cos_sin_cache,
+    int64_t const* positions, int64_t const* slot_mapping,
+    int64_t const* index_slot_mapping, cache_t* kv_cache, void* index_cache,
+    float const eps, int const rotary_dim, int const num_tokens, int const nq,
+    int const nkv, int const niq, int const block_size,
+    int64_t const kv_s_block, int64_t const kv_s_kv, int64_t const kv_s_token,
+    int64_t const kv_s_head, bool const has_index, bool const insert_kv,
+    bool const fp8_idx, cudaStream_t stream) {
+  // Index outputs are scalar_t (bf16) or e4m3 bytes (uint8_t); reinterpret the
+  // void* pointers per instantiation in the LAUNCH macro.
   // Slot count must match the kernel's compile-time gating.
   int const v_slots = insert_kv ? nkv : 0;
   int const idx_slots = has_index ? niq + 1 : 0;
@@ -440,25 +508,27 @@ void launchFusedMiniMaxM3(scalar_t* qkv, scalar_t* q_out, scalar_t* index_q_out,
   config.attrs = attrs;
   config.numAttrs = (sm_version >= 90) ? 1 : 0;
 
-  #define LAUNCH(IS_SPARSE, INSERT)                                           \
-    cudaLaunchKernelEx(                                                       \
-        &config,                                                              \
-        fusedMiniMaxM3QNormRopeKVInsertKernel<scalar_t, cache_t, kv_dt,       \
-                                              IS_SPARSE, INSERT>,             \
-        qkv, q_out, index_q_out, q_norm_w, k_norm_w, iq_norm_w, ik_norm_w,    \
-        cos_sin_cache, positions, slot_mapping, index_slot_mapping, kv_cache, \
-        index_cache, eps, rotary_dim, num_tokens, nq, nkv, niq, block_size,   \
-        kv_s_block, kv_s_kv, kv_s_token, kv_s_head)
+  #define LAUNCH(IS_SPARSE, INSERT, FP8, OUT_T)                                \
+    cudaLaunchKernelEx(                                                        \
+        &config,                                                               \
+        fusedMiniMaxM3QNormRopeKVInsertKernel<scalar_t, cache_t, kv_dt, OUT_T, \
+                                              IS_SPARSE, INSERT, FP8>,         \
+        qkv, q_out, reinterpret_cast<OUT_T*>(index_q_out), q_norm_w, k_norm_w, \
+        iq_norm_w, ik_norm_w, cos_sin_cache, positions, slot_mapping,          \
+        index_slot_mapping, kv_cache, reinterpret_cast<OUT_T*>(index_cache),   \
+        eps, rotary_dim, num_tokens, nq, nkv, niq, block_size, kv_s_block,     \
+        kv_s_kv, kv_s_token, kv_s_head)
 #else
   // ROCm: standard kernel launch syntax (no PDL/stream serialization).
   // clang-format off
-  #define LAUNCH(IS_SPARSE, INSERT)                                          \
-    fusedMiniMaxM3QNormRopeKVInsertKernel<scalar_t, cache_t, kv_dt,          \
-                                          IS_SPARSE, INSERT>                 \
+  #define LAUNCH(IS_SPARSE, INSERT, FP8, OUT_T)                              \
+    fusedMiniMaxM3QNormRopeKVInsertKernel<scalar_t, cache_t, kv_dt, OUT_T,   \
+                                          IS_SPARSE, INSERT, FP8>            \
         <<<grid, kBlockSize, 0, stream>>>(                                   \
-            qkv, q_out, index_q_out, q_norm_w, k_norm_w, iq_norm_w,          \
-            ik_norm_w, cos_sin_cache, positions, slot_mapping,               \
-            index_slot_mapping, kv_cache, index_cache, eps, rotary_dim,      \
+            qkv, q_out, reinterpret_cast<OUT_T*>(index_q_out), q_norm_w,     \
+            k_norm_w, iq_norm_w, ik_norm_w, cos_sin_cache, positions,        \
+            slot_mapping, index_slot_mapping, kv_cache,                      \
+            reinterpret_cast<OUT_T*>(index_cache), eps, rotary_dim,          \
             num_tokens, nq, nkv, niq, block_size, kv_s_block, kv_s_kv,       \
             kv_s_token, kv_s_head)
   // clang-format on
@@ -466,14 +536,22 @@ void launchFusedMiniMaxM3(scalar_t* qkv, scalar_t* q_out, scalar_t* index_q_out,
 
   if (has_index) {
     if (insert_kv) {
-      LAUNCH(true, true);  // sparse serving
+      if (fp8_idx) {
+        LAUNCH(true, true, true, uint8_t);  // sparse serving, fp8 index outputs
+      } else {
+        LAUNCH(true, true, false, scalar_t);  // sparse serving, bf16
+      }
     } else {
-      LAUNCH(true, false);  // sparse profiling
+      if (fp8_idx) {
+        LAUNCH(true, false, true, uint8_t);  // sparse profiling, fp8 index_q
+      } else {
+        LAUNCH(true, false, false, scalar_t);  // sparse profiling, bf16
+      }
     }
   } else {
     // Dense layer: never has an index branch and never inserts here (the
     // generic Attention layer owns the KV insert).
-    LAUNCH(false, false);
+    LAUNCH(false, false, false, scalar_t);
   }
 #undef LAUNCH
 }
@@ -485,8 +563,9 @@ void launchFusedMiniMaxM3(scalar_t* qkv, scalar_t* q_out, scalar_t* index_q_out,
   vllm::minimax_m3_fused_ops::launchFusedMiniMaxM3<st, CACHE_T, KV_DTYPE>(     \
       reinterpret_cast<st*>(qkv.data_ptr()),                                   \
       q_out.has_value() ? reinterpret_cast<st*>(q_out->data_ptr()) : nullptr,  \
-      index_q_out.has_value() ? reinterpret_cast<st*>(index_q_out->data_ptr()) \
-                              : nullptr,                                       \
+      index_q_out.has_value()                                                  \
+          ? reinterpret_cast<void*>(index_q_out->data_ptr())                   \
+          : nullptr,                                                           \
       reinterpret_cast<st const*>(q_norm_weight.data_ptr()),                   \
       reinterpret_cast<st const*>(k_norm_weight.data_ptr()),                   \
       has_index ? reinterpret_cast<st const*>(index_q_norm_weight->data_ptr()) \
@@ -502,11 +581,11 @@ void launchFusedMiniMaxM3(scalar_t* qkv, scalar_t* q_out, scalar_t* index_q_out,
                 : nullptr,                                                     \
       insert_kv ? reinterpret_cast<CACHE_T*>(kv_cache->data_ptr()) : nullptr,  \
       (insert_kv && has_index)                                                 \
-          ? reinterpret_cast<st*>(index_cache->data_ptr())                     \
+          ? reinterpret_cast<void*>(index_cache->data_ptr())                   \
           : nullptr,                                                           \
       static_cast<float>(eps), static_cast<int>(rotary_dim), num_tokens, nq,   \
       nkv, niq, static_cast<int>(block_size), kv_s_block, kv_s_kv, kv_s_token, \
-      kv_s_head, has_index, insert_kv, stream)
+      kv_s_head, has_index, insert_kv, fp8_idx, stream)
 
 // ────────────────────────────────────────────────────────────────────────────
 // Torch op wrapper
@@ -612,6 +691,7 @@ void fused_minimax_m3_qknorm_rope_kv_insert(
                  torch::headeronly::ScalarType::Long &&
              index_slot_mapping->numel() == slot_mapping->numel()),
         "index_slot_mapping must be int64 CUDA with slot_mapping length");
+    // Main attention KV cache: auto matches qkv, fp8 uses uint8 storage.
     if (kv_dt == vllm::Fp8KVCacheDataType::kAuto) {
       STD_TORCH_CHECK(kv_cache->scalar_type() == qkv.scalar_type(),
                       "auto kv_cache dtype must match qkv");
@@ -620,9 +700,13 @@ void fused_minimax_m3_qknorm_rope_kv_insert(
           kv_cache->scalar_type() == torch::headeronly::ScalarType::Byte,
           "fp8 kv_cache must use uint8 storage");
     }
-    STD_TORCH_CHECK(index_cache.has_value() &&
-                        index_cache->scalar_type() == qkv.scalar_type(),
-                    "insert mode requires matching index_cache");
+    // Indexer index-K cache: independent dtype -- qkv dtype or fp8 e4m3.
+    STD_TORCH_CHECK(
+        index_cache.has_value() &&
+            (index_cache->scalar_type() == qkv.scalar_type() ||
+             index_cache->scalar_type() ==
+                 torch::headeronly::ScalarType::Float8_e4m3fn),
+        "insert mode requires index_cache matching qkv dtype or fp8 e4m3");
     STD_TORCH_CHECK(kv_cache->dim() == 5 && kv_cache->stride(4) == 1,
                     "kv_cache must be [nb,2,bs,nkv,head_dim] with contiguous "
                     "head_dim (stride(4)==1)");
@@ -652,14 +736,31 @@ void fused_minimax_m3_qknorm_rope_kv_insert(
         "index_q_out requires the index branch (num_index_heads > 0)");
     STD_TORCH_CHECK(
         index_q_out->is_cuda() && index_q_out->is_contiguous() &&
-            index_q_out->scalar_type() == qkv.scalar_type(),
-        "index_q_out must be a contiguous CUDA tensor matching qkv dtype");
+            (index_q_out->scalar_type() == qkv.scalar_type() ||
+             index_q_out->scalar_type() ==
+                 torch::headeronly::ScalarType::Float8_e4m3fn),
+        "index_q_out must be contiguous CUDA, qkv dtype or fp8 e4m3");
     STD_TORCH_CHECK(index_q_out->numel() ==
                         static_cast<int64_t>(num_tokens) * niq * kHeadDim,
                     "index_q_out must have num_tokens * num_index_heads * 128 "
                     "elements");
   }
 
+  // fp8 index path: the index-K cache and index-Q outputs are e4m3 bytes while
+  // q/k/v + q_out stay qkv dtype. Both index outputs must agree.
+  auto const kFp8 = torch::headeronly::ScalarType::Float8_e4m3fn;
+  bool const fp8_idx =
+      (index_cache.has_value() && index_cache->scalar_type() == kFp8) ||
+      (index_q_out.has_value() && index_q_out->scalar_type() == kFp8);
+  if (fp8_idx) {
+    STD_TORCH_CHECK(
+        !index_cache.has_value() || index_cache->scalar_type() == kFp8,
+        "fp8 index path: index_cache must be fp8 e4m3");
+    STD_TORCH_CHECK(
+        !index_q_out.has_value() || index_q_out->scalar_type() == kFp8,
+        "fp8 index path: index_q_out must be fp8 e4m3");
+  }
+
   const torch::stable::accelerator::DeviceGuard device_guard(
       qkv.get_device_index());
   auto stream = get_current_cuda_stream(qkv.get_device_index());

docker/Dockerfile

@@ -792,7 +792,7 @@ RUN --mount=type=cache,target=/opt/uv/cache \
 # Install FlashInfer JIT cache (requires CUDA-version-specific index URL)
 # https://docs.flashinfer.ai/installation.html
 # From versions.json: .flashinfer.version
-ARG FLASHINFER_VERSION=0.6.12
+ARG FLASHINFER_VERSION=0.6.13rc2
 RUN --mount=type=cache,target=/opt/uv/cache \
     uv pip install --system flashinfer-jit-cache==${FLASHINFER_VERSION} \
         --index-url https://flashinfer.ai/whl/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.')

docker/Dockerfile.xpu

@@ -132,7 +132,7 @@ CMD ["/bin/bash"]
 FROM vllm-base AS ucx-nixl-build
 
 ARG UCX_VERSION=v1.21.0-rc2
-ARG NIXL_VERSION=v1.2.0
+ARG NIXL_VERSION=0.10.1
 
 # Build-time only: compiler, autotools, and verbs dev headers
 RUN apt-get update -y && apt-get install -y --no-install-recommends \
@@ -149,25 +149,25 @@ RUN apt-get update -y && apt-get install -y --no-install-recommends \
 # patchelf (installed via uv) is used by the NIXL wheel build to rewrite
 # RPATH entries, making the wheel portable across stages.
 RUN --mount=type=cache,target=/root/.cache/uv \
-    git clone --depth 1 --branch "${UCX_VERSION}" https://github.com/openucx/ucx /tmp/ucx_source && \
-    cd /tmp/ucx_source && \
+    git clone https://github.com/openucx/ucx /tmp/ucx_source && \
+    cd /tmp/ucx_source && git checkout "${UCX_VERSION}" && \
     bash autogen.sh && \
     ./configure --prefix=/tmp/ucx_install --with-ze=yes --enable-examples --enable-mt && \
-    make CFLAGS="-Wno-error=incompatible-pointer-types" -j"$(nproc)" && make install && \
-    git clone --depth 1 --branch "${NIXL_VERSION}" https://github.com/ai-dynamo/nixl /tmp/nixl_source && \
-    cd /tmp/nixl_source && \
+    make CFLAGS="-Wno-error=incompatible-pointer-types" -j8 && make install && \
+    git clone https://github.com/ai-dynamo/nixl /tmp/nixl_source && \
+    cd /tmp/nixl_source && git checkout "${NIXL_VERSION}" && \
     uv pip install --upgrade meson pybind11 patchelf && \
     uv pip install -r requirements.txt && \
     PKG_CONFIG_PATH=/tmp/ucx_install/lib/pkgconfig \
     LD_LIBRARY_PATH=/tmp/ucx_install/lib \
     python -m pip wheel --no-deps . -w /tmp/nixl_wheels/ && \
     find /tmp/ucx_install -type f \( -name '*.a' -o -name '*.la' \) -delete && \
-    rm -rf /tmp/ucx_install/{include,share,etc,bin} /tmp/ucx_install/lib/cmake \
-           /tmp/ucx_source /tmp/nixl_source
+    rm -rf /tmp/ucx_install/include /tmp/ucx_install/share /tmp/ucx_install/etc /tmp/ucx_install/lib/cmake /tmp/ucx_install/bin && \
+    rm -rf /tmp/ucx_source /tmp/nixl_source
 
 FROM vllm-base AS vllm-openai
 
-ARG NIXL_VERSION=v1.2.0
+ARG NIXL_VERSION=0.10.1
 
 # Copy compiled UCX runtime libraries and the pre-built NIXL wheel.
 # No compiler or autotools are installed in this stage.
@@ -192,7 +192,7 @@ RUN --mount=type=cache,target=/root/.cache/uv \
     librdmacm1t64 \
     && rm -rf /var/lib/apt/lists/* \
     && uv pip install --no-deps /tmp/nixl_wheels/nixl*.whl \
-    && uv pip install nixl==${NIXL_VERSION} && uv pip uninstall nixl-cu13 \
+    && uv pip install nixl==${NIXL_VERSION} \
     && rm -rf /tmp/nixl_wheels
 
 RUN --mount=type=cache,target=/root/.cache/uv \

docker/versions.json

@@ -68,7 +68,7 @@
       "default": "true"
     },
     "FLASHINFER_VERSION": {
-      "default": "0.6.12"
+      "default": "0.6.13rc2"
     },
     "GDRCOPY_CUDA_VERSION": {
       "default": "12.8"

docs/contributing/README.md

@@ -304,15 +304,9 @@ review process:
   resources. The reviewer will add `ready` label to the PR when the PR is
   ready to merge or a full CI run is needed.
 
-### Pull Request Limits and Escalation
+### Escalating Stalled Contributions
 
-vLLM uses GitHub's [pull request limit](https://github.blog/open-source/maintainers/how-pull-request-limits-are-cutting-down-the-noise/)
-for contributors without write access. The current cap is 6 open PRs. If this
-blocks well-intentioned critical work, contact a committer to request bypass
-list access.
-
-If you need an expedited review for an important contribution, please email us
-at:
+If you have an important contribution that has not yet received maintainer attention, please email us at:
 
 <pr-review-request@vllm.ai>
 

docs/contributing/model/basic.md

@@ -136,7 +136,7 @@ The model should also be added to the `MODELS_CONFIG_MAP` dictionary in [vllm/mo
 For case (2), we recommend using as a reference the implementation of [`JambaForCausalLM`](../../../vllm/model_executor/models/jamba.py) (for an example of a model that uses Mamba-1 and attention together) or [`NemotronHForCausalLM`](../../../vllm/model_executor/models/nemotron_h.py) (for an example of a model that uses Mamba-2 and attention together).
 These models should follow the same instructions as case (1), but they should inherit protocol `IsHybrid` (instead of `IsAttentionFree`) and it is *not* necessary to add them to the `MODELS_CONFIG_MAP` (their runtime defaults will be inferred from the protocol).
 
-For case (3), we recommend looking at the implementation of [`Lfm2ForCausalLM`](../../../vllm/model_executor/models/lfm2.py) as a reference, which uses a custom "mamba-like" layer `ShortConv`.
+For case (3), we recommend looking at the implementation of [`MiniMaxText01ForCausalLM`](../../../vllm/model_executor/models/minimax_text_01.py) or [`Lfm2ForCausalLM`](../../../vllm/model_executor/models/lfm2.py) as a reference, which use custom "mamba-like" layers `MiniMaxText01LinearAttention` and `ShortConv` respectively.
 Please follow the same guidelines as case (2) for implementing these models.
 We use "mamba-like" to refer to layers that possess a state that is updated in-place, rather than being appended-to (like KV cache for attention).
 For implementing new custom mamba-like layers, one should inherit from `MambaBase` and implement the methods `get_state_dtype`, `get_state_shape` to calculate the data types and state shapes at runtime, as well as `mamba_type` and `get_attn_backend`.
@@ -144,5 +144,5 @@ It is also necessary to implement the "attention meta-data" class which handles
 Please see [`LinearAttentionMetadata`](../../../vllm/v1/attention/backends/linear_attn.py) or [`ShortConvAttentionMetadata`](../../../vllm/v1/attention/backends/short_conv_attn.py) for examples of this.
 It is also worth noting that we should update `MambaAttentionBackendEnum` in [`registry.py`](../../../vllm/v1/attention/backends/registry.py) when adding a new mamba backend.
 Finally, if one wants to support torch compile and CUDA graphs, it necessary to wrap the call to the mamba-like layer inside a custom op and register it.
-Please see the calls to `direct_register_custom_op` in [vllm/model_executor/layers/mamba/linear/minimax_linear_attn.py](../../../vllm/model_executor/layers/mamba/linear/minimax_linear_attn.py) or [vllm/model_executor/layers/mamba/short_conv.py](../../../vllm/model_executor/layers/mamba/short_conv.py) for examples of this.
+Please see the calls to `direct_register_custom_op` in [vllm/model_executor/models/minimax_text_01.py](../../../vllm/model_executor/models/minimax_text_01.py) or [vllm/model_executor/layers/mamba/short_conv.py](../../../vllm/model_executor/layers/mamba/short_conv.py) for examples of this.
 The new custom op should then be added to the list `_attention_ops` in [vllm/config/compilation.py](../../../vllm/config/compilation.py) to ensure that piecewise CUDA graphs works as intended.

docs/design/attention_backends.md

@@ -170,8 +170,8 @@ Priority is **1 = highest** (tried first).
 | Backend | Version | Dtypes | KV Dtypes | Block Sizes | Head Sizes | Sink | Non-Causal | MM Prefix | DCP | Attention Types | Compute Cap. |
 | ------- | ------- | ------ | --------- | ----------- | ---------- | ---- | ---------- | --------- | --- | --------------- | ------------ |
 | `CPU_ATTN` | | fp16, bf16, fp32 | `auto`, `fp8`, `fp8_e4m3`, `fp8_e5m2` | %16 | 32, 64, 80, 96, 112, 128, 160, 192, 224, 256, 512 | ❌ | ❌ | ❌ | ❌ | All | N/A |
-| `FLASHINFER` | Native† | fp16, bf16 | `auto`, `float16`, `bfloat16`, `fp8`, `fp8_e4m3`, `fp8_e5m2` | 16, 32, 64, 128, 256, 512, 1024 | 64, 128, 256, 512 | ❌ | ✅ | ❌ | ✅ | Decoder | 7.x-9.x |
-| `FLASHINFER` | TRTLLM† | fp16, bf16 | `auto`, `float16`, `bfloat16`, `fp8`, `fp8_e4m3`, `fp8_e5m2`, `nvfp4` | 16, 32, 64, 128, 256, 512, 1024 | 64, 128, 256, 512 | ✅ | ✅ | ❌ | ✅ | Decoder | 10.x |
+| `FLASHINFER` | Native† | fp16, bf16 | `auto`, `float16`, `bfloat16`, `fp8`, `fp8_e4m3`, `fp8_e5m2` | 16, 32, 64, 128, 256, 512, 1024 | 64, 128, 256, 512 | ❌ | ❌ | ❌ | ✅ | Decoder | 7.x-9.x |
+| `FLASHINFER` | TRTLLM† | fp16, bf16 | `auto`, `float16`, `bfloat16`, `fp8`, `fp8_e4m3`, `fp8_e5m2`, `nvfp4` | 16, 32, 64, 128, 256, 512, 1024 | 64, 128, 256, 512 | ✅ | ❌ | ❌ | ✅ | Decoder | 10.x |
 | `FLASH_ATTN` | FA2* | fp16, bf16 | `auto`, `float16`, `bfloat16` | %16 | Any | ❌ | ✅ | ❌ | ✅ | All | ≥8.0 |
 | `FLASH_ATTN` | FA3* | fp16, bf16 | `auto`, `float16`, `bfloat16`, `fp8`, `fp8_e4m3`, `fp8_e5m2` | %16 | Any | ✅ | ✅ | ❌ | ✅ | All | 9.x |
 | `FLASH_ATTN` | FA4* | fp16, bf16 | `auto`, `float16`, `bfloat16` | %16 | Any | ✅ | ✅ | ❌ | ✅ | All | ≥10.0 |

docs/features/kv_offloading_usage.md

@@ -74,7 +74,6 @@ vllm serve <model> \
 | `max_tracker_size` | no | `64000` | single-tier | Max entries in the lookup tracker. |
 | `secondary_tiers` | no | `[]` | multi-tier | List of secondary tier configs (see below). |
 | `offload_prompt_only` | no | `true` | both | If `true`, only prompt (prefill) blocks are offloaded; decode blocks are skipped. |
-| `self_describing_kv_events` | no | `false` | single-tier | Opt-in. When `true` *and* KV cache events are enabled (`--kv-events-config` with `enable_kv_cache_events`), the connector emits self-describing block-granular `BlockStored`/`BlockRemoved` payloads (constituent block hashes, whole-chunk `token_ids`, per-block `block_size`, parent hash, LoRA + group/cache-spec metadata) instead of the placeholder fallback, so external KV-event consumers can index offloaded blocks. Inert unless events are enabled. Currently rejected by `TieringOffloadingSpec`. Full-attention groups only; sliding-window/SSM groups keep the placeholder fallback. In chunk mode (`block_size` > GPU block size), overlapping chunks re-announce shared per-block hashes, so consumers must reference-count (deduplicate) repeated store/remove announcements. |
 | `spec_module_path` | no | — | both | Python import path for a custom `OffloadingSpec` not in the built-in registry. Required only when `spec_name` is not built-in (advanced). |
 
 ## Secondary Tiers

docs/features/tool_calling.md

@@ -321,6 +321,15 @@ For Qwen2.5, the chat template in tokenizer_config.json has already included sup
 
 Flags: `--tool-call-parser hermes`
 
+### MiniMax Models (`minimax_m1`)
+
+Supported models:
+
+* `MiniMaxAi/MiniMax-M1-40k` (use with [examples/tool_chat_template_minimax_m1.jinja](../../examples/tool_chat_template_minimax_m1.jinja))
+* `MiniMaxAi/MiniMax-M1-80k` (use with [examples/tool_chat_template_minimax_m1.jinja](../../examples/tool_chat_template_minimax_m1.jinja))
+
+Flags: `--tool-call-parser minimax --chat-template examples/tool_chat_template_minimax_m1.jinja`
+
 ### DeepSeek-V3 Models (`deepseek_v3`)
 
 Supported models:

docs/models/pooling_models/token_embed.md

@@ -61,7 +61,7 @@ Models of any architecture can be converted into embedding models using `--conve
 | `ColModernVBertForRetrieval` | ColModernVBERT | T / I | `ModernVBERT/colmodernvbert-merged` | | |
 | `ColPaliForRetrieval` | ColPali | T / I | `vidore/colpali-v1.3-hf` | | |
 | `ColQwen3` | Qwen3-VL | T / I | `TomoroAI/tomoro-colqwen3-embed-4b`, `TomoroAI/tomoro-colqwen3-embed-8b` | | |
-| `ColQwen3_5` | ColQwen3.5 | T + I + V | `athrael-soju/colqwen3.5-4.5B-v3`, `vultr/VultronRetrieverPrime-Qwen3.5-8B` | | |
+| `ColQwen3_5` | ColQwen3.5 | T + I + V | `athrael-soju/colqwen3.5-4.5B-v3` | | |
 | `OpsColQwen3Model` | Qwen3-VL | T / I | `OpenSearch-AI/Ops-Colqwen3-4B`, `OpenSearch-AI/Ops-Colqwen3-8B` | | |
 | `Qwen3VLNemotronEmbedModel` | Qwen3-VL | T / I | `nvidia/nemotron-colembed-vl-4b-v2`, `nvidia/nemotron-colembed-vl-8b-v2` | ✅︎ | ✅︎ |
 | `*ForConditionalGeneration`<sup>C</sup>, `*ForCausalLM`<sup>C</sup>, etc. | Generative models | \* | N/A | \* | \* |

docs/models/supported_models.md

@@ -441,6 +441,7 @@ th {
 | `MiMoV2ForCausalLM` | MiMoV2Pro | `XiaomiMiMo/MiMo-V2.5-Pro`, etc. | | ✅︎ |
 | `MiniCPMForCausalLM` | MiniCPM | `openbmb/MiniCPM-2B-sft-bf16`, `openbmb/MiniCPM-2B-dpo-bf16`, `openbmb/MiniCPM-S-1B-sft`, etc. | ✅︎ | ✅︎ |
 | `MiniCPM3ForCausalLM` | MiniCPM3 | `openbmb/MiniCPM3-4B`, etc. | ✅︎ | ✅︎ |
+| `MiniMaxForCausalLM` | MiniMax-Text | `MiniMaxAI/MiniMax-Text-01-hf`, etc. | | |
 | `MiniMaxM2ForCausalLM` | MiniMax-M2, MiniMax-M2.1 | `MiniMaxAI/MiniMax-M2`, etc. | ✅︎ | ✅︎ |
 | `MistralForCausalLM` | Ministral-3, Mistral, Mistral-Instruct | `mistralai/Ministral-3-3B-Instruct-2512`, `mistralai/Mistral-7B-v0.1`, `mistralai/Mistral-7B-Instruct-v0.1`, etc. | ✅︎ | ✅︎ |
 | `MistralLarge3ForCausalLM` | Mistral-Large-3-675B-Base-2512, Mistral-Large-3-675B-Instruct-2512 | `mistralai/Mistral-Large-3-675B-Base-2512`, `mistralai/Mistral-Large-3-675B-Instruct-2512`, etc. | ✅︎ | ✅︎ |
@@ -486,6 +487,8 @@ th {
 | `TeleChat2ForCausalLM` | TeleChat2 | `Tele-AI/TeleChat2-3B`, `Tele-AI/TeleChat2-7B`, `Tele-AI/TeleChat2-35B`, etc. | ✅︎ | ✅︎ |
 | `TeleChat3ForCausalLM` | TeleChat3 | `Tele-AI/TeleChat3-36B-Thinking`, `Tele-AI/TeleChat3-Coder-36B-Thinking`, etc. | ✅︎ | ✅︎ |
 | `TeleFLMForCausalLM` | TeleFLM | `CofeAI/FLM-2-52B-Instruct-2407`, `CofeAI/Tele-FLM`, etc. | ✅︎ | ✅︎ |
+| `MiniMaxM1ForCausalLM` | MiniMax-Text | `MiniMaxAI/MiniMax-M1-40k`, `MiniMaxAI/MiniMax-M1-80k`, etc. | | |
+| `MiniMaxText01ForCausalLM` | MiniMax-Text | `MiniMaxAI/MiniMax-Text-01`, etc. | | |
 | `Zamba2ForCausalLM` | Zamba2 | `Zyphra/Zamba2-7B-instruct`, `Zyphra/Zamba2-2.7B-instruct`, `Zyphra/Zamba2-1.2B-instruct`, etc. | | |
 
 !!! note
@@ -592,7 +595,6 @@ These models primarily accept the [`LLM.generate`](./generative_models.md#llmgen
 | `MiMoV2OmniForCausalLM` | MiMo-V2.5-Omni | T + I<sup>E+</sup> + V<sup>E+</sup> + A<sup>+</sup> | `XiaomiMiMo/MiMo-V2.5-Omni` | | ✅︎ |
 | `MiniCPMO` | MiniCPM-O | T + I<sup>E+</sup> + V<sup>E+</sup> + A<sup>E+</sup> | `openbmb/MiniCPM-o-2_6`, etc. | ✅︎ | ✅︎ |
 | `MiniCPMV` | MiniCPM-V | T + I<sup>E+</sup> + V<sup>E+</sup> | `openbmb/MiniCPM-V-2` (see note), `openbmb/MiniCPM-Llama3-V-2_5`, `openbmb/MiniCPM-V-2_6`, `openbmb/MiniCPM-V-4`, `openbmb/MiniCPM-V-4_5`, etc. | ✅︎ | |
-| `MiniMaxM3SparseForConditionalGeneration` | MiniMax-M3 | T + I<sup>+</sup> + V<sup>+</sup> | `MiniMaxAI/MiniMax-M3`, `MiniMaxAI/MiniMax-M3-MXFP8`, etc. | | |
 | `MiniMaxVL01ForConditionalGeneration` | MiniMax-VL | T + I<sup>E+</sup> | `MiniMaxAI/MiniMax-VL-01`, etc. | | ✅︎ |
 | `Mistral3ForConditionalGeneration` | Mistral3 (HF Transformers) | T + I<sup>+</sup> | `mistralai/Mistral-Small-3.1-24B-Instruct-2503`, etc. | ✅︎ | ✅︎ |
 | `MolmoForCausalLM` | Molmo | T + I<sup>+</sup> | `allenai/Molmo-7B-D-0924`, `allenai/Molmo-7B-O-0924`, etc. | ✅︎ | ✅︎ |

docs/usage/v1_guide.md

@@ -128,7 +128,7 @@ Models that use Mamba-2 and Mamba-1 layers (e.g., `Mamba2ForCausalLM`, `MambaFor
 Hybrid models that combine Mamba-2 and Mamba-1 layers with standard attention layers are also supported (e.g., `BambaForCausalLM`,
 `Zamba2ForCausalLM`, `NemotronHForCausalLM`, `FalconH1ForCausalLM` and `GraniteMoeHybridForCausalLM`, `JambaForCausalLM`, `Plamo2ForCausalLM`).
 
-Hybrid models with mechanisms different to Mamba are also supported (e.g, `Lfm2ForCausalLM`).
+Hybrid models with mechanisms different to Mamba are also supported (e.g, `MiniMaxText01ForCausalLM`, `MiniMaxM1ForCausalLM`, `Lfm2ForCausalLM`).
 
 Please note that prefix caching is not yet supported for any of the above models.
 

examples/generate/multimodal/vision_language_offline.py

@@ -1481,6 +1481,39 @@ def run_minicpmv(questions: list[str], modality: str) -> ModelRequestData:
     return run_minicpmv_base(questions, modality, "openbmb/MiniCPM-V-2_6")
 
 
+def run_minimax_vl_01(questions: list[str], modality: str) -> ModelRequestData:
+    assert modality == "image"
+
+    model_name = "MiniMaxAI/MiniMax-VL-01"
+
+    engine_args = EngineArgs(
+        model=model_name,
+        max_num_seqs=2,
+        limit_mm_per_prompt={modality: 1},
+        trust_remote_code=True,
+        tensor_parallel_size=8,
+    )
+
+    tokenizer = AutoTokenizer.from_pretrained(model_name)
+    messages = [
+        [
+            {
+                "role": "user",
+                "content": [{"type": "image"}, {"type": "text", "text": question}],
+            }
+        ]
+        for question in questions
+    ]
+    prompts = tokenizer.apply_chat_template(
+        messages, add_generation_prompt=True, tokenize=False
+    )
+
+    return ModelRequestData(
+        engine_args=engine_args,
+        prompts=prompts,
+    )
+
+
 # Mistral-3 HF-format
 def run_mistral3(questions: list[str], modality: str) -> ModelRequestData:
     assert modality == "image"
@@ -2452,6 +2485,7 @@ model_example_map = {
     "mantis": run_mantis,
     "minicpmo": run_minicpmo,
     "minicpmv": run_minicpmv,
+    "minimax_vl_01": run_minimax_vl_01,
     "mistral3": run_mistral3,
     "molmo": run_molmo,
     "molmo2": run_molmo2,

examples/pooling/score/colqwen3_5_rerank_online.py

@@ -7,27 +7,11 @@ ColQwen3.5 is a multi-modal ColBERT-style model based on Qwen3.5.
 It produces per-token embeddings and uses MaxSim scoring for retrieval
 and reranking. Supports both text and image inputs.
 
-Works for any ColQwen3.5 checkpoint, e.g. `athrael-soju/colqwen3.5-4.5B-v3`
-or `vultr/VultronRetrieverPrime-Qwen3.5-8B`.
-
 Start the server with:
-    vllm serve athrael-soju/colqwen3.5-4.5B-v3 --max-model-len 4096 \
-        --mm-processor-kwargs '{"min_pixels": 65536, "max_pixels": 1835008}'
+    vllm serve athrael-soju/colqwen3.5-4.5B --max-model-len 4096
 
 Then run this script:
     python colqwen3_5_rerank_online.py
-
-Parity note (matching the native colpali ColQwen3_5Processor pipeline):
-  - Visual-token budget: ColQwen3_5Processor uses max_num_visual_tokens=1792,
-    i.e. max_pixels = 1792 * (patch_size*merge_size)^2 = 1792 * 32^2 = 1835008
-    (with min_pixels = shortest_edge = 65536). Pass these via --mm-processor-kwargs
-    as above; the default budget gives fewer visual tokens and lower retrieval ndcg.
-  - When you build prompts yourself (token_embed), reproduce the processor exactly:
-      image (document): wrap in the instruction template
-        "<|im_start|>user\n<|vision_start|><|image_pad|><|vision_end|>"
-        "Describe the image.<|im_end|><|endoftext|>"
-      query: append the augmentation suffix  <text> + "<|endoftext|>" * 10
-    Omitting these reproduces a silent ~2.5 ndcg@10 drop vs the native pipeline.
 """
 
 import requests

examples/tool_chat_template_minimax_m1.jinja

@@ -0,0 +1,91 @@
+{{ '<begin_of_document>' -}}
+{%- if custom_tools is defined %}
+    {%- set tools = custom_tools %}
+{%- endif %}
+{%- if not tools is defined %}
+    {%- set tools = none %}
+{%- endif %}
+
+{#- Extract system message #}
+{% set ns = namespace(system_prompt='') -%}
+{%- if messages[0]['role'] == 'system' %}
+    {%- if messages[0]['content'] is string %}
+        {%- set ns.system_prompt = messages[0]['content']|trim %}
+    {%- else %}
+        {%- set ns.system_prompt = messages[0]['content'][0]['text']|trim %}
+    {%- endif %}
+    {%- set messages = messages[1:] %}
+{%- else %}
+    {%- if tools is not none %}
+        {%- set ns.system_prompt = "You are a helpful assistant created by Minimax based on MiniMax-M1 model." %}
+    {%- else %}
+        {%- set ns.system_prompt = "You are a helpful assistant created by Minimax based on MiniMax-M1 model." %}
+    {%- endif %}
+{%- endif %}
+
+{#- System message #}
+{%- if ns.system_prompt != '' %}
+{{ '<beginning_of_sentence>system ai_setting=assistant\n' + ns.system_prompt + '<end_of_sentence>\n' -}}
+{%- endif %}
+
+{#- Tools configuration #}
+{%- if tools is not none %}
+{{ '<beginning_of_sentence>system tool_setting=tools\nYou are provided with these tools:\n<tools>\n' -}}
+{%- for tool in tools %}
+{{ tool | tojson ~ '\n' -}}
+{%- endfor %}
+{{ '</tools>\n\nIf you need to call tools, please respond with <tool_calls></tool_calls> XML tags, and provide tool-name and json-object of arguments, following the format below:\n<tool_calls>\n{"name": <tool-name>, "arguments": <args-json-object>}\n...\n</tool_calls><end_of_sentence>\n' -}}
+{%- endif %}
+
+{#- Process messages #}
+{%- for message in messages %}
+    {%- if not (message.role == 'ipython' or message.role == 'tool' or 'tool_calls' in message) %}
+        {%- if message['role'] == 'user' %}
+{{ '<beginning_of_sentence>user name=user\n' -}}
+{%- if message['content'] is string %}
+{{ message['content']|trim -}}
+{%- else %}
+{%- for content in message['content'] %}
+{%- if content['type'] == 'text' %}
+{{ content['text']|trim -}}
+{%- endif %}
+{%- endfor %}
+{%- endif %}
+{{ '<end_of_sentence>\n' -}}
+        {%- elif message['role'] == 'assistant' %}
+{{ '<beginning_of_sentence>ai name=assistant\n' -}}
+{%- if message['content'] is string %}
+{{ message['content']|trim -}}
+{%- else %}
+{%- for content in message['content'] | selectattr('type', 'equalto', 'text') %}
+{{ content['text']|trim -}}
+{%- endfor %}
+{%- endif %}
+{{ '<end_of_sentence>\n' -}}
+        {%- endif %}
+    {%- elif 'tool_calls' in message %}
+{{ '<beginning_of_sentence>ai name=assistant\n<tool_calls>\n' -}}
+{%- for tool_call in message.tool_calls %}
+{{ '{"name": "' + tool_call.function.name + '", "arguments": ' + tool_call.function.arguments | tojson + '}\n' -}}
+{%- endfor %}
+{{ '</tool_calls><end_of_sentence>\n' -}}
+    {%- elif message.role == "tool" or message.role == "ipython" %}
+{{ '<beginning_of_sentence>tool name=tools\n' -}}
+{%- if message.content is string %}
+{{ 'tool result: ' + message.content + '\n\n' -}}
+{%- else %}
+{%- for content in message['content'] %}
+{%- if content['type'] == 'text' %}
+{{ 'tool result: ' + content['text'] + '\n\n' -}}
+{%- elif content.get('name') %}
+{{ 'tool name: ' + content['name'] + '\ntool result: ' + content['text'] + '\n\n' -}}
+{%- endif %}
+{%- endfor %}
+{%- endif %}
+{{ '<end_of_sentence>\n' -}}
+    {%- endif %}
+{%- endfor %}
+
+{%- if add_generation_prompt %}
+{{ '<beginning_of_sentence>ai name=assistant\n' -}}
+{%- endif %}

requirements/cuda.txt

@@ -9,8 +9,8 @@ torchaudio==2.11.0
 # These must be updated alongside torch
 torchvision==0.26.0 # Required for phi3v processor. See https://github.com/pytorch/vision?tab=readme-ov-file#installation for corresponding version
 # FlashInfer should be updated together with the Dockerfile
-flashinfer-python==0.6.12
-flashinfer-cubin==0.6.12
+flashinfer-python==0.6.13rc2
+flashinfer-cubin==0.6.13rc2
 apache-tvm-ffi==0.1.9
 tilelang==0.1.9
 nvidia-cudnn-frontend>=1.19.1

rust/src/chat/src/renderer/hf/format.rs

@@ -386,6 +386,7 @@ mod tests {
             tool_chat_template_llama3.2_pythonic.jinja         => String
             tool_chat_template_llama4_json.jinja               => OpenAi
             tool_chat_template_llama4_pythonic.jinja           => OpenAi
+            tool_chat_template_minimax_m1.jinja                => OpenAi
             tool_chat_template_mistral.jinja                   => String
             tool_chat_template_mistral3.jinja                  => OpenAi
             tool_chat_template_mistral_parallel.jinja          => String

rust/src/chat/tests/templates/vllm_examples/tool_chat_template_minimax_m1.jinja

@@ -0,0 +1,91 @@
+{{ '<begin_of_document>' -}}
+{%- if custom_tools is defined %}
+    {%- set tools = custom_tools %}
+{%- endif %}
+{%- if not tools is defined %}
+    {%- set tools = none %}
+{%- endif %}
+
+{#- Extract system message #}
+{% set ns = namespace(system_prompt='') -%}
+{%- if messages[0]['role'] == 'system' %}
+    {%- if messages[0]['content'] is string %}
+        {%- set ns.system_prompt = messages[0]['content']|trim %}
+    {%- else %}
+        {%- set ns.system_prompt = messages[0]['content'][0]['text']|trim %}
+    {%- endif %}
+    {%- set messages = messages[1:] %}
+{%- else %}
+    {%- if tools is not none %}
+        {%- set ns.system_prompt = "You are a helpful assistant created by Minimax based on MiniMax-M1 model." %}
+    {%- else %}
+        {%- set ns.system_prompt = "You are a helpful assistant created by Minimax based on MiniMax-M1 model." %}
+    {%- endif %}
+{%- endif %}
+
+{#- System message #}
+{%- if ns.system_prompt != '' %}
+{{ '<beginning_of_sentence>system ai_setting=assistant\n' + ns.system_prompt + '<end_of_sentence>\n' -}}
+{%- endif %}
+
+{#- Tools configuration #}
+{%- if tools is not none %}
+{{ '<beginning_of_sentence>system tool_setting=tools\nYou are provided with these tools:\n<tools>\n' -}}
+{%- for tool in tools %}
+{{ tool | tojson ~ '\n' -}}
+{%- endfor %}
+{{ '</tools>\n\nIf you need to call tools, please respond with <tool_calls></tool_calls> XML tags, and provide tool-name and json-object of arguments, following the format below:\n<tool_calls>\n{"name": <tool-name>, "arguments": <args-json-object>}\n...\n</tool_calls><end_of_sentence>\n' -}}
+{%- endif %}
+
+{#- Process messages #}
+{%- for message in messages %}
+    {%- if not (message.role == 'ipython' or message.role == 'tool' or 'tool_calls' in message) %}
+        {%- if message['role'] == 'user' %}
+{{ '<beginning_of_sentence>user name=user\n' -}}
+{%- if message['content'] is string %}
+{{ message['content']|trim -}}
+{%- else %}
+{%- for content in message['content'] %}
+{%- if content['type'] == 'text' %}
+{{ content['text']|trim -}}
+{%- endif %}
+{%- endfor %}
+{%- endif %}
+{{ '<end_of_sentence>\n' -}}
+        {%- elif message['role'] == 'assistant' %}
+{{ '<beginning_of_sentence>ai name=assistant\n' -}}
+{%- if message['content'] is string %}
+{{ message['content']|trim -}}
+{%- else %}
+{%- for content in message['content'] | selectattr('type', 'equalto', 'text') %}
+{{ content['text']|trim -}}
+{%- endfor %}
+{%- endif %}
+{{ '<end_of_sentence>\n' -}}
+        {%- endif %}
+    {%- elif 'tool_calls' in message %}
+{{ '<beginning_of_sentence>ai name=assistant\n<tool_calls>\n' -}}
+{%- for tool_call in message.tool_calls %}
+{{ '{"name": "' + tool_call.function.name + '", "arguments": ' + tool_call.function.arguments | tojson + '}\n' -}}
+{%- endfor %}
+{{ '</tool_calls><end_of_sentence>\n' -}}
+    {%- elif message.role == "tool" or message.role == "ipython" %}
+{{ '<beginning_of_sentence>tool name=tools\n' -}}
+{%- if message.content is string %}
+{{ 'tool result: ' + message.content + '\n\n' -}}
+{%- else %}
+{%- for content in message['content'] %}
+{%- if content['type'] == 'text' %}
+{{ 'tool result: ' + content['text'] + '\n\n' -}}
+{%- elif content.get('name') %}
+{{ 'tool name: ' + content['name'] + '\ntool result: ' + content['text'] + '\n\n' -}}
+{%- endif %}
+{%- endfor %}
+{%- endif %}
+{{ '<end_of_sentence>\n' -}}
+    {%- endif %}
+{%- endfor %}
+
+{%- if add_generation_prompt %}
+{{ '<beginning_of_sentence>ai name=assistant\n' -}}
+{%- endif %}

rust/src/engine-core-client/src/protocol/mod.rs

@@ -277,12 +277,6 @@ pub struct EngineCoreSamplingParams {
     pub max_tokens: u32,
     /// Minimum number of tokens to generate before EOS or stop-token handling.
     pub min_tokens: u32,
-    /// Maximum number of reasoning ("thinking") tokens to emit before the
-    /// reasoning section is force-closed. `None` means unlimited; the
-    /// user-facing `-1` sentinel is normalized to `None` by the frontend before
-    /// reaching this DTO, so only non-negative values are sent. Enforced
-    /// engine-side (and only when a reasoning parser is configured).
-    pub thinking_token_budget: Option<u64>,
     /// Number of log probabilities to return per generated token.
     ///
     /// `None` disables sample logprobs. `-1` requests the full vocabulary.
@@ -351,7 +345,6 @@ impl EngineCoreSamplingParams {
             seed: None,
             max_tokens: 65536,
             min_tokens: 0,
-            thinking_token_budget: None,
             logprobs: None,
             prompt_logprobs: None,
             min_p: 0.0,

rust/src/engine-core-client/src/tests/client.rs

@@ -150,7 +150,6 @@ fn sample_request_with_id(request_id: &str) -> EngineCoreRequest {
             top_k: 8,
             max_tokens: 32,
             min_tokens: 1,
-            thinking_token_budget: Some(256),
             stop_token_ids: vec![151643],
             eos_token_id: Some(151645),
             all_stop_token_ids: BTreeSet::from([151643, 151645]),
@@ -2503,7 +2502,6 @@ fn python_msgpack_fixtures_match_rust_encoding() {
             seed: None,
             max_tokens: 16,
             min_tokens: 0,
-            thinking_token_budget: None,
             logprobs: None,
             prompt_logprobs: None,
             min_p: 0.0,

rust/src/engine-core-client/src/tests/python_compat.py

@@ -39,7 +39,6 @@ class EngineCoreSamplingParams(msgspec.Struct, dict=True, omit_defaults=True):
     seed: int | None = None
     max_tokens: int = 16
     min_tokens: int = 0
-    thinking_token_budget: int | None = None
     min_p: float = 0.0
     frequency_penalty: float = 0.0
     presence_penalty: float = 0.0
@@ -123,7 +122,6 @@ request = EngineCoreRequest(
         seed=None,
         max_tokens=32,
         min_tokens=1,
-        thinking_token_budget=256,
         min_p=0.0,
         frequency_penalty=0.0,
         presence_penalty=0.0,

rust/src/server/src/error.rs

@@ -103,7 +103,6 @@ fn is_request_validation_error(error: &vllm_text::Error) -> bool {
             | vllm_text::Error::EmptyPromptTokenIds { .. }
             | vllm_text::Error::Logprobs(_)
             | vllm_text::Error::OutOfVocab(_)
-            | vllm_text::Error::InvalidThinkingTokenBudget
             // An empty tokenized prompt detected later, at request prepare
             // time, surfaces through the transparent Llm wrapper.
             | vllm_text::Error::Llm(vllm_llm::Error::EmptyPromptTokenIds { .. })
@@ -128,18 +127,6 @@ mod tests {
         assert!(response.error.message.contains("9000"));
     }
 
-    #[test]
-    fn invalid_thinking_token_budget_maps_to_invalid_request() {
-        let api_error = text_submit_error(
-            "failed to submit completion request",
-            vllm_text::Error::InvalidThinkingTokenBudget,
-        );
-        assert_eq!(api_error.status_code(), StatusCode::BAD_REQUEST);
-        let response = api_error.to_error_response();
-        assert_eq!(response.error.error_type, "invalid_request_error");
-        assert!(response.error.message.contains("thinking_token_budget"));
-    }
-
     #[test]
     fn chat_wrapped_prompt_too_long_maps_to_invalid_request() {
         let error = vllm_chat::Error::Text(vllm_text::Error::PromptTooLong {

rust/src/server/src/routes/inference/generate/convert.rs

@@ -150,33 +150,6 @@ mod tests {
         );
     }
 
-    #[test]
-    fn prepare_generate_request_forwards_thinking_token_budget() {
-        let request: GenerateRequest = serde_json::from_value(json!({
-            "model": "Qwen/Qwen1.5-0.5B-Chat",
-            "token_ids": [11, 22, 33],
-            "sampling_params": {
-                "thinking_token_budget": 64
-            }
-        }))
-        .expect("parse request");
-
-        let prepared = prepare_generate_request(
-            request,
-            &served(&["Qwen/Qwen1.5-0.5B-Chat"]),
-            ResolvedRequestContext::default(),
-        )
-        .expect("prepare");
-
-        // The raw inference route shares `vllm_text::SamplingParams`, so the
-        // field is carried through to lowering exactly like the OpenAI routes
-        // (normalization/validation then happens in `lower_sampling_params`).
-        assert_eq!(
-            prepared.text_request.sampling_params.thinking_token_budget,
-            Some(64)
-        );
-    }
-
     #[test]
     fn prepare_generate_request_gates_continuous_usage_on_include_usage() {
         let request: GenerateRequest = serde_json::from_value(json!({

rust/src/server/src/routes/openai/chat_completions/convert.rs

@@ -115,7 +115,6 @@ pub(super) fn prepare_chat_request(
             seed: request.seed,
             max_tokens: request.max_completion_tokens,
             min_tokens: request.min_tokens,
-            thinking_token_budget: request.thinking_token_budget,
             logprobs: request.logprobs.then_some(top_logprobs),
             prompt_logprobs,
             min_p: request.min_p,
@@ -614,31 +613,6 @@ mod tests {
         assert_eq!(prepared.chat_request.sampling_params, expected);
     }
 
-    #[test]
-    fn prepare_chat_request_passes_through_thinking_token_budget() {
-        let prepare = |budget: Option<i64>| {
-            prepare_chat_request(
-                ChatCompletionRequest {
-                    thinking_token_budget: budget,
-                    ..base_request()
-                },
-                &served(&["Qwen/Qwen1.5-0.5B-Chat"]),
-                ResolvedRequestContext::default(),
-            )
-            .expect("request is valid")
-            .chat_request
-            .sampling_params
-            .thinking_token_budget
-        };
-
-        // The convert layer forwards the raw value verbatim (including the `-1`
-        // "unlimited" sentinel); normalization/validation happens during
-        // lowering (see `vllm_text::lower`).
-        assert_eq!(prepare(Some(64)), Some(64));
-        assert_eq!(prepare(Some(-1)), Some(-1));
-        assert_eq!(prepare(None), None);
-    }
-
     #[test]
     fn prepare_chat_request_accepts_developer_messages() {
         let request = ChatCompletionRequest {

rust/src/server/src/routes/openai/chat_completions/types.rs

@@ -165,10 +165,8 @@ pub struct ChatCompletionRequest {
     pub bad_words: Option<Vec<String>>,
 
     // -------- Extra vLLM Parameters --------
-    /// Token budget for reasoning/thinking. Accepts a non-negative integer, or
-    /// `-1` for unlimited (mirroring the Python frontend, which normalizes `-1`
-    /// to "no budget").
-    pub thinking_token_budget: Option<i64>,
+    /// Token budget for reasoning/thinking
+    pub thinking_token_budget: Option<u32>,
 
     /// Whether to include reasoning content in the response
     #[serde(default = "default_true")]

rust/src/server/src/routes/openai/chat_completions/validate.rs

@@ -108,6 +108,11 @@ pub(super) fn validate_request_compat(
         "truncate_prompt_tokens",
         "truncate_prompt_tokens is not supported.",
     )?;
+    reject_non_default(
+        request.thinking_token_budget.as_ref(),
+        "thinking_token_budget",
+        "thinking_token_budget is not supported.",
+    )?;
     reject_non_default(
         request.media_io_kwargs.as_ref(),
         "media_io_kwargs",

rust/src/server/src/routes/openai/completions/convert.rs

@@ -108,7 +108,6 @@ pub(super) fn prepare_completion_request(
             seed: request.seed,
             max_tokens,
             min_tokens: request.min_tokens,
-            thinking_token_budget: request.thinking_token_budget,
             logprobs,
             prompt_logprobs,
             min_p: request.min_p,
@@ -267,34 +266,6 @@ mod tests {
         assert!(!prepared.text_request.decode_options.skip_special_tokens);
     }
 
-    #[test]
-    fn prepare_completion_request_passes_through_thinking_token_budget() {
-        let prepare = |budget: serde_json::Value| {
-            let request: CompletionRequest = serde_json::from_value(json!({
-                "model": "Qwen/Qwen1.5-0.5B-Chat",
-                "prompt": "hello",
-                "thinking_token_budget": budget,
-            }))
-            .expect("parse request");
-            prepare_completion_request(
-                request,
-                &served(&["Qwen/Qwen1.5-0.5B-Chat"]),
-                ResolvedRequestContext::default(),
-            )
-            .expect("prepare")
-            .text_request
-            .sampling_params
-            .thinking_token_budget
-        };
-
-        // The convert layer forwards the raw value verbatim (including the `-1`
-        // "unlimited" sentinel); normalization/validation happens during
-        // lowering (see `vllm_text::lower`).
-        assert_eq!(prepare(json!(64)), Some(64));
-        assert_eq!(prepare(json!(-1)), Some(-1));
-        assert_eq!(prepare(json!(null)), None);
-    }
-
     #[test]
     fn prepare_completion_request_maps_stream_usage_and_token_format_options() {
         let request: CompletionRequest = serde_json::from_value(json!({

rust/src/server/src/routes/openai/completions/types.rs

@@ -146,11 +146,6 @@ pub struct CompletionRequest {
     /// Additional kwargs for structured outputs
     pub structured_outputs: Option<Value>,
 
-    /// Token budget for reasoning/thinking. Accepts a non-negative integer, or
-    /// `-1` for unlimited (mirroring the Python frontend, which normalizes `-1`
-    /// to "no budget").
-    pub thinking_token_budget: Option<i64>,
-
     /// Request scheduling priority (lower means earlier; default 0)
     pub priority: Option<i32>,
 

rust/src/text/src/error.rs

@@ -20,8 +20,6 @@ pub enum Error {
     Logprobs(#[from] LogprobsError),
     #[error(transparent)]
     OutOfVocab(#[from] OutOfVocabError),
-    #[error("`thinking_token_budget` must be a non-negative integer or -1 for unlimited.")]
-    InvalidThinkingTokenBudget,
     #[error("text request stream `{request_id}` closed before terminal output")]
     StreamClosedBeforeTerminalOutput { request_id: String },
     #[error(transparent)]

rust/src/text/src/lower.rs

@@ -87,7 +87,6 @@ pub fn lower_sampling_params(
         seed,
         max_tokens,
         min_tokens,
-        thinking_token_budget,
         logprobs,
         prompt_logprobs,
         min_p,
@@ -129,7 +128,6 @@ pub fn lower_sampling_params(
         prompt_len,
     )?;
     let min_tokens = min_tokens.unwrap_or(0);
-    let thinking_token_budget = normalize_thinking_token_budget(thinking_token_budget)?;
     let frequency_penalty = frequency_penalty.unwrap_or(0.0);
     let presence_penalty = presence_penalty.unwrap_or(0.0);
 
@@ -151,7 +149,6 @@ pub fn lower_sampling_params(
         seed,
         max_tokens,
         min_tokens,
-        thinking_token_budget,
         logprobs,
         prompt_logprobs,
         min_p,
@@ -173,21 +170,6 @@ pub fn lower_sampling_params(
     Ok(params)
 }
 
-/// Normalize the user-facing `thinking_token_budget` into the engine value.
-///
-/// Mirrors Python's `validate_thinking_token_budget`
-/// (<https://github.com/vllm-project/vllm/blob/ecf9d83520eb217401b47d8a5451a27c5231b8c2/vllm/sampling_params.py#L35-L55>):
-/// `None` and the `-1` "unlimited" sentinel both map to `None`; any other
-/// negative value is rejected; non-negative values pass through unchanged. Like
-/// Python's `int`, no upper bound is imposed.
-fn normalize_thinking_token_budget(value: Option<i64>) -> Result<Option<u64>> {
-    match value {
-        None | Some(-1) => Ok(None),
-        Some(budget) if budget >= 0 => Ok(Some(budget as u64)),
-        Some(_) => Err(Error::InvalidThinkingTokenBudget),
-    }
-}
-
 /// Convert bad-word strings into token-ID sequences, following the Python vLLM
 /// logic in `SamplingParams.update_from_tokenizer()`.
 ///
@@ -384,36 +366,6 @@ mod tests {
         )
     }
 
-    #[test]
-    fn lower_sampling_params_normalizes_thinking_token_budget() {
-        let lower = |budget: Option<i64>| {
-            lower_sampling_params_with_limits(
-                SamplingParams {
-                    thinking_token_budget: budget,
-                    ..SamplingParams::default()
-                },
-                sample_sampling_limits(),
-            )
-        };
-
-        // Non-negative budgets (including 0) pass through unchanged.
-        assert_eq!(lower(Some(256)).unwrap().thinking_token_budget, Some(256));
-        assert_eq!(lower(Some(0)).unwrap().thinking_token_budget, Some(0));
-        // `None` and the `-1` "unlimited" sentinel both disable the budget.
-        assert_eq!(lower(None).unwrap().thinking_token_budget, None);
-        assert_eq!(lower(Some(-1)).unwrap().thinking_token_budget, None);
-        // No upper bound is imposed, matching Python's `int`.
-        assert_eq!(
-            lower(Some(i64::from(u32::MAX) + 1)).unwrap().thinking_token_budget,
-            Some(u64::from(u32::MAX) + 1)
-        );
-        // Other negatives are rejected.
-        assert!(matches!(
-            lower(Some(-2)),
-            Err(Error::InvalidThinkingTokenBudget)
-        ));
-    }
-
     #[test]
     fn lower_text_request_applies_python_style_eos_hints() {
         let prepared = lower_text_request(
@@ -434,7 +386,6 @@ mod tests {
                 seed: None,
                 max_tokens: 999997,
                 min_tokens: 0,
-                thinking_token_budget: None,
                 logprobs: None,
                 prompt_logprobs: None,
                 min_p: 0.0,
@@ -486,7 +437,6 @@ mod tests {
                 seed: None,
                 max_tokens: 999997,
                 min_tokens: 0,
-                thinking_token_budget: None,
                 logprobs: None,
                 prompt_logprobs: None,
                 min_p: 0.0,
@@ -617,7 +567,6 @@ mod tests {
                 seed: None,
                 max_tokens: 40957,
                 min_tokens: 0,
-                thinking_token_budget: None,
                 logprobs: None,
                 prompt_logprobs: None,
                 min_p: 0.0,
@@ -679,7 +628,6 @@ mod tests {
                 seed: None,
                 max_tokens: 999997,
                 min_tokens: 0,
-                thinking_token_budget: None,
                 logprobs: None,
                 prompt_logprobs: None,
                 min_p: 0.0,
@@ -749,7 +697,6 @@ mod tests {
                 seed: None,
                 max_tokens: 32,
                 min_tokens: 2,
-                thinking_token_budget: None,
                 logprobs: None,
                 prompt_logprobs: None,
                 min_p: 0.1,
@@ -982,7 +929,6 @@ mod tests {
                 seed: None,
                 max_tokens: 128,
                 min_tokens: 0,
-                thinking_token_budget: None,
                 logprobs: None,
                 prompt_logprobs: None,
                 min_p: 0.1,

rust/src/text/src/output/decoded.rs

@@ -309,7 +309,7 @@ fn matches_stop_string(stops: &[String], output: &str, new_bytes: usize) -> Opti
         .find_map(|(ss_idx, (ss, len, start_off))| {
             output[start_off..]
                 .windows(len)
-                .position(|w| w == ss)
+                .rposition(|w| w == ss)
                 .map(|pos| (ss_idx, start_off + pos))
         })
 }
@@ -562,13 +562,6 @@ mod tests {
         assert_eq!(result, Some((0, 4)));
     }
 
-    #[test]
-    fn stop_string_matches_leftmost_with_multiple_new_bytes() {
-        let stops = vec!["\n".to_string()];
-        let result = matches_stop_string(&stops, "Answer\n\n", 2);
-        assert_eq!(result, Some((0, 6)));
-    }
-
     #[test]
     fn stop_string_matches_at_beginning() {
         let stops = vec!["say".to_string()];

rust/src/text/src/request.rs

@@ -56,12 +56,6 @@ pub struct SamplingParams {
     pub max_tokens: Option<u32>,
     /// Minimum number of tokens to generate before EOS or stop-token handling.
     pub min_tokens: Option<u32>,
-    /// Maximum number of reasoning ("thinking") tokens to emit before the
-    /// reasoning section is force-closed. `None` or the user-facing `-1`
-    /// "unlimited" sentinel both disable the budget. The raw value is carried
-    /// here; `-1` is normalized to `None` (and other negatives rejected) during
-    /// lowering (see `lower_sampling_params`).
-    pub thinking_token_budget: Option<i64>,
     /// Number of log probabilities to return per generated token.
     ///
     /// `None` disables sample logprobs. `-1` requests the full vocabulary.
@@ -122,7 +116,6 @@ impl Default for SamplingParams {
             seed: None,
             max_tokens: None,
             min_tokens: None,
-            thinking_token_budget: None,
             logprobs: None,
             prompt_logprobs: None,
             min_p: None,

setup.py

@@ -1171,7 +1171,15 @@ package_data = {
         "third_party/deep_gemm/include/**/*.h",
         "third_party/deep_gemm/include/**/*.hpp",
         # fmha_sm100 sparse CuTe-DSL helper kernels (vendored via cmake)
+        "third_party/fmha_sm100/csrc/**/*.cu",
+        "third_party/fmha_sm100/csrc/**/*.h",
+        "third_party/fmha_sm100/csrc/**/*.jinja",
+        "third_party/fmha_sm100/csrc/**/*.cu.jinja",
         "third_party/fmha_sm100/cute/**/*.cu",
+        "third_party/fmha_sm100/cutlass/include/**/*.h",
+        "third_party/fmha_sm100/cutlass/include/**/*.hpp",
+        "third_party/fmha_sm100/cutlass/tools/util/include/**/*.h",
+        "third_party/fmha_sm100/cutlass/tools/util/include/**/*.hpp",
     ]
 }
 

tests/compile/passes/test_double_aiter_rms_quant_fusion.py

@@ -22,7 +22,7 @@ import torch
 
 import vllm.config
 from tests.compile.backend import TestBackend
-from vllm._aiter_ops import rocm_aiter_ops
+from vllm._aiter_ops import is_aiter_found_and_supported, rocm_aiter_ops
 from vllm.compilation.passes.utility.noop_elimination import NoOpEliminationPass
 from vllm.compilation.passes.utility.post_cleanup import PostCleanupPass
 from vllm.config import (
@@ -83,8 +83,9 @@ class _ViewDoubleQuantModel(torch.nn.Module):
     [_NoViewDoubleQuantModel, _ViewDoubleQuantModel],
     ids=["no_view", "with_view"],
 )
-@pytest.mark.skip(
-    reason="Skipping for now because pytorch compiler removes one the two quant ops"
+@pytest.mark.skipif(
+    not is_aiter_found_and_supported(),
+    reason="Only test on ROCm with AITER installed and supported",
 )
 def test_double_aiter_rms_fp8_group_quant_fusion(
     model_cls: type[torch.nn.Module],

tests/distributed/test_pipeline_parallel.py

@@ -175,7 +175,7 @@ MULTIMODAL_MODELS = {
     "facebook/chameleon-7b": PPTestSettings.fast(),
     "adept/fuyu-8b": PPTestSettings.fast(),
     "zai-org/glm-4v-9b": PPTestSettings.fast(),
-    "OpenGVLab/InternVL3-1B": PPTestSettings.fast(),
+    "OpenGVLab/InternVL2-1B": PPTestSettings.fast(),
     "llava-hf/llava-1.5-7b-hf": PPTestSettings.fast(),
     "llava-hf/llava-v1.6-mistral-7b-hf": PPTestSettings.fast(),
     "llava-hf/LLaVA-NeXT-Video-7B-hf": PPTestSettings.fast(),
@@ -203,7 +203,7 @@ TEST_MODELS = [
     "intfloat/e5-mistral-7b-instruct",
     "BAAI/bge-multilingual-gemma2",
     # [MULTIMODAL GENERATION]
-    "OpenGVLab/InternVL3-1B",
+    "OpenGVLab/InternVL2-1B",
     "microsoft/Phi-3.5-vision-instruct",
     "fixie-ai/ultravox-v0_5-llama-3_2-1b",
     # [LANGUAGE GENERATION - HYBRID ARCH]

tests/entrypoints/anthropic/test_anthropic_messages_conversion.py

@@ -8,8 +8,6 @@ AnthropicServingMessages._convert_anthropic_to_openai_request().
 Also covers extended-thinking edge cases such as ``redacted_thinking``
 blocks echoed back by Anthropic clients, and streaming conversion in
 ``message_stream_converter``.
-
-Also covers cache usage computation in ``_build_anthropic_usage``.
 """
 
 import json
@@ -20,11 +18,7 @@ import pytest
 from vllm.entrypoints.anthropic.protocol import (
     AnthropicMessagesRequest,
 )
-from vllm.entrypoints.anthropic.serving import (
-    AnthropicServingMessages,
-    _build_anthropic_usage,
-    _get_cached_tokens,
-)
+from vllm.entrypoints.anthropic.serving import AnthropicServingMessages
 from vllm.entrypoints.openai.chat_completion.protocol import (
     ChatCompletionResponseStreamChoice,
     ChatCompletionStreamResponse,
@@ -33,7 +27,6 @@ from vllm.entrypoints.openai.engine.protocol import (
     DeltaFunctionCall,
     DeltaMessage,
     DeltaToolCall,
-    PromptTokenUsageInfo,
     UsageInfo,
 )
 
@@ -660,108 +653,6 @@ class TestThinkingBlockConversion:
         assert asst.get("content") == "Hi!"
 
 
-# ======================================================================
-# Cache usage computation
-# ======================================================================
-
-
-class TestGetCachedTokens:
-    """Tests for _get_cached_tokens helper."""
-
-    def test_none_usage(self):
-        assert _get_cached_tokens(None) is None
-
-    def test_no_prompt_tokens_details(self):
-        usage = UsageInfo(prompt_tokens=100, completion_tokens=10)
-        assert _get_cached_tokens(usage) is None
-
-    def test_cached_tokens_present(self):
-        usage = UsageInfo(
-            prompt_tokens=100,
-            completion_tokens=10,
-            prompt_tokens_details=PromptTokenUsageInfo(cached_tokens=80),
-        )
-        assert _get_cached_tokens(usage) == 80
-
-    def test_cached_tokens_zero(self):
-        """Zero cached tokens should return 0, not None."""
-        usage = UsageInfo(
-            prompt_tokens=100,
-            completion_tokens=10,
-            prompt_tokens_details=PromptTokenUsageInfo(cached_tokens=0),
-        )
-        assert _get_cached_tokens(usage) == 0
-
-    def test_cached_tokens_none_in_details(self):
-        usage = UsageInfo(
-            prompt_tokens=100,
-            completion_tokens=10,
-            prompt_tokens_details=PromptTokenUsageInfo(cached_tokens=None),
-        )
-        assert _get_cached_tokens(usage) is None
-
-
-class TestBuildAnthropicUsage:
-    """Tests for _build_anthropic_usage helper.
-
-    Anthropic defines: total_input = input_tokens + cache_read + cache_creation
-    vLLM's prompt_tokens is the total.
-    """
-
-    def test_no_cache_info(self):
-        """When cache info is unavailable, return raw prompt_tokens."""
-        result = _build_anthropic_usage(100, 10, None)
-        assert result.input_tokens == 100
-        assert result.output_tokens == 10
-        assert result.cache_read_input_tokens is None
-        assert result.cache_creation_input_tokens is None
-
-    def test_cache_hit(self):
-        """When cache is hit, input_tokens excludes cached tokens."""
-        usage = UsageInfo(
-            prompt_tokens=100,
-            completion_tokens=10,
-            prompt_tokens_details=PromptTokenUsageInfo(cached_tokens=80),
-        )
-        result = _build_anthropic_usage(100, 10, usage)
-        assert result.input_tokens == 20  # 100 - 80
-        assert result.output_tokens == 10
-        assert result.cache_read_input_tokens == 80
-        assert result.cache_creation_input_tokens == 0
-
-    def test_zero_cached_tokens(self):
-        """Zero cached tokens should still set cache_creation to 0."""
-        usage = UsageInfo(
-            prompt_tokens=100,
-            completion_tokens=10,
-            prompt_tokens_details=PromptTokenUsageInfo(cached_tokens=0),
-        )
-        result = _build_anthropic_usage(100, 10, usage)
-        assert result.input_tokens == 100  # 100 - 0
-        assert result.cache_read_input_tokens == 0
-        assert result.cache_creation_input_tokens == 0
-
-    def test_all_tokens_cached(self):
-        """When all tokens are cached, input_tokens should be 0."""
-        usage = UsageInfo(
-            prompt_tokens=100,
-            completion_tokens=10,
-            prompt_tokens_details=PromptTokenUsageInfo(cached_tokens=100),
-        )
-        result = _build_anthropic_usage(100, 10, usage)
-        assert result.input_tokens == 0
-        assert result.cache_read_input_tokens == 100
-        assert result.cache_creation_input_tokens == 0
-
-    def test_no_prompt_tokens_details(self):
-        """UsageInfo without prompt_tokens_details returns no cache info."""
-        usage = UsageInfo(prompt_tokens=100, completion_tokens=10)
-        result = _build_anthropic_usage(100, 10, usage)
-        assert result.input_tokens == 100
-        assert result.cache_read_input_tokens is None
-        assert result.cache_creation_input_tokens is None
-
-
 class TestInlineSystemMessageInMessagesArray:
     """Verify that ``role: system`` messages embedded inside the ``messages``
     array are preserved in their original position.
@@ -1207,135 +1098,6 @@ class TestMessageStartIncludesTypeAndRole:
         assert message["role"] == "assistant"
 
 
-class TestStreamingCacheUsageSemantics:
-    """Locks in the documented streaming behavior of cache usage fields.
-
-    vLLM's OpenAI chat completion streaming only attaches
-    ``prompt_tokens_details`` to the terminal usage chunk. The Anthropic layer
-    mirrors that contract: cache fields are omitted on ``message_start`` (key
-    absence signals "unknown") and populated on ``message_delta`` (the final
-    cumulative count). This is intentionally consistent with vLLM's OpenAI
-    behavior, even though Anthropic's upstream API populates cache fields on
-    ``message_start``; closing that gap requires plumbing cache info into the
-    first chunk at the OpenAI layer, which is out of scope here.
-    """
-
-    @pytest.mark.asyncio
-    async def test_streaming_cache_fields_absent_then_populated(self):
-        """First chunk lacks prompt_tokens_details (vLLM contract);
-        message_start omits cache fields. The final chunk carries
-        prompt_tokens_details, so message_delta carries resolved values."""
-
-        async def sse_input():
-            yield _make_stream_chunk(
-                delta=DeltaMessage(role="assistant", content="hi"),
-                usage=UsageInfo(prompt_tokens=100, total_tokens=100),
-            )
-            yield _make_stream_chunk(finish_reason="stop")
-            yield _make_stream_chunk(
-                choices=[],
-                usage=UsageInfo(
-                    prompt_tokens=100,
-                    completion_tokens=5,
-                    total_tokens=105,
-                    prompt_tokens_details=PromptTokenUsageInfo(cached_tokens=80),
-                ),
-            )
-            yield "data: [DONE]"
-
-        converter = _make_stream_converter()
-        output = []
-        async for event in converter.message_stream_converter(sse_input()):
-            output.append(event)
-        events = _parse_sse_events(output)
-
-        # message_start: cache fields unknown → omitted from JSON entirely.
-        start_usage = events[0][1]["message"]["usage"]
-        assert events[0][0] == "message_start"
-        assert start_usage["input_tokens"] == 100
-        assert "cache_read_input_tokens" not in start_usage
-        assert "cache_creation_input_tokens" not in start_usage
-
-        # message_delta: authoritative usage with cache fields populated.
-        delta_usage = next(
-            data["usage"] for ev, data in events if ev == "message_delta"
-        )
-        assert delta_usage["input_tokens"] == 20  # 100 - 80
-        assert delta_usage["cache_read_input_tokens"] == 80
-        assert delta_usage["cache_creation_input_tokens"] == 0
-
-    @pytest.mark.asyncio
-    async def test_streaming_no_cache_hit(self):
-        """When the final chunk reports cached_tokens=0, message_delta carries
-        cache fields = 0 (cache miss); message_start still omits them."""
-
-        async def sse_input():
-            yield _make_stream_chunk(
-                delta=DeltaMessage(role="assistant"),
-                usage=UsageInfo(prompt_tokens=50, total_tokens=50),
-            )
-            yield _make_stream_chunk(finish_reason="stop")
-            yield _make_stream_chunk(
-                choices=[],
-                usage=UsageInfo(
-                    prompt_tokens=50,
-                    completion_tokens=5,
-                    total_tokens=55,
-                    prompt_tokens_details=PromptTokenUsageInfo(cached_tokens=0),
-                ),
-            )
-            yield "data: [DONE]"
-
-        converter = _make_stream_converter()
-        output = []
-        async for event in converter.message_stream_converter(sse_input()):
-            output.append(event)
-        events = _parse_sse_events(output)
-
-        start_usage = events[0][1]["message"]["usage"]
-        delta_usage = next(
-            data["usage"] for ev, data in events if ev == "message_delta"
-        )
-        assert start_usage["input_tokens"] == 50
-        assert "cache_read_input_tokens" not in start_usage
-        assert "cache_creation_input_tokens" not in start_usage
-        assert delta_usage["input_tokens"] == 50  # 50 - 0
-        assert delta_usage["cache_read_input_tokens"] == 0
-        assert delta_usage["cache_creation_input_tokens"] == 0
-
-    @pytest.mark.asyncio
-    async def test_streaming_no_prompt_tokens_details_at_all(self):
-        """If --enable-prompt-tokens-details is off, no chunk carries cache
-        info; both message_start and message_delta omit cache fields."""
-
-        async def sse_input():
-            yield _make_stream_chunk(
-                delta=DeltaMessage(role="assistant"),
-                usage=UsageInfo(prompt_tokens=30, total_tokens=30),
-            )
-            yield _make_stream_chunk(finish_reason="stop")
-            yield _make_stream_chunk(
-                choices=[],
-                usage=UsageInfo(prompt_tokens=30, completion_tokens=2, total_tokens=32),
-            )
-            yield "data: [DONE]"
-
-        converter = _make_stream_converter()
-        output = []
-        async for event in converter.message_stream_converter(sse_input()):
-            output.append(event)
-        events = _parse_sse_events(output)
-
-        start_usage = events[0][1]["message"]["usage"]
-        delta_usage = next(
-            data["usage"] for ev, data in events if ev == "message_delta"
-        )
-        assert "cache_read_input_tokens" not in start_usage
-        assert "cache_creation_input_tokens" not in start_usage
-        assert "cache_read_input_tokens" not in delta_usage
-        assert "cache_creation_input_tokens" not in delta_usage
-
-
 # ======================================================================
 # Auto-detection of system-first template requirement
 # ======================================================================

tests/entrypoints/pooling/classify/test_online_vision.py

@@ -25,7 +25,7 @@ def server():
         "--runner",
         "pooling",
         "--max-model-len",
-        "16384",
+        "5000",
         "--enforce-eager",
         "--limit-mm-per-prompt",
         json.dumps({"video": MAXIMUM_VIDEOS}),
@@ -143,4 +143,4 @@ def test_chat_video_url_request(server: RemoteOpenAIServer, model_name: str):
     assert output.model == model_name
     assert len(output.data) == 1
     assert len(output.data[0].probs) == 2
-    assert output.usage.prompt_tokens == 8993
+    assert output.usage.prompt_tokens == 4807

tests/entrypoints/serve/render/test_derender.py

@@ -8,7 +8,6 @@ import pytest
 import pytest_asyncio
 
 from tests.utils import RemoteLaunchRenderServer
-from vllm.tokenizers import get_tokenizer
 
 MODEL_NAME = "hmellor/tiny-random-LlamaForCausalLM"
 
@@ -487,438 +486,3 @@ async def test_derender_completion_kv_transfer_params_passthrough(client):
     )
     assert response.status_code == 200
     assert response.json()["kv_transfer_params"] == kv
-
-
-# ---------------------------------------------------------------------------
-# E2E: render -> derender roundtrip with parser (reasoning + tool calls)
-# ---------------------------------------------------------------------------
-
-PARSER_MODEL = "deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B"
-
-_E2E_TOOLS = [
-    {
-        "type": "function",
-        "function": {
-            "name": "get_weather",
-            "description": "Get weather for a city",
-            "parameters": {
-                "type": "object",
-                "properties": {"city": {"type": "string"}},
-            },
-        },
-    }
-]
-
-
-@pytest.fixture(scope="module")
-def parser_server():
-    args = [
-        "--enable-auto-tool-choice",
-        "--tool-call-parser",
-        "hermes",
-        "--reasoning-parser",
-        "deepseek_r1",
-    ]
-    with RemoteLaunchRenderServer(PARSER_MODEL, args) as remote_server:
-        yield remote_server
-
-
-@pytest_asyncio.fixture
-async def parser_client(parser_server):
-    async with httpx.AsyncClient(
-        base_url=parser_server.url_for(""), timeout=60.0
-    ) as http_client:
-        yield http_client
-
-
-@pytest.fixture(scope="module")
-def parser_tokenizer():
-    return get_tokenizer(PARSER_MODEL)
-
-
-def _encode(tokenizer, text: str) -> list[int]:
-    return tokenizer.encode(text, add_special_tokens=False)
-
-
-def _decoded(tokenizer, token_ids: list[int]) -> str:
-    return tokenizer.decode(token_ids, skip_special_tokens=True)
-
-
-def _require_markers_survive(tokenizer, text: str, *markers: str) -> list[int]:
-    """Encode text and skip the test if any marker is lost in roundtrip."""
-    ids = _encode(tokenizer, text)
-    decoded = tokenizer.decode(ids, skip_special_tokens=False)
-    for m in markers:
-        if m not in decoded:
-            pytest.skip(f"Marker {m!r} lost in encode->decode roundtrip")
-    return ids
-
-
-async def _e2e_render_chat(
-    client: httpx.AsyncClient,
-    model: str,
-    messages: list[dict],
-) -> dict:
-    resp = await client.post(
-        "/v1/chat/completions/render",
-        json={"model": model, "messages": messages},
-    )
-    assert resp.status_code == 200, resp.text
-    return resp.json()
-
-
-def _e2e_generate_response(
-    token_ids: list[int],
-    request_id: str = "chatcmpl-e2e-test",
-) -> dict:
-    return {
-        "request_id": request_id,
-        "choices": [
-            {
-                "index": 0,
-                "token_ids": token_ids,
-                "finish_reason": "stop",
-            }
-        ],
-    }
-
-
-@pytest.mark.asyncio
-async def test_e2e_plain_roundtrip(parser_client, parser_tokenizer):
-    """Plain text without reasoning markers roundtrips correctly."""
-    messages = [{"role": "user", "content": "What is 2+2?"}]
-    gen_req = await _e2e_render_chat(parser_client, PARSER_MODEL, messages)
-
-    answer = "The answer is four."
-    output_ids = _encode(parser_tokenizer, answer)
-    expected = _decoded(parser_tokenizer, output_ids)
-
-    resp = await parser_client.post(
-        "/v1/chat/completions/derender",
-        json={
-            "model": PARSER_MODEL,
-            "generate_response": _e2e_generate_response(output_ids),
-            "prompt_tokens": len(gen_req["token_ids"]),
-        },
-    )
-    assert resp.status_code == 200, resp.text
-    content = resp.json()["choices"][0]["message"]["content"]
-    assert content == expected
-
-
-@pytest.mark.asyncio
-async def test_e2e_token_identity(parser_client, parser_tokenizer):
-    """encode(derender(token_ids)) == token_ids (RL invariant)."""
-    messages = [{"role": "user", "content": "Hi"}]
-    gen_req = await _e2e_render_chat(parser_client, PARSER_MODEL, messages)
-
-    answer = "Hello! How can I help?"
-    output_ids = _encode(parser_tokenizer, answer)
-
-    resp = await parser_client.post(
-        "/v1/chat/completions/derender",
-        json={
-            "model": PARSER_MODEL,
-            "generate_response": _e2e_generate_response(output_ids),
-            "prompt_tokens": len(gen_req["token_ids"]),
-        },
-    )
-    assert resp.status_code == 200
-    content = resp.json()["choices"][0]["message"]["content"]
-    re_encoded = _encode(parser_tokenizer, content)
-    assert output_ids == re_encoded
-
-
-@pytest.mark.asyncio
-async def test_e2e_non_ascii_roundtrip(parser_client, parser_tokenizer):
-    """CJK + emoji roundtrip without U+FFFD."""
-    messages = [{"role": "user", "content": "Reply in Chinese"}]
-    gen_req = await _e2e_render_chat(parser_client, PARSER_MODEL, messages)
-
-    answer = "你好世界 😀"
-    output_ids = _encode(parser_tokenizer, answer)
-
-    resp = await parser_client.post(
-        "/v1/chat/completions/derender",
-        json={
-            "model": PARSER_MODEL,
-            "generate_response": _e2e_generate_response(output_ids),
-            "prompt_tokens": len(gen_req["token_ids"]),
-        },
-    )
-    assert resp.status_code == 200
-    content = resp.json()["choices"][0]["message"]["content"]
-    assert "�" not in content
-
-
-@pytest.mark.asyncio
-async def test_e2e_parsed_reasoning(parser_client, parser_tokenizer):
-    """<think>...</think> splits into reasoning + content."""
-    messages = [{"role": "user", "content": "What is 2+3?"}]
-    gen_req = await _e2e_render_chat(parser_client, PARSER_MODEL, messages)
-
-    reasoning_text = "The user wants 2 plus 3. That is 5."
-    answer_text = "The answer is 5."
-    output_text = f"<think>{reasoning_text}</think>{answer_text}"
-    output_ids = _require_markers_survive(parser_tokenizer, output_text, "</think>")
-
-    resp = await parser_client.post(
-        "/v1/chat/completions/derender",
-        json={
-            "model": PARSER_MODEL,
-            "generate_response": _e2e_generate_response(output_ids),
-            "prompt_tokens": len(gen_req["token_ids"]),
-            "chat_request": {
-                "model": PARSER_MODEL,
-                "messages": messages,
-                "include_reasoning": True,
-            },
-        },
-    )
-    assert resp.status_code == 200, resp.text
-    msg = resp.json()["choices"][0]["message"]
-    assert msg["reasoning"] is not None
-    assert reasoning_text in msg["reasoning"]
-    assert answer_text in msg["content"]
-    assert "<think>" not in msg["content"]
-
-
-@pytest.mark.asyncio
-async def test_e2e_parsed_tool_call(parser_client, parser_tokenizer):
-    """<tool_call> extracted into tool_calls field."""
-    messages = [{"role": "user", "content": "Weather in Paris?"}]
-    gen_req = await _e2e_render_chat(parser_client, PARSER_MODEL, messages)
-
-    output_text = (
-        "<think>Let me check the weather.</think>"
-        '<tool_call>\n{"name": "get_weather", '
-        '"arguments": {"city": "Paris"}}\n</tool_call>'
-    )
-    output_ids = _require_markers_survive(
-        parser_tokenizer,
-        output_text,
-        "</think>",
-        "<tool_call>",
-        "</tool_call>",
-    )
-
-    resp = await parser_client.post(
-        "/v1/chat/completions/derender",
-        json={
-            "model": PARSER_MODEL,
-            "generate_response": _e2e_generate_response(output_ids),
-            "prompt_tokens": len(gen_req["token_ids"]),
-            "chat_request": {
-                "model": PARSER_MODEL,
-                "messages": messages,
-                "tools": _E2E_TOOLS,
-                "tool_choice": "auto",
-            },
-        },
-    )
-    assert resp.status_code == 200, resp.text
-    choice = resp.json()["choices"][0]
-    assert choice["message"]["tool_calls"]
-    assert choice["message"]["tool_calls"][0]["function"]["name"] == "get_weather"
-
-
-@pytest.mark.asyncio
-async def test_e2e_parsed_reasoning_and_tool_call(parser_client, parser_tokenizer):
-    """Reasoning + tool call in the same output."""
-    messages = [{"role": "user", "content": "Weather in Paris?"}]
-    gen_req = await _e2e_render_chat(parser_client, PARSER_MODEL, messages)
-
-    reasoning_text = "I should look up the weather."
-    tool_text = (
-        '<tool_call>\n{"name": "get_weather", '
-        '"arguments": {"city": "Paris"}}\n</tool_call>'
-    )
-    output_text = f"<think>{reasoning_text}</think>{tool_text}"
-    output_ids = _require_markers_survive(
-        parser_tokenizer, output_text, "</think>", "<tool_call>"
-    )
-
-    resp = await parser_client.post(
-        "/v1/chat/completions/derender",
-        json={
-            "model": PARSER_MODEL,
-            "generate_response": _e2e_generate_response(output_ids),
-            "prompt_tokens": len(gen_req["token_ids"]),
-            "chat_request": {
-                "model": PARSER_MODEL,
-                "messages": messages,
-                "tools": _E2E_TOOLS,
-                "tool_choice": "auto",
-                "include_reasoning": True,
-            },
-        },
-    )
-    assert resp.status_code == 200, resp.text
-    choice = resp.json()["choices"][0]
-    assert choice["message"]["reasoning"] is not None
-    assert reasoning_text in choice["message"]["reasoning"]
-    assert choice["message"]["tool_calls"]
-
-
-@pytest.mark.asyncio
-async def test_e2e_no_chat_request_fallback(parser_client, parser_tokenizer):
-    """Without chat_request, derender falls back to plain detokenization."""
-    messages = [{"role": "user", "content": "Hello"}]
-    gen_req = await _e2e_render_chat(parser_client, PARSER_MODEL, messages)
-
-    answer = "Hi there!"
-    output_ids = _encode(parser_tokenizer, answer)
-
-    resp = await parser_client.post(
-        "/v1/chat/completions/derender",
-        json={
-            "model": PARSER_MODEL,
-            "generate_response": _e2e_generate_response(output_ids),
-            "prompt_tokens": len(gen_req["token_ids"]),
-        },
-    )
-    assert resp.status_code == 200
-    content = resp.json()["choices"][0]["message"]["content"]
-    assert "Hi" in content
-
-
-# ---------------------------------------------------------------------------
-# E2E: HarmonyParser + GPT-OSS
-# ---------------------------------------------------------------------------
-
-HARMONY_MODEL = "openai/gpt-oss-20b"
-
-
-def _ensure_harmony_vocab():
-    """Pre-cache the o200k_base BPE file needed by openai-harmony.
-
-    The Rust tiktoken-rs backend downloads from Azure Blob Storage, which
-    may be unreachable in some environments.  When the cache is cold we
-    fetch the file ourselves and place it in ``/tmp/tiktoken-rs-cache/``
-    using the SHA-1(URL) filename that tiktoken-rs expects.
-    """
-    import hashlib
-    import urllib.request
-    from pathlib import Path
-
-    url = "https://openaipublic.blob.core.windows.net/encodings/o200k_base.tiktoken"
-    cache_dir = Path("/tmp/tiktoken-rs-cache")
-    cache_key = hashlib.sha1(url.encode()).hexdigest()
-    cache_file = cache_dir / cache_key
-    if not cache_file.exists():
-        cache_dir.mkdir(parents=True, exist_ok=True)
-        urllib.request.urlretrieve(url, cache_file)
-
-
-@pytest.fixture(scope="module")
-def harmony_server():
-    _ensure_harmony_vocab()
-    args = [
-        "--trust-remote-code",
-        "--enable-auto-tool-choice",
-        "--tool-call-parser",
-        "openai",
-        "--reasoning-parser",
-        "openai_gptoss",
-    ]
-    with RemoteLaunchRenderServer(HARMONY_MODEL, args) as remote_server:
-        yield remote_server
-
-
-@pytest_asyncio.fixture
-async def harmony_client(harmony_server):
-    async with httpx.AsyncClient(
-        base_url=harmony_server.url_for(""), timeout=60.0
-    ) as http_client:
-        yield http_client
-
-
-@pytest.fixture(scope="module")
-def harmony_tokenizer():
-    return get_tokenizer(HARMONY_MODEL, trust_remote_code=True)
-
-
-def _harmony_extract_assistant_ids(
-    tokenizer, assistant_msg: dict, user_content: str = "test"
-) -> list[int]:
-    """Extract assistant token IDs via apply_chat_template diff."""
-    prompt = [{"role": "user", "content": user_content}]
-    full = prompt + [assistant_msg]
-    text_prompt = tokenizer.apply_chat_template(
-        prompt, add_generation_prompt=True, tokenize=False
-    )
-    text_full = tokenizer.apply_chat_template(
-        full, add_generation_prompt=False, tokenize=False
-    )
-    prompt_ids = tokenizer.encode(text_prompt)
-    full_ids = tokenizer.encode(text_full)
-    assistant_ids = list(full_ids[len(prompt_ids) :])
-    if not assistant_ids:
-        pytest.skip("Could not extract assistant tokens for Harmony")
-    return assistant_ids
-
-
-@pytest.mark.asyncio
-async def test_e2e_harmony_plain_roundtrip(harmony_client, harmony_tokenizer):
-    """GPT-OSS content-only roundtrip."""
-    messages = [{"role": "user", "content": "What is 2+2?"}]
-    gen_req = await _e2e_render_chat(harmony_client, HARMONY_MODEL, messages)
-
-    assistant_msg = {"role": "assistant", "content": "Four."}
-    output_ids = _harmony_extract_assistant_ids(harmony_tokenizer, assistant_msg)
-
-    resp = await harmony_client.post(
-        "/v1/chat/completions/derender",
-        json={
-            "model": HARMONY_MODEL,
-            "generate_response": _e2e_generate_response(output_ids),
-            "prompt_tokens": len(gen_req["token_ids"]),
-            "chat_request": {
-                "model": HARMONY_MODEL,
-                "messages": messages,
-            },
-        },
-    )
-    assert resp.status_code == 200, resp.text
-    content = resp.json()["choices"][0]["message"]["content"]
-    assert content is not None and len(content) > 0
-    assert "Four" in content
-
-
-@pytest.mark.asyncio
-async def test_e2e_harmony_reasoning(harmony_client, harmony_tokenizer):
-    """GPT-OSS reasoning: analysis channel extracted."""
-    messages = [{"role": "user", "content": "Add 2 and 3."}]
-    gen_req = await _e2e_render_chat(harmony_client, HARMONY_MODEL, messages)
-
-    reasoning_text = "The user wants 2 plus 3."
-    answer_text = "The answer is 5."
-    assistant_msg = {
-        "role": "assistant",
-        "thinking": reasoning_text,
-        "content": answer_text,
-    }
-    output_ids = _harmony_extract_assistant_ids(harmony_tokenizer, assistant_msg)
-
-    decoded = harmony_tokenizer.decode(output_ids)
-    if reasoning_text not in decoded:
-        pytest.skip("Harmony template did not render thinking")
-
-    resp = await harmony_client.post(
-        "/v1/chat/completions/derender",
-        json={
-            "model": HARMONY_MODEL,
-            "generate_response": _e2e_generate_response(output_ids),
-            "prompt_tokens": len(gen_req["token_ids"]),
-            "chat_request": {
-                "model": HARMONY_MODEL,
-                "messages": messages,
-                "include_reasoning": True,
-            },
-        },
-    )
-    assert resp.status_code == 200, resp.text
-    msg = resp.json()["choices"][0]["message"]
-    assert msg["reasoning"] is not None
-    assert reasoning_text in msg["reasoning"]
-    assert answer_text in (msg["content"] or "")

tests/entrypoints/serve/tokenize/test_serving_tokenization.py

@@ -15,7 +15,7 @@ from vllm.entrypoints.serve.tokenize.protocol import (
     TokenizeChatRequest,
     TokenizeCompletionRequest,
 )
-from vllm.entrypoints.serve.tokenize.serving import ServingTokenization
+from vllm.entrypoints.serve.tokenize.serving import OpenAIServingTokenization
 from vllm.v1.engine.async_llm import AsyncLLM
 
 MODEL_NAME = "openai-community/gpt2"
@@ -58,7 +58,7 @@ class MockModelConfig:
         return self.diff_sampling_param or {}
 
 
-def _build_serving_tokenization(engine: AsyncLLM) -> ServingTokenization:
+def _build_serving_tokenization(engine: AsyncLLM) -> OpenAIServingTokenization:
     models = OpenAIServingModels(
         engine_client=engine,
         base_model_paths=BASE_MODEL_PATHS,
@@ -71,7 +71,8 @@ def _build_serving_tokenization(engine: AsyncLLM) -> ServingTokenization:
         chat_template=None,
         chat_template_content_format="auto",
     )
-    return ServingTokenization(
+    return OpenAIServingTokenization(
+        engine,
         models,
         openai_serving_render=serving_render,
         request_logger=None,

tests/kernels/attention/test_attention_selector.py

@@ -15,14 +15,16 @@ from vllm.config import (
 from vllm.platforms import current_platform
 from vllm.platforms.cpu import CpuPlatform
 
-if current_platform.is_cuda():
+# CudaPlatform and RocmPlatform import their respective compiled C extensions
+# at module level, raising ModuleNotFoundError on incompatible builds.
+try:
     from vllm.platforms.cuda import CudaPlatform
-else:
+except (ImportError, ModuleNotFoundError):
     CudaPlatform = None
 
-if current_platform.is_rocm():
+try:
     from vllm.platforms.rocm import RocmPlatform
-else:
+except (ImportError, ModuleNotFoundError):
     RocmPlatform = None
 
 from vllm.v1.attention.backends.registry import AttentionBackendEnum
@@ -432,15 +434,9 @@ def test_per_head_quant_scales_backend_selection(
     [
         ("FLASH_ATTN", True, True),  # FlashAttn supports non-causal
         ("FLASH_ATTN", False, True),  # FlashAttn also works with causal
-    ]
-    + (
-        [
-            ("FLASHINFER", True, True),  # FlashInfer supports non-causal
-            ("FLASHINFER", False, True),  # FlashInfer works with causal
-        ]
-        if CudaPlatform is not None
-        else []
-    ),
+        ("FLASHINFER", True, False),  # FlashInfer does not support non-causal
+        ("FLASHINFER", False, True),  # FlashInfer works with causal
+    ],
 )
 def test_non_causal_backend_selection(
     backend_name: str, use_non_causal: bool, should_succeed: bool
@@ -463,12 +459,11 @@ def test_non_causal_backend_selection(
         attention_config=attention_config, cache_config=cache_config
     )
 
-    platform = CudaPlatform or RocmPlatform
-    if platform is None:
-        pytest.skip("CudaPlatform and RocmPlatform are not available")
+    if CudaPlatform is None:
+        pytest.skip("CudaPlatform not available")
     with (
         set_current_vllm_config(vllm_config),
-        patch("vllm.platforms.current_platform", platform()),
+        patch("vllm.platforms.current_platform", CudaPlatform()),
     ):
         if should_succeed:
             backend = get_attn_backend(

tests/kernels/attention/test_cutlass_mla_decode.py

@@ -212,69 +212,3 @@ def test_cutlass_mla_decode(
     print(
         f"{t:.3f} ms, {FLOPS / 10**9 / t:.0f} TFLOPS,", f"{bytes / 10**6 / t:.0f} GB/s"
     )
-
-
-@pytest.mark.skipif(
-    not current_platform.has_device_capability(100),
-    reason=CUTLASS_MLA_UNSUPPORTED_REASON,
-)
-@torch.inference_mode()
-def test_cutlass_mla_decode_cross_layer_view():
-    """The kernel must read the cache's page-dim stride instead of assuming
-    pages are packed back-to-back. A per-layer view into a cross-layer
-    (block-major) cache has stride(0) inflated by num_layers; outputs must
-    match a contiguous cache holding the same data exactly."""
-    device = torch.device("cuda:0")
-    torch.set_default_dtype(torch.bfloat16)
-    torch.set_default_device(device)
-    torch.manual_seed(42)
-
-    b, mean_sk, d, dv, block_size = 4, 512, 576, 512, 64
-    num_layers, layer_idx = 3, 1
-    scale = math.sqrt(d) ** (-1)
-
-    num_pages = b * (mean_sk // block_size)
-    cache_seqlens = torch.full((b,), mean_sk, dtype=torch.int32)
-    block_table = torch.arange(num_pages, dtype=torch.int32).view(
-        b, mean_sk // block_size
-    )
-
-    kv_contig = torch.randn(num_pages, block_size, d)
-    # Neighbor layers hold random data so packed-pages addressing reads
-    # garbage rather than zeros.
-    kv_cross_layer = torch.randn(num_pages, num_layers, block_size, d)
-    kv_view = kv_cross_layer[:, layer_idx]
-    kv_view.copy_(kv_contig)
-    assert kv_view.stride(0) == num_layers * block_size * d
-
-    q_nope = torch.randn(b, 128, dv)
-    q_pe = torch.randn(b, 128, d - dv)
-    sm_count = num_compute_units(device.index)
-    workspace_size = ops.sm100_cutlass_mla_get_workspace_size(
-        mean_sk, b, sm_count, num_kv_splits=1
-    )
-    workspace = torch.empty(workspace_size, dtype=torch.uint8)
-
-    def run(cache):
-        out = torch.empty(b, 128, dv)
-        lse = torch.empty(b, 128, dtype=torch.float32)
-        ops.sm100_cutlass_mla_decode(
-            out,
-            lse,
-            q_nope,
-            q_pe,
-            cache,
-            cache_seqlens,
-            block_table,
-            workspace,
-            scale,
-            1,
-        )
-        return out, lse
-
-    out_contig, lse_contig = run(kv_contig)
-    out_view, lse_view = run(kv_view)
-
-    # Same data and same compute order; only addressing differs.
-    assert torch.equal(out_contig, out_view)
-    assert torch.equal(lse_contig, lse_view)

tests/kernels/attention/test_minimax_m3.py

@@ -134,6 +134,7 @@ def _reference_index_topk(
     topk: int,
     init_blocks: int,
     local_blocks: int,
+    sm_scale: float = 1.0,
 ) -> torch.Tensor:
     total_q, num_idx_heads, _ = idx_q.shape
     out = torch.full(
@@ -149,7 +150,7 @@ def _reference_index_topk(
         num_blocks = (seq_len + BLOCK_SIZE - 1) // BLOCK_SIZE
         pages = block_table[req_id, :num_blocks]
         k = index_kv_cache[pages].reshape(num_blocks * BLOCK_SIZE, -1)
-        score = torch.einsum("qhd,kd->hqk", q.float(), k.float())
+        score = torch.einsum("qhd,kd->hqk", q.float(), k.float()) * sm_scale
 
         q_pos = prefix_len + torch.arange(q_len, device=idx_q.device)
         k_pos = torch.arange(k.shape[0], device=idx_q.device)
@@ -244,6 +245,270 @@ def test_prefill_index_topk_correctness():
     _assert_topk_indices_equal_unordered(actual, expected)
 
 
+# MSA indexer (SM100): fmha_sm100 OnlyScore for the per-block scores, then the
+# Triton minimax_m3_index_topk for selection (no sparse_topk_select). Uses a
+# deterministic construction (idx_q == 1, distinct e4m3-exact per-block values)
+# so scores are strictly monotonic in the block id -> exact top-k agreement.
+def _fmha_indexer_topk(
+    idx_q: torch.Tensor,  # [total_q, H, 128] bf16/e4m3
+    index_cache: torch.Tensor,  # [num_pages, 128, 128] bf16/e4m3
+    block_table: torch.Tensor,
+    q_lens: torch.Tensor,
+    seq_lens: torch.Tensor,
+    prefix_lens: torch.Tensor,
+    sm_scale: float,
+    topk: int,
+) -> torch.Tensor:
+    """Replicate MiniMaxM3IndexerMSAImpl's score path (single decode/prefill side)."""
+    from vllm.third_party.fmha_sm100.api import _fmha_sm100, _fmha_sm100_plan
+
+    num_idx_heads, head_dim = idx_q.shape[1], idx_q.shape[2]
+    nvp = [(s + 127) // 128 for s in seq_lens.tolist()]
+    kv_indices = torch.cat([block_table[r, : nvp[r]] for r in range(len(nvp))]).to(
+        torch.int32
+    )
+
+    qo = q_lens.cpu().to(torch.int32)
+    kv = seq_lens.cpu().to(torch.int32)
+    plan = _fmha_sm100_plan(
+        qo,
+        kv,
+        num_idx_heads,
+        num_kv_heads=1,
+        qo_offset=kv - qo,
+        page_size=128,
+        output_maxscore=True,
+        causal=True,
+        num_kv_splits=1,
+    )
+    k_pages = index_cache.view(index_cache.shape[0], 1, 128, head_dim)
+    _, max_score = _fmha_sm100(
+        idx_q,
+        k_pages,
+        k_pages,
+        plan,
+        kv_indices=kv_indices,
+        output_o=False,
+        output_maxscore=True,
+        sm_scale=sm_scale,
+    )
+
+    batch = q_lens.numel()
+    cu = torch.zeros(batch + 1, dtype=torch.int32, device=idx_q.device)
+    cu[1:] = q_lens.to(torch.int32).cumsum(0)
+    # max_score [H, k_tiles, total_q] -> transpose to [H, total_q, k_tiles].
+    return minimax_m3_index_topk(
+        max_score.transpose(1, 2),
+        cu,
+        prefix_lens.to(torch.int32),
+        int(q_lens.max()),
+        topk,
+        0,  # init_blocks
+        0,  # local_blocks
+    )
+
+
+# e4m3-exact, strictly-increasing per-block values: with idx_q == 1 (also exact)
+# the per-block scores are exact and distinct in BOTH bf16 and e4m3, so the fp8
+# score path selects the same top-k as the reference (no quantization ties).
+_E4M3_EXACT_VALUES = [
+    *range(1, 17),  # 1..16  (step 1)
+    *range(18, 33, 2),  # 18..32 (step 2)
+    *range(36, 65, 4),  # 36..64 (step 4)
+    *range(72, 129, 8),  # 72..128 (step 8)
+]
+
+
+@pytest.mark.skipif(
+    not current_platform.is_device_capability_family(100),
+    reason="fmha_sm100 indexer requires SM100 (Blackwell).",
+)
+@pytest.mark.parametrize("index_dtype", [torch.bfloat16, torch.float8_e4m3fn])
+@pytest.mark.parametrize(
+    ("q_lens", "prefix_lens"),
+    [
+        ((4, 3), (2048, 2560)),  # prefill: every token sees >= 16 causal blocks
+        ((1, 1, 1), (2048, 3000, 4096)),  # decode: one query token per request
+    ],
+)
+def test_fmha_sm100_indexer_matches_reference(q_lens, prefix_lens, index_dtype):
+    torch.manual_seed(0)
+    num_idx_heads, head_dim = 4, HEAD_DIM
+    device = "cuda"
+
+    q_lens_t = torch.tensor(q_lens, device=device, dtype=torch.int32)
+    prefix_lens_t = torch.tensor(prefix_lens, device=device, dtype=torch.int32)
+    seq_lens = prefix_lens_t + q_lens_t
+    batch = len(q_lens)
+    max_blocks = (int(seq_lens.max()) + BLOCK_SIZE - 1) // BLOCK_SIZE
+    assert max_blocks <= len(_E4M3_EXACT_VALUES)
+    num_pages = batch * max_blocks
+    block_table = torch.randperm(num_pages, device=device, dtype=torch.int32).reshape(
+        batch, max_blocks
+    )
+
+    idx_q = torch.ones(
+        int(q_lens_t.sum()), num_idx_heads, head_dim, device=device, dtype=index_dtype
+    )
+    index_cache = torch.empty(
+        num_pages, BLOCK_SIZE, head_dim, device=device, dtype=index_dtype
+    )
+    for r in range(batch):
+        for b in range(max_blocks):
+            index_cache[block_table[r, b]] = float(_E4M3_EXACT_VALUES[b])
+
+    sm_scale = head_dim**-0.5
+    actual = _fmha_indexer_topk(
+        idx_q,
+        index_cache,
+        block_table,
+        q_lens_t,
+        seq_lens,
+        prefix_lens_t,
+        sm_scale,
+        TOPK,
+    )
+    expected = _reference_index_topk(
+        idx_q,
+        index_cache,
+        block_table,
+        q_lens_t,
+        seq_lens,
+        prefix_lens_t,
+        TOPK,
+        init_blocks=0,
+        local_blocks=0,
+        sm_scale=sm_scale,
+    )
+    _assert_topk_indices_equal_unordered(actual, expected)
+
+
+# Full impl-level parity: drive both MiniMaxM3IndexerMSAImpl (fmha_sm100 score +
+# Triton top-k) and MiniMaxM3IndexerTritonImpl through their real metadata
+# builders on the SAME CommonAttentionMetadata + index cache, and assert the
+# selected blocks agree. This exercises all the metadata the impl/kernels consume
+# (decode/prefill split, cu_seqlens_q rebasing, prefix_lens, kv_indices gather,
+# decode_pages split) -- a metadata bug on either side shifts the causal window
+# or the block->page mapping and breaks the comparison.
+@pytest.mark.skipif(
+    not current_platform.is_device_capability_family(100),
+    reason="fmha_sm100 indexer requires SM100 (Blackwell).",
+)
+@pytest.mark.parametrize("topk", [8, 16])
+def test_msa_indexer_impl_matches_triton(topk, monkeypatch):
+    import vllm.models.minimax_m3.common.indexer as indexer_mod
+    from tests.v1.attention.utils import (
+        BatchSpec,
+        create_common_attn_metadata,
+        create_vllm_config,
+    )
+    from vllm.config import set_current_vllm_config
+    from vllm.forward_context import set_forward_context
+    from vllm.models.minimax_m3.common.indexer import (
+        MiniMaxM3IndexerTritonImpl,
+        MiniMaxM3IndexerTritonMetadataBuilder,
+    )
+    from vllm.models.minimax_m3.nvidia.indexer_msa import (
+        MiniMaxM3IndexerMSAImpl,
+        MiniMaxM3IndexerMSAMetadataBuilder,
+    )
+
+    torch.manual_seed(0)
+    device = torch.device("cuda")
+    num_idx_heads, head_dim = 4, HEAD_DIM
+    # TP=1: avoid requiring an initialized distributed group in a unit test.
+    monkeypatch.setattr(indexer_mod, "get_tensor_model_parallel_world_size", lambda: 1)
+
+    vllm_config = create_vllm_config(
+        block_size=BLOCK_SIZE, max_model_len=8192, max_num_batched_tokens=8192
+    )
+    vllm_config.model_config.hf_config.sparse_attention_config = {
+        "sparse_num_index_heads": num_idx_heads
+    }
+
+    # Decode-first mixed batch: 2 decode reqs (q_len 1) then 2 prefill reqs. Long
+    # prefixes so every token sees > TOPK causal blocks (non-trivial selection).
+    batch = BatchSpec(seq_lens=[2305, 2561, 2624, 2720], query_lens=[1, 1, 64, 96])
+    common = create_common_attn_metadata(
+        batch, BLOCK_SIZE, device, arange_block_indices=True
+    )
+    num_tokens = batch.compute_num_tokens()
+
+    # Deterministic index cache: distinct, monotonic per-logical-block values so
+    # the top-k is unambiguous (both kernels pick the same blocks, no fp ties).
+    block_table = common.block_table_tensor
+    num_pages = int(block_table.max().item()) + 1
+    index_cache = torch.zeros(
+        num_pages, BLOCK_SIZE, head_dim, device=device, dtype=DTYPE
+    )
+    for r, seq_len in enumerate(batch.seq_lens):
+        for b in range((seq_len + BLOCK_SIZE - 1) // BLOCK_SIZE):
+            index_cache[block_table[r, b]] = float(b + 1)
+    index_q = torch.ones(
+        num_tokens, num_idx_heads * head_dim, device=device, dtype=DTYPE
+    )
+
+    spec = MLAAttentionSpec(
+        block_size=BLOCK_SIZE, num_kv_heads=1, head_size=head_dim, dtype=DTYPE
+    )
+    impl_kwargs = dict(
+        num_kv_heads=num_idx_heads,
+        scale=head_dim**-0.5,
+        topk_blocks=topk,
+        sparse_block_size=BLOCK_SIZE,
+        num_index_heads=num_idx_heads,
+        index_head_dim=head_dim,
+        init_blocks=0,
+        local_blocks=0,
+    )
+
+    with set_current_vllm_config(vllm_config):
+        msa_impl = MiniMaxM3IndexerMSAImpl(prefix="idx_msa", **impl_kwargs)
+        triton_impl = MiniMaxM3IndexerTritonImpl(prefix="idx_triton", **impl_kwargs)
+        msa_builder = MiniMaxM3IndexerMSAMetadataBuilder(
+            spec, [msa_impl.index_cache.prefix], vllm_config, device
+        )
+        triton_builder = MiniMaxM3IndexerTritonMetadataBuilder(
+            spec, [triton_impl.index_cache.prefix], vllm_config, device
+        )
+
+    # Both impls score against the same index keys.
+    msa_impl.index_cache.kv_cache = index_cache
+    triton_impl.index_cache.kv_cache = index_cache
+
+    # Exercise the shared persistent top-k buffer for BOTH impls: each must write
+    # decode ([:, :nd]) and prefill ([:, nd:]) into its buffer and return views.
+    # Separate buffers so the two forwards don't clobber each other.
+    nd = sum(q for q in batch.query_lens if q <= 1)
+    msa_impl.topk_indices_buffer = torch.full(
+        (num_idx_heads, num_tokens, topk), -2, dtype=torch.int32, device=device
+    )
+    triton_impl.topk_indices_buffer = torch.full(
+        (num_idx_heads, num_tokens, topk), -2, dtype=torch.int32, device=device
+    )
+
+    attn_metadata = {
+        msa_impl.index_cache.prefix: msa_builder.build(0, common),
+        triton_impl.index_cache.prefix: triton_builder.build(0, common),
+    }
+    with set_forward_context(attn_metadata, vllm_config):
+        msa_decode, msa_prefill = msa_impl(index_q)
+        tri_decode, tri_prefill = triton_impl(index_q)
+
+    assert msa_decode is not None and tri_decode is not None
+    assert msa_prefill is not None and tri_prefill is not None
+    _assert_topk_indices_equal_unordered(msa_decode, tri_decode)
+    _assert_topk_indices_equal_unordered(msa_prefill, tri_prefill)
+    # decode/prefill outputs are views into each impl's persistent buffer.
+    for impl, dec, pre in (
+        (msa_impl, msa_decode, msa_prefill),
+        (triton_impl, tri_decode, tri_prefill),
+    ):
+        buf = impl.topk_indices_buffer
+        assert dec.data_ptr() == buf[:, :nd, :].data_ptr()
+        assert pre.data_ptr() == buf[:, nd:, :].data_ptr()
+
+
 @pytest.mark.parametrize(
     ("decode_query_len", "max_decode_query_len"),
     [
@@ -317,6 +582,65 @@ def test_decode_index_topk_correctness(
     _assert_topk_indices_equal_unordered(actual, expected)
 
 
+@pytest.mark.skipif(
+    not current_platform.is_device_capability_family(100),
+    reason="fp8 e4m3 indexer cache is the SM100 (MSA) path.",
+)
+@pytest.mark.parametrize("num_idx_heads", [1, 4])
+def test_decode_index_topk_fp8(num_idx_heads: int):
+    """The fp8 (e4m3) indexer cache feeds the Triton decode kernel on the MSA
+    path. The kernel must score in fp32 (no scaling) so its top-k matches a
+    reference computed from the dequantized fp8 values."""
+    torch.manual_seed(0)
+    topk, init_blocks, local_blocks, head_dim = 8, 0, 1, 128
+    decode_query_len = 1
+    active_seq_lens = torch.tensor((129, 1025, 4097), device="cuda", dtype=torch.int32)
+    q_lens = torch.full_like(active_seq_lens, decode_query_len)
+    prefix_lens = active_seq_lens - decode_query_len
+    batch = active_seq_lens.numel()
+    max_seq_len = int(active_seq_lens.max())
+    max_blocks = (max_seq_len + BLOCK_SIZE - 1) // BLOCK_SIZE
+    num_pages = batch * max_blocks
+    block_table = torch.randperm(num_pages, device="cuda", dtype=torch.int32).reshape(
+        batch, max_blocks
+    )
+    idx_q = torch.randn(
+        batch * decode_query_len, num_idx_heads, head_dim, device="cuda"
+    ).to(torch.float8_e4m3fn)
+    index_kv_cache = torch.randn(num_pages, BLOCK_SIZE, head_dim, device="cuda").to(
+        torch.float8_e4m3fn
+    )
+
+    actual = minimax_m3_index_decode(
+        idx_q,
+        index_kv_cache,
+        block_table,
+        active_seq_lens,
+        max_seq_len=max_seq_len,
+        topk=topk,
+        init_blocks=init_blocks,
+        local_blocks=local_blocks,
+        num_kv_heads=num_idx_heads,
+        sm_scale=head_dim**-0.5,
+        decode_query_len=decode_query_len,
+    )
+    # Reference from the DEQUANTIZED fp8 values (the kernel computes the fp8 QK
+    # in fp32, so it must match an fp32 matmul of the same e4m3 values).
+    expected = _reference_index_topk(
+        idx_q.float(),
+        index_kv_cache.float(),
+        block_table,
+        q_lens,
+        active_seq_lens,
+        prefix_lens,
+        topk,
+        init_blocks,
+        local_blocks,
+        head_dim**-0.5,
+    )
+    _assert_topk_indices_equal_unordered(actual, expected)
+
+
 # Sparse attention kernels.
 def _reference_sparse_attn(
     q: torch.Tensor,

tests/kernels/attention/test_mla_cross_layer_kernel_equivalence.py

@@ -1,566 +0,0 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
-"""Bit-exact kernel equivalence for MLA decode/write kernels on the
-cross-layer (block-major) KV cache layout.
-
-The cross-layer layout carves each layer's per-block page out of a single
-unified slot, so the per-layer view has an inflated ``stride(0)`` (the full
-unified slot) and a non-zero storage offset. These tests confirm the MLA
-kernels behind the backends that opt in to the layout (FlashMLA dense,
-FlashInfer MLA dense, FlashMLA fp8 sparse, plus the ``concat_and_cache_mla``
-write) honor that strided view bit-identically to a contiguous per-layer
-cache, and that writes do not bleed into neighbouring layers' segments.
-"""
-
-import pytest
-import torch
-
-pytestmark = pytest.mark.skipif(
-    not torch.cuda.is_available(), reason="MLA cache kernels require CUDA"
-)
-
-
-def test_concat_and_cache_mla_into_unified_slot_view():
-    """concat_and_cache_mla must write correctly into a per-layer view whose
-    block stride is the full unified slot (block-major), with zero bleed into
-    the other layers' segments of the same slot."""
-    from vllm import _custom_ops as ops
-
-    torch.manual_seed(0)
-    dev = "cuda"
-    kv_lora_rank = 512
-    pe = 64
-    entry = kv_lora_rank + pe
-    page = 64
-    num_blocks = 32
-    ntok = 200
-
-    kv_c = torch.randn(ntok, kv_lora_rank, device=dev, dtype=torch.bfloat16)
-    k_pe = torch.randn(ntok, pe, device=dev, dtype=torch.bfloat16)
-    slot = torch.randperm(num_blocks * page, device=dev, dtype=torch.int64)[:ntok]
-    scale = torch.tensor(1.0, device=dev)
-
-    def write(cache):
-        ops.concat_and_cache_mla(kv_c, k_pe, cache, slot, "auto", scale)
-
-    # Contiguous per-layer reference: (num_blocks, page, entry).
-    ref = torch.zeros(num_blocks, page, entry, device=dev, dtype=torch.bfloat16)
-    write(ref)
-
-    # Unified slot holding three layer pages per block. Carve the middle
-    # layer's view (non-zero offset, block stride == full unified slot).
-    layer_page_elems = page * entry
-    n_layers = 3
-    unified_slot_elems = n_layers * layer_page_elems
-    big = torch.zeros(num_blocks, unified_slot_elems, device=dev, dtype=torch.bfloat16)
-    flat = big.view(-1)
-    offset = layer_page_elems  # middle layer
-    view = torch.as_strided(
-        flat,
-        size=(num_blocks, page, entry),
-        stride=(unified_slot_elems, entry, 1),
-        storage_offset=offset,
-    )
-    assert not view.is_contiguous()
-    assert view.stride(0) == unified_slot_elems
-    write(view)
-
-    # Bit-exact equivalence and zero bleed into the neighbour segments.
-    max_diff = (ref.float() - view.float()).abs().max().item()
-    assert max_diff == 0.0, f"max|Δ| = {max_diff}"
-
-    neighbour_lo = torch.as_strided(
-        flat, (num_blocks, layer_page_elems), (unified_slot_elems, 1), 0
-    )
-    neighbour_hi = torch.as_strided(
-        flat,
-        (num_blocks, layer_page_elems),
-        (unified_slot_elems, 1),
-        2 * layer_page_elems,
-    )
-    assert neighbour_lo.abs().max().item() == 0.0
-    assert neighbour_hi.abs().max().item() == 0.0
-
-
-def test_flashmla_dense_decode_unified_slot_view():
-    """FlashMLA dense decode (FLASHMLA backend, e.g. Kimi-K2-style dense MLA
-    on Hopper) must read a unified-slot block-major view bit-identically to a
-    contiguous per-layer cache."""
-    import vllm.v1.attention.ops.flashmla as fm
-
-    ok, reason = fm.is_flashmla_dense_supported()
-    if not ok:
-        pytest.skip(reason)
-
-    torch.manual_seed(0)
-    dev = "cuda"
-    dt = torch.bfloat16
-    head_dim = 576
-    hdv = 512
-    h_q = 128
-    page = 64
-    num_blocks = 64
-    bs = 4
-    n_layers = 3
-    layer = 1
-
-    q = torch.randn(bs, 1, h_q, head_dim, device=dev, dtype=dt) * 0.1
-    kv_data = torch.randn(num_blocks, page, 1, head_dim, device=dev, dtype=dt) * 0.1
-
-    # (A) contiguous per-layer reference.
-    cache_contiguous = kv_data.clone().contiguous()
-
-    # (B) unified slot: view one layer -> inflated stride(0), non-zero offset.
-    unified = (
-        torch.randn(num_blocks, n_layers, page, 1, head_dim, device=dev, dtype=dt) * 0.1
-    )
-    unified[:, layer].copy_(kv_data)
-    cache_view = unified[:, layer]
-    assert not cache_view.is_contiguous()
-    assert cache_view.stride(0) == n_layers * page * 1 * head_dim
-
-    max_blk = num_blocks // bs
-    block_table = torch.arange(num_blocks, device=dev, dtype=torch.int32).view(
-        bs, max_blk
-    )
-    cache_seqlens = torch.full((bs,), max_blk * page, device=dev, dtype=torch.int32)
-
-    def run(kc):
-        meta, num_splits = fm.get_mla_metadata()
-        out, _ = fm.flash_mla_with_kvcache(
-            q=q,
-            k_cache=kc,
-            block_table=block_table,
-            cache_seqlens=cache_seqlens,
-            head_dim_v=hdv,
-            tile_scheduler_metadata=meta,
-            num_splits=num_splits,
-            softmax_scale=head_dim**-0.5,
-            causal=True,
-        )
-        return out.clone().float()
-
-    out_ref = run(cache_contiguous)
-    out_view = run(cache_view)
-    assert torch.isfinite(out_ref).all()
-    assert out_ref.abs().max().item() > 0.0
-    assert (out_ref - out_view).abs().max().item() == 0.0
-
-
-def test_flashinfer_mla_dense_decode_unified_slot_view():
-    """FlashInfer MLA dense decode must read a unified-slot block-major view
-    (inflated stride(0), non-zero storage offset) bit-identically to a
-    contiguous per-layer cache."""
-    try:
-        from flashinfer.decode import trtllm_batch_decode_with_kv_cache_mla
-    except ImportError:
-        pytest.skip("flashinfer is not available")
-    from vllm.platforms import current_platform
-
-    if not current_platform.is_device_capability_family(100):
-        pytest.skip("FlashInfer trtllm-gen MLA requires sm100")
-
-    torch.manual_seed(0)
-    dev = "cuda"
-    dt = torch.bfloat16
-    kv_lora_rank = 512
-    qk_rope_head_dim = 64
-    qk_nope_head_dim = 128
-    head_dim = kv_lora_rank + qk_rope_head_dim  # 576
-    num_qo_heads = 128
-    page = 64
-    num_blocks = 64
-    bs = 4
-    n_layers = 3  # >1 so the per-layer view's block stride is inflated.
-    layer = 1
-
-    q = torch.randn(bs, 1, num_qo_heads, head_dim, device=dev, dtype=dt)
-    kv_data = torch.randn(num_blocks, 1, page, head_dim, device=dev, dtype=dt)
-
-    # (A) contiguous per-layer reference.
-    kv_contiguous = kv_data.clone().contiguous()
-
-    # (B) unified slot: block b of every layer packed together; view one layer
-    # -> stride(0) is n_layers x larger and storage offset is non-zero.
-    unified = torch.randn(num_blocks, n_layers, 1, page, head_dim, device=dev, dtype=dt)
-    unified[:, layer].copy_(kv_data)
-    kv_view = unified[:, layer]
-    assert not kv_view.is_contiguous()
-    assert kv_view.stride(0) == n_layers * 1 * page * head_dim
-
-    max_blk = num_blocks // bs
-    block_tables = torch.arange(num_blocks, device=dev, dtype=torch.int32).view(
-        bs, max_blk
-    )
-    seq_lens = torch.full((bs,), max_blk * page, device=dev, dtype=torch.int32)
-    ws = torch.empty(128 * 1024 * 1024, dtype=torch.int8, device=dev)
-    scale = head_dim**-0.5
-
-    def run(kv):
-        return trtllm_batch_decode_with_kv_cache_mla(
-            query=q,
-            kv_cache=kv,
-            workspace_buffer=ws,
-            qk_nope_head_dim=qk_nope_head_dim,
-            kv_lora_rank=kv_lora_rank,
-            qk_rope_head_dim=qk_rope_head_dim,
-            block_tables=block_tables,
-            seq_lens=seq_lens,
-            max_seq_len=int(seq_lens.max().item()),
-            bmm1_scale=scale,
-            bmm2_scale=1.0,
-        ).clone()
-
-    out_ref = run(kv_contiguous).float()
-    out_view = run(kv_view).float()
-    assert torch.isfinite(out_ref).all()
-    assert (out_ref - out_view).abs().max().item() == 0.0
-
-
-def test_flashmla_fp8_sparse_decode_unified_slot_view():
-    """FlashMLA fp8 sparse decode (DeepSeek V3.2/V4 DSA path) must read a
-    unified-slot block-major view bit-identically to a contiguous fp8_ds_mla
-    cache, with finite nonzero output."""
-    import vllm.v1.attention.ops.flashmla as fm
-
-    ok, reason = fm.is_flashmla_sparse_supported()
-    if not ok:
-        pytest.skip(reason)
-
-    torch.manual_seed(0)
-    dev = "cuda"
-    entry = 656  # fp8_ds_mla bytes per token
-    page = 64
-    num_blocks = 32
-    h_q = 128
-    head_dim = 576
-    hdv = 512
-    batch = 2
-    topk = 128
-    n_layers = 3
-    layer = 1
-
-    q = torch.randn(batch, 1, h_q, head_dim, device=dev, dtype=torch.bfloat16) * 0.1
-
-    # Structurally valid fp8 ds_mla payload: 512B fp8 + 16B f32 scales + 128B
-    # bf16 rope (random bytes corrupt the scale region and yield NaNs).
-    nope = (torch.randn(num_blocks, page, 1, 512, device=dev) * 0.1).to(
-        torch.float8_e4m3fn
-    )
-    scales = torch.ones(num_blocks, page, 1, 4, device=dev, dtype=torch.float32)
-    rope = (torch.randn(num_blocks, page, 1, 64, device=dev) * 0.1).to(torch.bfloat16)
-    payload = torch.cat(
-        [
-            nope.view(torch.uint8).view(num_blocks, page, 1, 512),
-            scales.view(torch.uint8).view(num_blocks, page, 1, 16),
-            rope.view(torch.uint8).view(num_blocks, page, 1, 128),
-        ],
-        dim=-1,
-    ).contiguous()
-    assert payload.shape[-1] == entry and payload.dtype == torch.uint8
-
-    # (A) contiguous reference.
-    cache_contiguous = payload.clone().contiguous()
-
-    # (B) unified slot: view one layer -> inflated stride(0), non-zero offset.
-    unified = torch.randint(
-        0, 256, (num_blocks, n_layers, page, 1, entry), device=dev, dtype=torch.uint8
-    )
-    unified[:, layer].copy_(payload)
-    cache_view = unified[:, layer]
-    assert not cache_view.is_contiguous()
-    assert cache_view.stride(0) == n_layers * page * 1 * entry
-
-    # Sparse indices: each batch uses its own disjoint blocks.
-    blocks_per_batch = num_blocks // batch
-    idx = torch.full((batch, 1, topk), -1, device=dev, dtype=torch.int32)
-    for b in range(batch):
-        slots: list[int] = []
-        for blk in range(b * blocks_per_batch, (b + 1) * blocks_per_batch):
-            slots.extend(blk * page + off for off in range(page))
-        slots_t = torch.tensor(slots[:topk], device=dev, dtype=torch.int32)
-        idx[b, 0, : slots_t.numel()] = slots_t
-
-    def run(kc):
-        meta, num_splits = fm.get_mla_metadata()
-        out, _ = fm.flash_mla_with_kvcache(
-            q=q,
-            k_cache=kc,
-            block_table=None,
-            cache_seqlens=None,
-            head_dim_v=hdv,
-            tile_scheduler_metadata=meta,
-            is_fp8_kvcache=True,
-            indices=idx,
-            softmax_scale=head_dim**-0.5,
-        )
-        return out.clone().float()
-
-    out_ref = run(cache_contiguous)
-    out_view = run(cache_view)
-    assert torch.isfinite(out_ref).all()
-    assert out_ref.abs().max().item() > 0.0
-    assert (out_ref - out_view).abs().max().item() == 0.0
-
-
-def test_indexer_k_quant_and_cache_into_unified_slot_view():
-    """indexer_k_quant_and_cache (DeepSeek V3.2/V4 DSA indexer K write) must
-    write correctly into a per-layer view whose block stride is the full
-    unified slot, with zero bleed into the other layers' segments."""
-    from vllm import _custom_ops as ops
-
-    torch.manual_seed(0)
-    dev = "cuda"
-    head_dim = 128
-    quant_block_size = 128
-    block_size = 64
-    num_blocks = 16
-    ntok = 100
-    # Indexer cache layout per token: head_dim fp8 bytes followed by
-    # head_dim * 4 / quant_block_size scale bytes.
-    cache_stride = head_dim + head_dim * 4 // quant_block_size
-
-    k = torch.randn(ntok, head_dim, device=dev, dtype=torch.bfloat16)
-    slot = torch.randperm(num_blocks * block_size, device=dev, dtype=torch.int64)[:ntok]
-
-    def write(cache):
-        ops.indexer_k_quant_and_cache(k, cache, slot, quant_block_size, "ue8m0")
-
-    # Contiguous per-layer reference.
-    ref = torch.zeros(
-        num_blocks, block_size, cache_stride, device=dev, dtype=torch.uint8
-    )
-    write(ref)
-
-    # Unified slot holding three layer pages per block; carve the middle one.
-    n_layers = 3
-    layer = 1
-    unified = torch.zeros(
-        num_blocks, n_layers, block_size, cache_stride, device=dev, dtype=torch.uint8
-    )
-    view = unified[:, layer]
-    assert not view.is_contiguous()
-    assert view.stride(0) == n_layers * block_size * cache_stride
-    write(view)
-
-    assert torch.equal(ref, view.contiguous())
-    # Zero bleed into the neighbour layers' segments.
-    assert unified[:, 0].abs().max().item() == 0
-    assert unified[:, 2].abs().max().item() == 0
-
-
-def test_flashattn_mla_dense_decode_unified_slot_view():
-    """FA3 decode (FLASH_ATTN_MLA backend) must read a unified-slot
-    block-major view bit-identically to a contiguous per-layer cache."""
-    try:
-        from vllm.vllm_flash_attn import flash_attn_varlen_func
-    except ImportError:
-        pytest.skip("vllm_flash_attn is not available")
-    from vllm.v1.attention.backends.fa_utils import flash_attn_supports_mla
-
-    if not flash_attn_supports_mla():
-        pytest.skip("FA3 MLA requires a Hopper device")
-
-    torch.manual_seed(0)
-    dev = "cuda"
-    dt = torch.bfloat16
-    kv_lora_rank = 512
-    rope_dim = 64
-    entry = kv_lora_rank + rope_dim  # 576
-    h_q = 16
-    page = 64
-    num_blocks = 64
-    bs = 4
-    n_layers = 3
-    layer = 1
-
-    q_pe = torch.randn(bs, h_q, rope_dim, device=dev, dtype=dt) * 0.1
-    q_nope = torch.randn(bs, h_q, kv_lora_rank, device=dev, dtype=dt) * 0.1
-    kv_data = torch.randn(num_blocks, page, entry, device=dev, dtype=dt) * 0.1
-
-    # (A) contiguous per-layer reference.
-    cache_contiguous = kv_data.clone().contiguous()
-
-    # (B) unified slot: view one layer -> inflated stride(0), non-zero offset.
-    unified = torch.randn(num_blocks, n_layers, page, entry, device=dev, dtype=dt) * 0.1
-    unified[:, layer].copy_(kv_data)
-    cache_view = unified[:, layer]
-    assert not cache_view.is_contiguous()
-    assert cache_view.stride(0) == n_layers * page * entry
-
-    max_blk = num_blocks // bs
-    block_table = torch.arange(num_blocks, device=dev, dtype=torch.int32).view(
-        bs, max_blk
-    )
-    seq_lens = torch.full((bs,), max_blk * page, device=dev, dtype=torch.int32)
-    cu_seqlens_q = torch.arange(bs + 1, device=dev, dtype=torch.int32)
-
-    def run(cache):
-        kv_c_cache = cache[..., :kv_lora_rank]
-        k_pe_cache = cache[..., kv_lora_rank:]
-        out = flash_attn_varlen_func(
-            q=q_pe,
-            k=k_pe_cache.unsqueeze(-2),  # Add head dim of 1
-            v=kv_c_cache.unsqueeze(-2),  # Add head dim of 1
-            q_v=q_nope,
-            max_seqlen_q=1,
-            cu_seqlens_q=cu_seqlens_q,
-            max_seqlen_k=int(seq_lens.max().item()),
-            seqused_k=seq_lens,
-            block_table=block_table,
-            softmax_scale=entry**-0.5,
-            causal=True,
-            fa_version=3,
-        )
-        return out.clone().float()
-
-    out_ref = run(cache_contiguous)
-    out_view = run(cache_view)
-    assert torch.isfinite(out_ref).all()
-    assert out_ref.abs().max().item() > 0.0
-    assert (out_ref - out_view).abs().max().item() == 0.0
-
-
-def test_flashmla_dense_fp8_decode_unified_slot_view():
-    """FlashMLA dense fp8 decode (FLASHMLA backend with quantized KV cache)
-    must read a unified-slot block-major view bit-identically to a contiguous
-    per-layer fp8 cache."""
-    import vllm.v1.attention.ops.flashmla as fm
-
-    ok, reason = fm.is_flashmla_dense_supported()
-    if not ok:
-        pytest.skip(reason)
-
-    torch.manual_seed(0)
-    dev = "cuda"
-    head_dim = 576
-    hdv = 512
-    h_q = 128
-    page = 64
-    num_blocks = 64
-    bs = 4
-    n_layers = 3
-    layer = 1
-
-    q = torch.randn(bs, 1, h_q, head_dim, device=dev, dtype=torch.bfloat16) * 0.1
-    kv_data = (torch.randn(num_blocks, page, head_dim, device=dev) * 0.1).to(
-        torch.float8_e4m3fn
-    )
-
-    # (A) contiguous per-layer reference.
-    cache_contiguous = kv_data.clone().contiguous()
-
-    # (B) unified slot: view one layer -> inflated stride(0), non-zero offset.
-    unified = (torch.randn(num_blocks, n_layers, page, head_dim, device=dev) * 0.1).to(
-        torch.float8_e4m3fn
-    )
-    unified[:, layer].copy_(kv_data)
-    cache_view = unified[:, layer]
-    assert not cache_view.is_contiguous()
-    assert cache_view.stride(0) == n_layers * page * head_dim
-
-    max_blk = num_blocks // bs
-    block_table = torch.arange(num_blocks, device=dev, dtype=torch.int32).view(
-        bs, max_blk
-    )
-    cache_seqlens = torch.full((bs,), max_blk * page, device=dev, dtype=torch.int32)
-    descale = torch.ones(1, device=dev, dtype=torch.float32)
-
-    def run(kc):
-        tile_md, num_splits = fm.get_mla_metadata_dense_fp8(cache_seqlens, h_q, 1)
-        out, _ = fm.flash_mla_with_kvcache_fp8(
-            q=q,
-            k_cache=kc.unsqueeze(-2),  # Add head dim of 1
-            block_table=block_table,
-            cache_seqlens=cache_seqlens,
-            head_dim_v=hdv,
-            tile_scheduler_metadata=tile_md,
-            num_splits=num_splits,
-            softmax_scale=head_dim**-0.5,
-            causal=True,
-            descale_q=descale,
-            descale_k=descale,
-        )
-        return out.clone().float()
-
-    out_ref = run(cache_contiguous)
-    out_view = run(cache_view)
-    assert torch.isfinite(out_ref).all()
-    assert out_ref.abs().max().item() > 0.0
-    assert (out_ref - out_view).abs().max().item() == 0.0
-
-
-def test_flashinfer_mla_dense_fp8_decode_unified_slot_view():
-    """FlashInfer MLA dense decode with an fp8 KV cache must read a
-    unified-slot block-major view bit-identically to a contiguous per-layer
-    cache."""
-    try:
-        from flashinfer.decode import trtllm_batch_decode_with_kv_cache_mla
-    except ImportError:
-        pytest.skip("flashinfer is not available")
-    from vllm.platforms import current_platform
-
-    if not current_platform.is_device_capability_family(100):
-        pytest.skip("FlashInfer trtllm-gen MLA requires sm100")
-
-    torch.manual_seed(0)
-    dev = "cuda"
-    kv_lora_rank = 512
-    qk_rope_head_dim = 64
-    qk_nope_head_dim = 128
-    head_dim = kv_lora_rank + qk_rope_head_dim  # 576
-    num_qo_heads = 128
-    page = 64
-    num_blocks = 64
-    bs = 4
-    n_layers = 3
-    layer = 1
-
-    # With a quantized KV cache the decode query is quantized to fp8 as well
-    # (trtllm-gen has no bf16-query x fp8-cache decode kernel).
-    q = (torch.randn(bs, 1, num_qo_heads, head_dim, device=dev) * 0.1).to(
-        torch.float8_e4m3fn
-    )
-    kv_data = (torch.randn(num_blocks, 1, page, head_dim, device=dev) * 0.1).to(
-        torch.float8_e4m3fn
-    )
-
-    # (A) contiguous per-layer reference.
-    kv_contiguous = kv_data.clone().contiguous()
-
-    # (B) unified slot: view one layer -> inflated stride(0), non-zero offset.
-    unified = (
-        torch.randn(num_blocks, n_layers, 1, page, head_dim, device=dev) * 0.1
-    ).to(torch.float8_e4m3fn)
-    unified[:, layer].copy_(kv_data)
-    kv_view = unified[:, layer]
-    assert not kv_view.is_contiguous()
-    assert kv_view.stride(0) == n_layers * 1 * page * head_dim
-
-    max_blk = num_blocks // bs
-    block_tables = torch.arange(num_blocks, device=dev, dtype=torch.int32).view(
-        bs, max_blk
-    )
-    seq_lens = torch.full((bs,), max_blk * page, device=dev, dtype=torch.int32)
-    ws = torch.empty(128 * 1024 * 1024, dtype=torch.int8, device=dev)
-    scale = head_dim**-0.5
-
-    def run(kv):
-        return trtllm_batch_decode_with_kv_cache_mla(
-            query=q,
-            kv_cache=kv,
-            workspace_buffer=ws,
-            qk_nope_head_dim=qk_nope_head_dim,
-            kv_lora_rank=kv_lora_rank,
-            qk_rope_head_dim=qk_rope_head_dim,
-            block_tables=block_tables,
-            seq_lens=seq_lens,
-            max_seq_len=int(seq_lens.max().item()),
-            bmm1_scale=scale,
-            bmm2_scale=1.0,
-        ).clone()
-
-    out_ref = run(kv_contiguous).float()
-    out_view = run(kv_view).float()
-    assert torch.isfinite(out_ref).all()
-    assert (out_ref - out_view).abs().max().item() == 0.0

tests/kernels/attention/test_prefix_prefill.py

@@ -5,12 +5,10 @@ import math
 import random
 import time
 from collections.abc import Callable
-from contextlib import nullcontext
 
 import pytest
 import torch
 import torch.nn.functional as F
-from torch.nn.attention import SDPBackend, sdpa_kernel
 
 from vllm.platforms import current_platform
 from vllm.utils.torch_utils import STR_DTYPE_TO_TORCH_DTYPE, set_random_seed
@@ -559,21 +557,15 @@ def test_contexted_kv_attention_alibi(
             query_len, seq_len, alibi_slopes, device, dtype
         )
 
-        # Compute attention. On ROCm we force use of the Math SDPA backend rather than
-        # the Flash or Mem-Efficient backends for increased numerical accuracy
-        if current_platform.is_rocm():
-            sdpa_context = sdpa_kernel(SDPBackend.MATH)
-        else:
-            sdpa_context = nullcontext()
-        with sdpa_context:
-            out = F.scaled_dot_product_attention(
-                q_sdpa,
-                k_sdpa,
-                v_sdpa,
-                attn_mask=alibi_mask,
-                dropout_p=0.0,
-                scale=scale,
-            )
+        # Compute attention
+        out = F.scaled_dot_product_attention(
+            q_sdpa,
+            k_sdpa,
+            v_sdpa,
+            attn_mask=alibi_mask,
+            dropout_p=0.0,
+            scale=scale,
+        )
 
         # Reshape output back to [query_len, num_heads, head_size]
         out = out.view(num_heads, query_len, head_size).permute(1, 0, 2)

tests/kernels/attention/test_rocm_triton_attn_dsv4.py

@@ -90,9 +90,7 @@ def _ref_sparse_prefill_ragged(
     return out.to(torch.bfloat16)
 
 
-def _pack_fp8_ds_mla_cache(
-    kv: torch.Tensor, block_size: int, is_extra: bool = False
-) -> torch.Tensor:
+def _pack_fp8_ds_mla_cache(kv: torch.Tensor, block_size: int) -> torch.Tensor:
     assert kv.shape[-1] == HEAD_DIM
     num_tokens = kv.shape[0]
     num_blocks = (num_tokens + block_size - 1) // block_size
@@ -103,9 +101,7 @@ def _pack_fp8_ds_mla_cache(
     )
     cache_flat = cache.view(torch.uint8).flatten()
     kv_nope_fp8 = (
-        kv[:, :NOPE_HEAD_DIM]
-        .to(torch.float8_e4m3fn if is_extra else current_platform.fp8_dtype())
-        .view(torch.uint8)
+        kv[:, :NOPE_HEAD_DIM].to(current_platform.fp8_dtype()).view(torch.uint8)
     )
     kv_rope_u8 = kv[:, NOPE_HEAD_DIM:].contiguous().view(torch.uint8)
 
@@ -124,7 +120,7 @@ def _pack_fp8_ds_mla_cache(
 
 
 def _read_fp8_ds_mla_cache(
-    cache: torch.Tensor, slot: int, block_size: int, is_extra: bool = False
+    cache: torch.Tensor, slot: int, block_size: int
 ) -> torch.Tensor:
     cache_flat = cache.view(torch.uint8).flatten()
     block_idx = slot // block_size
@@ -133,9 +129,7 @@ def _read_fp8_ds_mla_cache(
     token_base = block_base + pos * 576
 
     nope_u8 = cache_flat[token_base : token_base + NOPE_HEAD_DIM]
-    nope = nope_u8.view(
-        torch.float8_e4m3fn if is_extra else current_platform.fp8_dtype()
-    ).to(torch.float32)
+    nope = nope_u8.view(current_platform.fp8_dtype()).to(torch.float32)
     rope_u8 = cache_flat[
         token_base + NOPE_HEAD_DIM : token_base + NOPE_HEAD_DIM + ROPE_HEAD_DIM * 2
     ]
@@ -163,9 +157,7 @@ def _ref_sparse_decode_ragged(
         ]
         if extra_cache is not None and extra_rows is not None:
             row_kv.extend(
-                _read_fp8_ds_mla_cache(
-                    extra_cache, int(slot), block_size, is_extra=True
-                )
+                _read_fp8_ds_mla_cache(extra_cache, int(slot), block_size)
                 for slot in extra_rows[query_idx]
             )
 
@@ -334,7 +326,7 @@ def test_sparse_attn_decode_ragged_kernel() -> None:
     main_kv = torch.randn(6, HEAD_DIM, dtype=torch.bfloat16, device=device) * 0.125
     extra_kv = torch.randn(5, HEAD_DIM, dtype=torch.bfloat16, device=device) * 0.125
     main_cache = _pack_fp8_ds_mla_cache(main_kv, block_size)
-    extra_cache = _pack_fp8_ds_mla_cache(extra_kv, block_size, is_extra=True)
+    extra_cache = _pack_fp8_ds_mla_cache(extra_kv, block_size)
     main_indices = torch.tensor([0, 2, 4, 1], dtype=torch.int32, device=device)
     main_indptr = torch.tensor([0, 2, 4], dtype=torch.int32, device=device)
     extra_indices = torch.tensor([1, 3, 0], dtype=torch.int32, device=device)
@@ -485,7 +477,7 @@ def test_sparse_attn_decode_split_k_kernel(
         rows = [[1, 3, 0, 5, 2, 4], [3, 0, 6]]
         extra_kv = torch.randn(7, HEAD_DIM, dtype=torch.bfloat16, device=device) * 0.125
         extra_rows = rows
-        extra_cache = _pack_fp8_ds_mla_cache(extra_kv, block_size, is_extra=True)
+        extra_cache = _pack_fp8_ds_mla_cache(extra_kv, block_size)
         extra_indices, extra_indptr = _ragged_from_rows(rows, device)
 
     attn_sink = (

tests/kernels/attention/test_triton_decode_attention.py

@@ -231,95 +231,3 @@ def test_decode_attention_fp8(B, L, H_Q, H_KV, D_QK, D_V, CACHE_SIZE, PAGE_SIZE)
 
     # FP8 tolerances match test_mla_backends.py test_backend_correctness.
     torch.testing.assert_close(o_ref, o_fp8, atol=5e-1, rtol=1e-2)
-
-
-@pytest.mark.parametrize(
-    "H_Q,H_KV,D_QK,D_V,is_mla",
-    [
-        (16, 1, 576, 512, True),  # MLA path (grouped kernel, v = trans(k))
-        (32, 8, 128, 128, False),  # GQA path (grouped kernel)
-        (32, 32, 128, 128, False),  # MHA path (normal kernel)
-    ],
-)
-@pytest.mark.parametrize("PAGE_SIZE", [16])
-def test_decode_attention_cross_layer_view(H_Q, H_KV, D_QK, D_V, is_mla, PAGE_SIZE):
-    """The kernel must honor the cache's page-dim stride, not assume pages are
-    packed back-to-back. A per-layer view into a cross-layer (block-major)
-    cache has stride(0) inflated by num_layers; outputs must match a
-    contiguous cache holding the same data exactly."""
-    B = 3
-    seq_len = 1027
-    CACHE_SIZE = 16384
-    NUM_LAYERS = 3
-    LAYER_IDX = 1
-    dtype = torch.bfloat16
-    sm_scale = 1.0 / (D_QK**0.5)
-    num_kv_splits = 8
-    num_pages = CACHE_SIZE // PAGE_SIZE
-
-    num_pages_per_batch = cdiv(seq_len, PAGE_SIZE)
-    req_to_page = torch.randint(
-        0, num_pages, (B, num_pages_per_batch), device=DEVICE_TYPE
-    )
-
-    q = torch.randn(B, H_Q, D_QK, dtype=dtype, device=DEVICE_TYPE)
-    b_seq_len = torch.full((B,), seq_len, device=DEVICE_TYPE)
-
-    # Reference: contiguous paged cache.
-    k_ref = torch.randn(
-        num_pages, PAGE_SIZE, H_KV, D_QK, dtype=dtype, device=DEVICE_TYPE
-    )
-    if is_mla:
-        v_ref = k_ref[..., :D_V]
-    else:
-        v_ref = torch.randn(
-            num_pages, PAGE_SIZE, H_KV, D_V, dtype=dtype, device=DEVICE_TYPE
-        )
-
-    # Cross-layer cache: all layers' pages for a block are adjacent. The
-    # per-layer view has the same shape as the contiguous cache but
-    # stride(0) is NUM_LAYERS x larger. Neighbor layers hold random data so
-    # any packed-pages addressing reads garbage rather than zeros.
-    k_xl = torch.randn(
-        num_pages, NUM_LAYERS, PAGE_SIZE, H_KV, D_QK, dtype=dtype, device=DEVICE_TYPE
-    )
-    k_view = k_xl[:, LAYER_IDX]
-    k_view.copy_(k_ref)
-    assert k_view.stride(0) == NUM_LAYERS * PAGE_SIZE * H_KV * D_QK
-    if is_mla:
-        v_view = k_view[..., :D_V]
-    else:
-        v_xl = torch.randn(
-            num_pages, NUM_LAYERS, PAGE_SIZE, H_KV, D_V, dtype=dtype, device=DEVICE_TYPE
-        )
-        v_view = v_xl[:, LAYER_IDX]
-        v_view.copy_(v_ref)
-
-    def run(k_buffer, v_buffer):
-        o = torch.zeros(B, H_Q, D_V, dtype=dtype, device=DEVICE_TYPE)
-        lse = torch.zeros(B, H_Q, dtype=dtype, device=DEVICE_TYPE)
-        attn_logits = torch.empty(
-            (B, H_Q, num_kv_splits, D_V + 1), dtype=torch.float32, device=DEVICE_TYPE
-        )
-        decode_attention_fwd(
-            q,
-            k_buffer,
-            v_buffer,
-            o,
-            lse,
-            req_to_page,
-            b_seq_len,
-            attn_logits,
-            num_kv_splits,
-            sm_scale,
-            PAGE_SIZE,
-            is_mla=is_mla,
-        )
-        return o, lse
-
-    o_ref, lse_ref = run(k_ref, v_ref)
-    o_xl, lse_xl = run(k_view, v_view)
-
-    # Same data and same compute order; only addressing differs.
-    assert torch.equal(o_ref, o_xl)
-    assert torch.equal(lse_ref, lse_xl)

tests/kernels/attention/test_triton_unified_attention.py

@@ -18,7 +18,11 @@ HEAD_SIZES = [128, 256]
 BLOCK_SIZES = [16]
 
 DTYPES = [torch.bfloat16]
-QDTYPES = [None, current_platform.fp8_dtype()]
+QDTYPES = (
+    [None, torch.float8_e4m3fn]
+    if not current_platform.is_rocm()
+    else [None, torch.float8_e4m3fnuz]
+)
 FP8_DTYPE = current_platform.fp8_dtype()
 
 # one value large enough to test overflow in index calculation.

tests/kernels/core/test_minimax_reduce_rms.py

@@ -10,12 +10,8 @@ from torch.multiprocessing import spawn
 from tests.kernels.utils import opcheck
 from tests.utils import ensure_current_vllm_config, init_test_distributed_environment
 from vllm.distributed import cleanup_dist_env_and_memory
-from vllm.model_executor.layers.minimax_rms_norm import (
-    MiniMaxText01RMSNormTP,
-    rms_norm_tp,
-)
+from vllm.model_executor.layers.minimax_rms_norm import MiniMaxText01RMSNormTP
 from vllm.platforms import current_platform
-from vllm.triton_utils import HAS_TRITON
 from vllm.utils.network_utils import get_open_port
 from vllm.utils.torch_utils import set_random_seed
 
@@ -58,19 +54,8 @@ def _worker_forward_qk(
     torch.manual_seed(seed + 1000 + local_rank)
     qkv = torch.randn(num_tokens, hq + hk + hk, dtype=dtype, device="cuda")
 
-    # Reference: eager all-reduce path. ``forward_qk`` no longer all-reduces
-    # the variance (it is the tp==1 / already-reduced building block), so the
-    # multi-rank reference must use the eager path that performs the global
-    # variance all-reduce, matching the fused kernel below.
-    ref_q, ref_k = rms_norm_tp._minimax_qk_norm_tp_eager(
-        qkv.clone(),
-        q_norm.weight,
-        k_norm.weight,
-        hq,
-        hk,
-        world_size,
-        eps,
-    )
+    q_ref, k_ref, v_ref = qkv.clone().split([hq, hk, hk], dim=-1)
+    ref_q, ref_k = MiniMaxText01RMSNormTP.forward_qk(q_norm, k_norm, q_ref, k_ref)
 
     # Set up Lamport workspace.
     from vllm.distributed.parallel_state import get_tp_group
@@ -165,44 +150,3 @@ def test_minimax_reduce_rms_qk(
         nprocs=world_size,
         join=True,
     )
-
-
-@pytest.mark.skipif(
-    not current_platform.is_cuda() or not HAS_TRITON,
-    reason="CUDA and Triton required",
-)
-@pytest.mark.parametrize("num_tokens", [1, 7, 128, 333, 2049])
-@pytest.mark.parametrize("hidden_dims", [(3072, 512), (768, 256), (3000, 500)])
-@pytest.mark.parametrize("dtype", [torch.bfloat16, torch.float16])
-@pytest.mark.parametrize("tp_world", [1, 4, 8])
-@pytest.mark.parametrize("eps", [1e-6])
-@pytest.mark.parametrize("seed", [42])
-def test_minimax_qk_norm_triton_fallback(
-    monkeypatch, num_tokens, hidden_dims, dtype, tp_world, eps, seed
-):
-    """Single-GPU check: Triton fallback kernels vs the pure-torch reference.
-
-    The all-reduce is a TP communication barrier, so it is monkeypatched to
-    identity here; both the Triton path and the reference see the same
-    (patched) reduction. This validates the kernel math and the folded
-    ``/ tp_world`` scaling without needing multiple ranks -- ``hidden_dims``
-    are the per-rank q/k segment widths.
-    """
-    monkeypatch.setattr(rms_norm_tp, "_all_reduce_variance", lambda v: v)
-
-    q_size, kv_size = hidden_dims
-    device = "cuda"
-    torch.manual_seed(seed)
-    qkv = torch.randn(num_tokens, q_size + 2 * kv_size, dtype=dtype, device=device)
-    q_weight = torch.randn(q_size, dtype=dtype, device=device)
-    k_weight = torch.randn(kv_size, dtype=dtype, device=device)
-
-    q_triton, k_triton = rms_norm_tp._minimax_qk_norm_tp_fallback(
-        qkv, q_weight, k_weight, q_size, kv_size, 0, tp_world, eps
-    )
-    q_ref, k_ref = rms_norm_tp._minimax_qk_norm_tp_eager(
-        qkv, q_weight, k_weight, q_size, kv_size, tp_world, eps
-    )
-
-    torch.testing.assert_close(q_triton, q_ref, atol=3e-2, rtol=3e-2)
-    torch.testing.assert_close(k_triton, k_ref, atol=3e-2, rtol=3e-2)

tests/kernels/moe/test_ocp_mx_moe.py

@@ -9,7 +9,6 @@ import pytest
 import torch
 from packaging import version
 
-from vllm._aiter_ops import is_aiter_found
 from vllm.platforms import current_platform
 from vllm.utils.flashinfer import has_flashinfer
 
@@ -32,15 +31,17 @@ HOPPER_MXFP4_BF16_AVAILABLE = (
 # ROCm platform and dependencies
 ROCM_AVAILABLE = current_platform.is_rocm()
 ROCM_TRITON_KERNELS_AVAILABLE = False
-ROCM_AITER_AVAILABLE = is_aiter_found()
+ROCM_AITER_AVAILABLE = False
 ROCM_GFX950 = False
 
 if ROCM_AVAILABLE:
+    from vllm._aiter_ops import rocm_aiter_ops
     from vllm.platforms.rocm import on_gfx950
     from vllm.utils.import_utils import has_triton_kernels
 
     ROCM_TRITON_KERNELS_AVAILABLE = has_triton_kernels()
     ROCM_GFX950 = on_gfx950()
+    ROCM_AITER_AVAILABLE = rocm_aiter_ops.is_enabled()
 
     if ROCM_AITER_AVAILABLE:
         from aiter.ops.triton.moe.quant_moe import upcast_from_mxfp
@@ -82,7 +83,7 @@ def enable_pickle(monkeypatch):
     [
         ModelCase("fxmarty/qwen_1.5-moe-a2.7b-mxfp4", tp=2),
         ModelCase("fxmarty/deepseek_r1_3_layers_mxfp4", tp=8),
-        ModelCase("mawong-amd/Llama-4-Scout-17B-16E-Instruct-2-layers-mxfp4", tp=1),
+        ModelCase("fxmarty/Llama-4-Scout-17B-16E-Instruct-2-layers-mxfp4", tp=1),
         ModelCase("fxmarty/Llama-3.1-70B-Instruct-2-layers-mxfp6", tp=1),
         ModelCase("fxmarty/Llama-3.1-70B-Instruct-2-layers-mxfp6", tp=4),
     ],
@@ -101,7 +102,6 @@ def test_mxfp4_loading_and_execution_moe(vllm_runner, model_case: ModelCase):
         tensor_parallel_size=model_case.tp,
         load_format="dummy",
         compilation_config={"cudagraph_capture_sizes": [16]},
-        gpu_memory_utilization=0.8,  # mxfp6 models use more scratch space
     ) as llm:
         # Disabled as check_model is broken: https://github.com/vllm-project/vllm/pull/18465#issuecomment-3329880562
         # def check_model(model):
@@ -1267,7 +1267,7 @@ def test_rocm_mxfp4_moe_oracle(
 
     This test validates that the oracle functions work end-to-end:
     - select_mxfp4_moe_backend() selects a valid backend
-    - convert_gpt_oss_weight_to_mxfp4_moe_kernel_format() converts weights without error
+    - convert_to_mxfp4_moe_kernel_format() converts weights without error
     - make_mxfp4_moe_quant_config() builds a valid quant config
     - make_mxfp4_moe_kernel() creates a kernel that runs without error
     - The kernel output is within accuracy tolerance of reference
@@ -1287,7 +1287,7 @@ def test_rocm_mxfp4_moe_oracle(
     from vllm.model_executor.layers.fused_moe.oracle.mxfp4 import (
         Mxfp4MoeBackend,
         backend_to_kernel_cls,
-        convert_gpt_oss_weight_to_mxfp4_moe_kernel_format,
+        convert_to_mxfp4_moe_kernel_format,
         make_mxfp4_moe_kernel,
         make_mxfp4_moe_quant_config,
     )
@@ -1387,7 +1387,7 @@ def test_rocm_mxfp4_moe_oracle(
 
     # Convert weights using oracle
     w13_conv, w2_conv, w13_scale_conv, w2_scale_conv, w13_bias_conv, w2_bias_conv = (
-        convert_gpt_oss_weight_to_mxfp4_moe_kernel_format(
+        convert_to_mxfp4_moe_kernel_format(
             mxfp4_backend=backend,
             layer=layer,  # type: ignore[arg-type]
             w13_weight=w13_quant,
@@ -1423,7 +1423,7 @@ def test_rocm_mxfp4_moe_oracle(
             mxfp4_backend=backend,
             experts_cls=experts_cls,
             routing_tables=None,
-            layer=None,
+            shared_experts=None,
         )
 
         # Create inputs

tests/kernels/moe/test_trtllm_nvfp4_moe.py

@@ -5,8 +5,9 @@ Tests for the FlashInfer TRTLLM NvFP4 MoE backend
 (`TrtLlmNvFp4ExpertsModular`).
 
 Covers the activations the wrapper claims to support — SiLU, RELU^2 (non-gated),
-and GELU — including a Gemma4-shaped case (128 experts, top-k 8,
-intermediate_size 704) that exercises the non-256-aligned padding path.
+GELU, and clamped SwiGLU-OAI (MiniMax-M3) — including a Gemma4-shaped case
+(128 experts, top-k 8, intermediate_size 704) that exercises the non-256-aligned
+padding path.
 """
 
 import pytest
@@ -80,6 +81,29 @@ if _CLAMP_OP_NAME not in op_registry:
 
 SILU_WITH_CLAMP = op_registry[_CLAMP_OP_NAME]
 
+# Clamped SwiGLU-OAI (MiniMax-M3): non-default alpha/beta so the kernel must
+# honor gemm1_alpha (raw) and gemm1_beta (folded by g1_alphas), not just clamp.
+_SWIGLU_ALPHA = 1.702
+_SWIGLU_BETA = 1.0
+_OAI_OP_NAME = "test_swigluoai_with_clamp"
+
+if _OAI_OP_NAME not in op_registry:
+
+    @CustomOp.register(_OAI_OP_NAME)
+    class _SwigluOAIWithClampTest(SiluAndMulWithClamp):
+        custom_op_name = _OAI_OP_NAME
+
+        def __init__(self, *, compile_native: bool = True) -> None:
+            super().__init__(
+                _SWIGLU_LIMIT,
+                alpha=_SWIGLU_ALPHA,
+                beta=_SWIGLU_BETA,
+                compile_native=compile_native,
+            )
+
+
+SWIGLUOAI_REF = op_registry[_OAI_OP_NAME]
+
 
 ACTIVATION_CASES = [
     pytest.param(MoEActivation.SILU, MoEActivation.SILU, None, id="silu"),
@@ -91,6 +115,12 @@ ACTIVATION_CASES = [
         id="relu2_no_mul",
     ),
     pytest.param(MoEActivation.GELU, MoEActivation.GELU, None, id="gelu"),
+    pytest.param(
+        MoEActivation.SWIGLUOAI_UNINTERLEAVE,
+        SWIGLUOAI_REF,
+        _SWIGLU_LIMIT,
+        id="swigluoai_uninterleave",
+    ),
 ]
 
 
@@ -148,6 +178,10 @@ def test_trtllm_fp4_moe_no_graph(
             is_scale_swizzled=False,
         )
         quant_config.gemm1_clamp_limit = swiglu_limit
+        is_oai = activation == MoEActivation.SWIGLUOAI_UNINTERLEAVE
+        if is_oai:
+            quant_config.gemm1_alpha = _SWIGLU_ALPHA
+            quant_config.gemm1_beta = _SWIGLU_BETA
         if swiglu_limit is not None:
             assert quant_config.g1_alphas is not None
             assert quant_config.a2_gscale is not None
@@ -192,6 +226,27 @@ def test_trtllm_fp4_moe_no_graph(
         fake_layer.w2_input_scale = torch.ones_like(quant_config.g2_alphas)
         trtllm_inner.process_weights_after_loading(fake_layer)
 
+        if is_oai:
+            # alpha stays raw; beta and clamp are folded by g1_alphas
+            # (== _LARGE_OUTPUT1_SCALE here), so the fold is load-bearing.
+            assert torch.allclose(
+                trtllm_inner.gemm1_alpha,
+                torch.full_like(trtllm_inner.gemm1_alpha, _SWIGLU_ALPHA),
+            )
+            assert torch.allclose(
+                trtllm_inner.gemm1_beta,
+                torch.full_like(
+                    trtllm_inner.gemm1_beta, _SWIGLU_BETA / _LARGE_OUTPUT1_SCALE
+                ),
+            )
+            assert torch.allclose(
+                trtllm_inner.gemm1_clamp_limit,
+                torch.full_like(
+                    trtllm_inner.gemm1_clamp_limit,
+                    _SWIGLU_LIMIT / _LARGE_OUTPUT1_SCALE,
+                ),
+            )
+
         trtllm_experts = mk.FusedMoEKernel(
             maybe_make_prepare_finalize(
                 moe=moe_config,

tests/kernels/test_fused_minimax_m3_qknorm_rope_kv_insert.py

@@ -278,3 +278,99 @@ def test_sparse_full(num_tokens, block_size, kv_cache_dtype):
     torch.testing.assert_close(
         index_cache.view(-1, HEAD_DIM), expected_index_cache, rtol=0, atol=0
     )
+
+
+# ── Test 3: fp8 (e4m3) index outputs ─────────────────────────────────────────
+# The fp8 score path stores index_q and the index-K cache as e4m3 while q/k/v +
+# q_out stay bf16. Asserts: (1) q/k/v/q_out are bit-identical to the bf16 run
+# (the index dtype must not perturb the main branch), and (2) the e4m3 index
+# outputs dequantize close to the bf16 reference.
+
+
+@pytest.mark.skipif(
+    not torch.cuda.is_available() or torch.cuda.get_device_capability() < (8, 9),
+    reason="e4m3 conversion requires CUDA SM89+.",
+)
+@pytest.mark.parametrize("num_tokens", [1, 7, 64, 513])
+@pytest.mark.parametrize("block_size", [16, 64])
+def test_sparse_full_fp8_index(num_tokens, block_size):
+    torch.manual_seed(1)
+    device, dtype, eps = "cuda", torch.bfloat16, 1e-6
+    base, max_pos = 5_000_000.0, 4096
+    num_heads, num_kv_heads, num_idx_heads = 16, 4, 4
+
+    q_w = torch.randn(HEAD_DIM, dtype=dtype, device=device) * 0.1
+    k_w = torch.randn(HEAD_DIM, dtype=dtype, device=device) * 0.1
+    iq_w = torch.randn(HEAD_DIM, dtype=dtype, device=device) * 0.1
+    ik_w = torch.randn(HEAD_DIM, dtype=dtype, device=device) * 0.1
+    cos_sin = make_cos_sin_cache(max_pos, ROTARY_DIM, base, dtype, device)
+    positions = torch.randint(
+        0, max_pos, (num_tokens,), dtype=torch.int64, device=device
+    )
+
+    qsz, kvsz = num_heads * HEAD_DIM, num_kv_heads * HEAD_DIM
+    iqsz, iksz = num_idx_heads * HEAD_DIM, HEAD_DIM
+    qkv0 = torch.randn(
+        num_tokens, qsz + 2 * kvsz + iqsz + iksz, dtype=dtype, device=device
+    )
+
+    num_blocks = (num_tokens + block_size - 1) // block_size + 1
+    slot_mapping = torch.randperm(
+        num_blocks * block_size, dtype=torch.int64, device=device
+    )[:num_tokens]
+    index_slot_mapping = torch.roll(slot_mapping, shifts=1)
+
+    def run(index_dtype):
+        qkv = qkv0.clone()
+        kv_cache = torch.zeros(
+            num_blocks,
+            2,
+            block_size,
+            num_kv_heads,
+            HEAD_DIM,
+            dtype=dtype,
+            device=device,
+        )
+        index_cache = torch.zeros(
+            num_blocks, block_size, HEAD_DIM, dtype=index_dtype, device=device
+        )
+        q_out = torch.empty(num_tokens, qsz, dtype=dtype, device=device)
+        index_q = torch.empty(num_tokens, iqsz, dtype=index_dtype, device=device)
+        ops.fused_minimax_m3_qknorm_rope_kv_insert(
+            qkv,
+            q_w,
+            k_w,
+            cos_sin,
+            positions,
+            num_heads,
+            num_kv_heads,
+            ROTARY_DIM,
+            eps,
+            iq_w,
+            ik_w,
+            num_idx_heads,
+            slot_mapping,
+            index_slot_mapping,
+            kv_cache,
+            index_cache,
+            block_size,
+            q_out,
+            index_q,
+        )
+        return qkv, kv_cache, index_cache, q_out, index_q
+
+    qkv_bf, kvc_bf, idxc_bf, qo_bf, iq_bf = run(torch.bfloat16)
+    qkv_fp, kvc_fp, idxc_fp, qo_fp, iq_fp = run(torch.float8_e4m3fn)
+
+    assert iq_fp.dtype == torch.float8_e4m3fn
+    assert idxc_fp.dtype == torch.float8_e4m3fn
+
+    # (1) The main branch (q/k/v in qkv, q_out, kv cache) must be bit-identical:
+    # the index output dtype must not perturb anything else.
+    torch.testing.assert_close(qo_fp, qo_bf, rtol=0, atol=0)
+    torch.testing.assert_close(qkv_fp, qkv_bf, rtol=0, atol=0)
+    torch.testing.assert_close(kvc_fp, kvc_bf, rtol=0, atol=0)
+
+    # (2) Dequantized e4m3 index outputs match the bf16 reference within fp8 ulp.
+    torch.testing.assert_close(iq_fp.float(), iq_bf.float(), rtol=0.13, atol=0.05)
+    torch.testing.assert_close(idxc_fp.float(), idxc_bf.float(), rtol=0.13, atol=0.05)

tests/kernels/test_mhc_kernels.py

@@ -8,8 +8,8 @@ from vllm.model_executor.kernels.mhc.tilelang import (
     _tilelang_hc_prenorm_gemm,
     _torch_hc_prenorm_gemm,
 )
-from vllm.model_executor.layers.mhc import HAS_TILELANG_MHC
 from vllm.platforms import current_platform
+from vllm.utils.import_utils import has_tilelang
 from vllm.utils.torch_utils import set_random_seed
 
 DEVICE = current_platform.device_type
@@ -97,8 +97,8 @@ def hc_head_ref(
 
 
 @pytest.mark.skipif(
-    not HAS_TILELANG_MHC,
-    reason="TileLang MHC support required",
+    not (current_platform.is_cuda_alike() and has_tilelang()),
+    reason="CUDA or ROCm and tilelang required",
 )
 @pytest.mark.parametrize("num_tokens", [1, 4, 8, 128])
 @pytest.mark.parametrize("hidden_size", [4096, 7168])
@@ -150,8 +150,8 @@ def test_mhc_pre_tilelang(num_tokens, hidden_size, hc_mult):
 
 
 @pytest.mark.skipif(
-    not HAS_TILELANG_MHC,
-    reason="TileLang MHC support required",
+    not (current_platform.is_cuda_alike() and has_tilelang()),
+    reason="CUDA or ROCm and tilelang required",
 )
 @pytest.mark.parametrize(
     ("num_tokens", "hidden_size"),
@@ -190,8 +190,8 @@ def test_hc_prenorm_gemm_tilelang(num_tokens, hidden_size):
 
 
 @pytest.mark.skipif(
-    not HAS_TILELANG_MHC,
-    reason="TileLang MHC support required",
+    not (current_platform.is_cuda_alike() and has_tilelang()),
+    reason="CUDA or ROCm and tilelang required",
 )
 @pytest.mark.parametrize("num_tokens", [1, 4, 8, 128])
 @pytest.mark.parametrize("hidden_size", [4096, 7168])
@@ -217,8 +217,8 @@ def test_mhc_post_tilelang(num_tokens, hidden_size, hc_mult):
 
 
 @pytest.mark.skipif(
-    not HAS_TILELANG_MHC,
-    reason="TileLang MHC support required",
+    not (current_platform.is_cuda_alike() and has_tilelang()),
+    reason="CUDA or ROCm and tilelang required",
 )
 @pytest.mark.parametrize("num_tokens", [1, 4, 8, 128])
 @pytest.mark.parametrize("hidden_size", [4096, 7168])
@@ -324,8 +324,8 @@ def test_hc_head_triton(num_tokens, hidden_size, hc_mult):
 
 
 @pytest.mark.skipif(
-    not HAS_TILELANG_MHC,
-    reason="TileLang MHC support required",
+    not (current_platform.is_cuda_alike() and has_tilelang()),
+    reason="CUDA or ROCm and tilelang required",
 )
 @pytest.mark.parametrize("num_tokens", [1, 4, 8, 128])
 @pytest.mark.parametrize("hidden_size", [4096, 7168])

tests/lora/test_lora_manager.py

@@ -124,7 +124,6 @@ def test_replace_submodules(default_vllm_config, dist_init, dummy_model):
             max_lora_rank=8, max_cpu_loras=8, max_loras=8, lora_dtype=DEFAULT_DTYPE
         ),
         torch.device(DEVICES[0]),
-        default_vllm_config,
     )
     model = manager.model
     assert isinstance(model.get_submodule("dense1"), ColumnParallelLinearWithLoRA)
@@ -153,7 +152,6 @@ def test_wrap_replicated_linear_subclasses(default_vllm_config, dist_init, dummy
             max_lora_rank=8, max_cpu_loras=8, max_loras=8, lora_dtype=DEFAULT_DTYPE
         ),
         torch.device(DEVICES[0]),
-        default_vllm_config,
     )
 
     assert isinstance(
@@ -174,7 +172,6 @@ def test_wrap_gate_linear(default_vllm_config, dist_init, dummy_model):
             max_lora_rank=8, max_cpu_loras=8, max_loras=8, lora_dtype=DEFAULT_DTYPE
         ),
         torch.device(DEVICES[0]),
-        default_vllm_config,
     )
 
     assert isinstance(
@@ -222,7 +219,6 @@ def test_dedup_shared_module_across_paths(default_vllm_config, dist_init, dummy_
             max_lora_rank=8, max_cpu_loras=8, max_loras=8, lora_dtype=DEFAULT_DTYPE
         ),
         torch.device(DEVICES[0]),
-        default_vllm_config,
     )
 
     canonical = manager.model.get_submodule("moe.gate")
@@ -267,7 +263,6 @@ def test_lm_head_exempt_from_dedup(default_vllm_config, dist_init, dummy_model):
             max_lora_rank=8, max_cpu_loras=8, max_loras=8, lora_dtype=DEFAULT_DTYPE
         ),
         torch.device(DEVICES[0]),
-        default_vllm_config,
     )
 
     # lm_head's special handling still ran: logits_processor got wrapped
@@ -298,7 +293,6 @@ def test_skip_unsupported_matched_modules(default_vllm_config, dist_init, dummy_
             max_lora_rank=8, max_cpu_loras=8, max_loras=8, lora_dtype=DEFAULT_DTYPE
         ),
         torch.device(DEVICES[0]),
-        default_vllm_config,
     )
 
     # Should not crash and should keep unsupported matched modules unchanged.
@@ -331,7 +325,6 @@ def test_target_modules_fail_closed_on_unsupported_matched_modules(
                 target_modules=["dense1"],
             ),
             torch.device(DEVICES[0]),
-            default_vllm_config,
         )
 
 
@@ -381,7 +374,6 @@ def test_lora_model_manager(default_vllm_config, dist_init, dummy_model, device)
             max_lora_rank=8, max_cpu_loras=3, max_loras=2, lora_dtype=DEFAULT_DTYPE
         ),
         device=device,
-        vllm_config=default_vllm_config,
     )
     assert all(x is None for x in manager.lora_index_to_id)
     assert manager.add_adapter(model_lora1)
@@ -450,7 +442,6 @@ def test_lora_lru_cache_model_manager(
             max_lora_rank=8, max_cpu_loras=3, max_loras=2, lora_dtype=DEFAULT_DTYPE
         ),
         device=device,
-        vllm_config=default_vllm_config,
     )
     assert all(x is None for x in manager.lora_index_to_id)
     assert manager.add_adapter(model_lora1)
@@ -544,7 +535,6 @@ def test_lru_lora_model_manager(default_vllm_config, dist_init, dummy_model, dev
             max_lora_rank=8, max_cpu_loras=2, max_loras=2, lora_dtype=DEFAULT_DTYPE
         ),
         device=device,
-        vllm_config=default_vllm_config,
     )
     assert all(x is None for x in manager.lora_index_to_id)
 
@@ -652,7 +642,9 @@ def test_lru_lora_model_manager(default_vllm_config, dist_init, dummy_model, dev
 
 
 @pytest.mark.parametrize("device", DEVICES)
-def test_lru_cache_worker_adapter_manager(dist_init, dummy_model, device, tmp_path):
+def test_lru_cache_worker_adapter_manager(
+    default_vllm_config, dist_init, dummy_model, device, tmp_path
+):
     lora_config = LoRAConfig(
         max_lora_rank=8, max_cpu_loras=4, max_loras=4, lora_dtype=DEFAULT_DTYPE
     )
@@ -678,7 +670,7 @@ def test_lru_cache_worker_adapter_manager(dist_init, dummy_model, device, tmp_pa
     worker_adapter_manager.max_num_seqs = 4
     worker_adapter_manager.max_num_batched_tokens = 2
 
-    worker_adapter_manager.create_lora_manager(dummy_model, vllm_config)
+    worker_adapter_manager.create_lora_manager(dummy_model)
 
     mapping = LoRAMapping([], [])
     worker_adapter_manager.set_active_adapters(
@@ -766,7 +758,9 @@ def test_lru_cache_worker_adapter_manager(dist_init, dummy_model, device, tmp_pa
 
 
 @pytest.mark.parametrize("device", DEVICES)
-def test_worker_adapter_manager(dist_init, dummy_model_gate_up, device, tmp_path):
+def test_worker_adapter_manager(
+    default_vllm_config, dist_init, dummy_model_gate_up, device, tmp_path
+):
     # Should remove every LoRA not specified in the request.
     lora_config = LoRAConfig(
         max_lora_rank=8, max_cpu_loras=4, max_loras=4, lora_dtype=DEFAULT_DTYPE
@@ -780,7 +774,7 @@ def test_worker_adapter_manager(dist_init, dummy_model_gate_up, device, tmp_path
 
     worker_adapter_manager = WorkerLoRAManager(vllm_config, device, EMBEDDING_MODULES)
     worker_adapter_manager.vocab_size = dummy_model_gate_up.unpadded_vocab_size
-    worker_adapter_manager.create_lora_manager(dummy_model_gate_up, vllm_config)
+    worker_adapter_manager.create_lora_manager(dummy_model_gate_up)
 
     dummy_lora_files = f"{tmp_path}/lora_adapter"
     os.makedirs(dummy_lora_files, exist_ok=True)
@@ -900,7 +894,6 @@ def test_packed_loras(default_vllm_config, dist_init, dummy_model_gate_up, devic
             max_lora_rank=8, max_cpu_loras=2, max_loras=2, lora_dtype=DEFAULT_DTYPE
         ),
         device=device,
-        vllm_config=default_vllm_config,
     )
     model = manager.model
 
@@ -951,7 +944,6 @@ def _test_target_modules(
     device: str,
     expected_lora: list[tuple[str, type]],
     expected_no_lora: list[tuple[str, type]],
-    vllm_config,
 ):
     """Create a LoRAModelManager and assert which modules have LoRA applied."""
     LoRAModelManager(
@@ -967,7 +959,6 @@ def _test_target_modules(
             target_modules=target_modules,
         ),
         device=device,
-        vllm_config=vllm_config,
     )
     for module_path, lora_cls in expected_lora:
         assert isinstance(model.get_submodule(module_path), lora_cls)
@@ -990,7 +981,6 @@ def test_target_modules_config(default_vllm_config, dist_init, dummy_model, devi
             ("dense2", RowParallelLinearWithLoRA),
             ("layer1.dense2", RowParallelLinearWithLoRA),
         ],
-        vllm_config=default_vllm_config,
     )
 
 
@@ -1008,7 +998,6 @@ def test_target_modules_multiple(default_vllm_config, dist_init, dummy_model, de
             ("layer1.dense2", RowParallelLinearWithLoRA),
         ],
         expected_no_lora=[],
-        vllm_config=default_vllm_config,
     )
 
 
@@ -1028,7 +1017,6 @@ def test_target_modules_none_uses_all(
             ("layer1.dense2", RowParallelLinearWithLoRA),
         ],
         expected_no_lora=[],
-        vllm_config=default_vllm_config,
     )
 
 
@@ -1048,5 +1036,4 @@ def test_target_modules_match_packed_runtime_modules(
             ("layer1.dense1", ColumnParallelLinearWithLoRA),
             ("layer1.dense2", RowParallelLinearWithLoRA),
         ],
-        vllm_config=default_vllm_config,
     )

tests/models/multimodal/generation/test_common.py

@@ -810,6 +810,29 @@ VLM_TEST_SETTINGS = {
         hf_output_post_proc=model_utils.minicpmv_trunc_hf_output,
         patch_hf_runner=model_utils.minicpmv_26_patch_hf_runner,
     ),
+    "minimax_vl_01": VLMTestInfo(
+        models=["MiniMaxAI/MiniMax-VL-01"],
+        prompt_formatter=lambda img_prompt: f"<beginning_of_sentence>user: {img_prompt} assistant:<end_of_sentence>",  # noqa: E501
+        img_idx_to_prompt=lambda _: "<image>",
+        test_type=(VLMTestType.IMAGE, VLMTestType.MULTI_IMAGE),
+        max_model_len=8192,
+        max_num_seqs=4,
+        dtype="bfloat16",
+        hf_output_post_proc=model_utils.minimax_vl_01_hf_output,
+        patch_hf_runner=model_utils.minimax_vl_01_patch_hf_runner,
+        auto_cls=AutoModelForImageTextToText,
+        marks=[
+            large_gpu_mark(min_gb=80),
+            # TODO: [ROCm] Fix pickle issue with ROCm spawn and tp>1
+            pytest.mark.skipif(
+                current_platform.is_rocm(),
+                reason=(
+                    "ROCm: Model too large for single GPU; "
+                    "multi-GPU blocked by HF _LazyConfigMapping pickle issue with spawn"
+                ),
+            ),
+        ],
+    ),
     "molmo": VLMTestInfo(
         models=["allenai/Molmo-7B-D-0924"],
         test_type=(VLMTestType.IMAGE, VLMTestType.MULTI_IMAGE),

tests/models/multimodal/generation/vlm_utils/model_utils.py

@@ -245,6 +245,13 @@ def minicpmv_trunc_hf_output(hf_output: RunnerOutput, model: str) -> RunnerOutpu
     return output_ids, output_str, out_logprobs
 
 
+def minimax_vl_01_hf_output(hf_output: RunnerOutput, model: str) -> RunnerOutput:
+    output_ids, output_str, out_logprobs = hf_output
+    if output_str.endswith("<end_of_sentence>"):
+        output_str = output_str.split("<end_of_sentence>")[0]
+    return output_ids, output_str, out_logprobs
+
+
 def ultravox_trunc_hf_output(hf_output: RunnerOutput, model: str) -> RunnerOutput:
     output_ids, output_str, out_logprobs = hf_output
 
@@ -1016,6 +1023,17 @@ def minicpmv_26_patch_hf_runner(hf_model: HfRunner) -> HfRunner:
     return hf_model
 
 
+def minimax_vl_01_patch_hf_runner(hf_model: HfRunner) -> HfRunner:
+    orig_generate = hf_model.model.generate
+
+    def _generate(self, *args, image_sizes=None, **kwargs):
+        return orig_generate(*args, decode_text=False, **kwargs)
+
+    hf_model.model.generate = types.MethodType(_generate, hf_model.model)
+
+    return hf_model
+
+
 def molmo_patch_hf_runner(hf_model: HfRunner) -> HfRunner:
     """Patches and returns an instance of the HfRunner to use for Molmo."""
     hf_processor = hf_model.processor

tests/models/multimodal/pooling/test_colqwen3_5.py

@@ -152,21 +152,3 @@ def test_colqwen3_5_relevance_ordering(
     dtype: str,
 ) -> None:
     _run_relevance_test(vllm_runner, model, dtype=dtype)
-
-
-def test_colqwen3_5_config_enables_bidirectional_attention() -> None:
-    """ColQwen3.5 retrieval must be served BIDIRECTIONAL (is_causal=False) so the
-    full_attention layers build with AttentionType.ENCODER_ONLY. This guards the
-    silent-causal regression (no GPU / model load needed)."""
-    from types import SimpleNamespace
-
-    from vllm.model_executor.models.config import (
-        MODELS_CONFIG_MAP,
-        ColQwen3_5Config,
-    )
-
-    assert MODELS_CONFIG_MAP["ColQwen3_5"] is ColQwen3_5Config
-
-    model_config = SimpleNamespace(hf_config=SimpleNamespace())
-    ColQwen3_5Config.verify_and_update_model_config(model_config)
-    assert model_config.hf_config.is_causal is False

tests/models/multimodal/processing/test_minimax_vl_01.py

@@ -0,0 +1,113 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
+
+import pytest
+from PIL import Image
+
+from vllm.multimodal import MULTIMODAL_REGISTRY
+from vllm.multimodal.parse import ImageSize
+from vllm.multimodal.processing import BaseMultiModalProcessor
+
+from ....conftest import ImageTestAssets
+from ...utils import build_model_context
+
+
+@pytest.mark.parametrize("model_id", ["MiniMaxAI/MiniMax-VL-01"])
+@pytest.mark.parametrize("num_imgs", [1, 2])
+def test_processor_override(
+    image_assets: ImageTestAssets,
+    model_id: str,
+    num_imgs: int,
+):
+    ctx = build_model_context(
+        model_id,
+        mm_processor_kwargs=None,
+        limit_mm_per_prompt={"image": num_imgs},
+    )
+    processor = MULTIMODAL_REGISTRY.create_processor(ctx.model_config)
+    prompt = "<image>" * num_imgs
+    image = Image.new("RGB", size=(364, 364))
+    mm_data = {"image": [image] * num_imgs}
+
+    processed_inputs = processor(
+        prompt,
+        mm_items=processor.info.parse_mm_data(mm_data),
+        hf_processor_mm_kwargs={},
+    )
+    image_placeholders = processed_inputs["mm_placeholders"]["image"]
+
+    assert len(image_placeholders) == num_imgs
+
+
+def _validate_image_prompt_replacements_one(
+    processor: BaseMultiModalProcessor,
+    num_imgs: int,
+    failed_size_excs: list[tuple[ImageSize, Exception]],
+    image_size: ImageSize,
+) -> None:
+    prompt = "<image>" * num_imgs
+    image = Image.new("RGB", size=image_size)
+    mm_data = {"image": [image] * num_imgs}
+
+    try:
+        processed_inputs = processor(
+            prompt,
+            mm_items=processor.info.parse_mm_data(mm_data),
+            hf_processor_mm_kwargs={},
+        )
+
+        image_placeholders = processed_inputs["mm_placeholders"]["image"]
+        assert len(image_placeholders) == num_imgs
+
+    except Exception as exc:
+        failed_size_excs.append((image_size, exc))
+
+
+def _test_image_prompt_replacements(
+    processor,
+    *,
+    num_imgs: int,
+    image_sizes: list[ImageSize],
+) -> None:
+    failed_size_excs = list[tuple[ImageSize, Exception]]()
+
+    for size in image_sizes:
+        _validate_image_prompt_replacements_one(
+            processor, num_imgs, failed_size_excs, size
+        )
+
+    if failed_size_excs:
+        msg = "Found failing image sizes:" + "\n========\n".join(
+            f"[{size}]\n{exc}" for size, exc in failed_size_excs
+        )
+        raise AssertionError(msg)
+
+
+@pytest.mark.parametrize("model_id", ["MiniMaxAI/MiniMax-VL-01"])
+@pytest.mark.parametrize("num_imgs", [1, 2])
+def test_processor_prompt_replacements_regression(model_id, num_imgs):
+    ctx = build_model_context(
+        model_id,
+        mm_processor_kwargs=None,
+        limit_mm_per_prompt={"image": num_imgs},
+    )
+    processor = MULTIMODAL_REGISTRY.create_processor(ctx.model_config)
+
+    image_ratios = [
+        (171, 152),
+        (184, 161),
+        (198, 176),
+        (333, 296),
+        (369, 328),
+        (488, 183),
+        (2560, 1669),
+    ]
+    image_sizes = [
+        size for w, h in image_ratios for size in [ImageSize(w, h), ImageSize(h, w)]
+    ]
+
+    _test_image_prompt_replacements(
+        processor,
+        num_imgs=num_imgs,
+        image_sizes=image_sizes,
+    )

tests/models/multimodal/processing/test_qwen3_vl.py

@@ -138,49 +138,3 @@ def test_processor_multi_video(
         assert video_phs[i].offset >= prev_end, (
             f"Placeholder {i} overlaps with placeholder {i - 1}"
         )
-
-
-@pytest.mark.parametrize("model_id", [MODEL_ID])
-@pytest.mark.parametrize(
-    "hf_mm_kwargs",
-    [{"num_frames": [8, 16]}, {"fps": [2.0, 4.0]}],
-)
-def test_processor_multi_video_list_kwargs(
-    model_id: str,
-    hf_mm_kwargs: dict[str, Any],
-) -> None:
-    """Regression test: a multi-video request with list-valued per-video
-    ``mm_processor_kwargs`` (one ``fps``/``num_frames`` per video) must not
-    crash.
-
-    Before the fix, ``_call_hf_processor`` copied the whole kwargs to every
-    video without slicing, so ``_get_video_second_idx`` received the list
-    where a scalar was expected and raised ``TypeError``.
-    """
-    ctx = build_model_context(
-        model_id,
-        limit_mm_per_prompt={"image": 0, "video": 2},
-    )
-    processor = MULTIMODAL_REGISTRY.create_processor(ctx.model_config)
-
-    prompt = (
-        "<|vision_start|><|video_pad|><|vision_end|>"
-        "<|vision_start|><|video_pad|><|vision_end|>"
-    )
-    mm_data = {
-        "video": [
-            _build_video_mm_data(num_frames=16)["video"][0],
-            _build_video_mm_data(num_frames=32)["video"][0],
-        ]
-    }
-
-    processed = processor(
-        prompt,
-        mm_items=processor.info.parse_mm_data(mm_data),
-        hf_processor_mm_kwargs=hf_mm_kwargs,
-    )
-
-    video_phs = processed["mm_placeholders"].get("video", [])
-    assert len(video_phs) == 2, (
-        f"Expected exactly 2 video placeholders, got {len(video_phs)}"
-    )

tests/models/registry.py

@@ -421,6 +421,15 @@ _TEXT_GENERATION_EXAMPLE_MODELS = {
         },
         trust_remote_code=True,
     ),
+    "MiniMaxForCausalLM": _HfExamplesInfo("MiniMaxAI/MiniMax-Text-01-hf"),
+    "MiniMaxText01ForCausalLM": _HfExamplesInfo(
+        "MiniMaxAI/MiniMax-Text-01",
+        trust_remote_code=True,
+        revision="a59aa9cbc53b9fb8742ca4e9e1531b9802b6fdc3",
+    ),
+    "MiniMaxM1ForCausalLM": _HfExamplesInfo(
+        "MiniMaxAI/MiniMax-M1-40k", trust_remote_code=True
+    ),
     "MiniMaxM2ForCausalLM": _HfExamplesInfo(
         "MiniMaxAI/MiniMax-M2",
         trust_remote_code=True,
@@ -1104,6 +1113,10 @@ _MULTIMODAL_EXAMPLE_MODELS = {
         "openbmb/MiniCPM-V-4_6",
         min_transformers_version="5.7.0",
     ),
+    "MiniMaxVL01ForConditionalGeneration": _HfExamplesInfo(
+        "MiniMaxAI/MiniMax-VL-01",
+        trust_remote_code=True,
+    ),
     "MiniMaxM3SparseForConditionalGeneration": _HfExamplesInfo(
         "MiniMaxAI/MiniMax-M3",
         trust_remote_code=True,
@@ -1517,16 +1530,6 @@ _SPECULATIVE_DECODING_EXAMPLE_MODELS = {
         "Qwen/Qwen3-VL-8B-Instruct",
         speculative_model="taobao-mnn/Qwen3-VL-8B-Instruct-Eagle3",
     ),
-    "Eagle3Qwen3ForCausalLM": _HfExamplesInfo(
-        "Qwen/Qwen3-8B",
-        trust_remote_code=True,
-        speculative_model=(
-            "inference-optimization/"
-            "Qwen3-8B-from-Qwen3-8B_regen-speculators.eagle3-qwen3arch-ckpt1"
-        ),
-        tokenizer="Qwen/Qwen3-8B",
-        use_original_num_layers=True,
-    ),
     # [PEagle]
     "PEagleDraftModel": _HfExamplesInfo(
         "Qwen/Qwen3-8B",
@@ -1542,15 +1545,6 @@ _SPECULATIVE_DECODING_EXAMPLE_MODELS = {
         tokenizer="Qwen/Qwen3-8B",
         use_original_num_layers=True,
     ),
-    "PeagleQwen3ForCausalLM": _HfExamplesInfo(
-        "Qwen/Qwen3-8B",
-        trust_remote_code=True,
-        speculative_model=(
-            "inference-optimization/Qwen3-8B-speculators.peagle-qwen3arch-ckpt4"
-        ),
-        tokenizer="Qwen/Qwen3-8B",
-        use_original_num_layers=True,
-    ),
     # [MTP]
     "DeepSeekMTPModel": _HfExamplesInfo(
         "luccafong/deepseek_mtp_main_random",

tests/models/test_initialization.py

@@ -98,6 +98,11 @@ def can_initialize(
         vllm_config.validate_block_size()
         return scheduler_kv_cache_config
 
+    if model_arch == "MiniMaxVL01ForConditionalGeneration":
+        pytest.skip(
+            "pickle error when loading `transformers.models.auto.CONFIG_MAPPING`"
+        )
+
     if model_arch == "MoonshotKimiaForCausalLM":
         pytest.skip(
             "Kimi-Audio requires SpeechToTextConfig "

tests/models/utils.py

@@ -507,13 +507,7 @@ def dummy_hf_overrides(
     # Only set MoE related config when the model has MoE layers.
     # Otherwise all models detected as MoE by _get_transformers_backend_cls.
     if model_arch_config.num_experts > 0:
-        num_experts_per_tok = 2
-        if model_arch in (
-            "Llama4ForConditionalGeneration",
-            "Llama4ForCausalLM",
-            "EagleLlama4ForCausalLM",
-        ):
-            num_experts_per_tok = 1
+        num_experts_per_tok = 1 if model_arch == "Llama4ForConditionalGeneration" else 2
         update_dict.update(
             {
                 "num_experts": num_experts,

tests/multimodal/test_video.py

@@ -15,7 +15,6 @@ from vllm.multimodal.video import (
     DynamicVideoBackend,
     GLM46VVideoBackend,
     Molmo2VideoBackend,
-    Qwen2VLVideoBackend,
     Qwen3VLVideoBackend,
     VideoLoader,
     VideoSourceMetadata,
@@ -71,12 +70,11 @@ def test_video_loader_type_doesnt_exist():
 
 
 @pytest.mark.parametrize(
-    "model_repo, expected_loader_cls, hf_sample_kwargs",
+    "model_repo, expected_loader_cls",
     [
         pytest.param(
             "allenai/Molmo2-4B",
             Molmo2VideoBackend,
-            None,
             marks=pytest.mark.skip(
                 reason="Video processor not aligned, investigate later.",
             ),
@@ -85,44 +83,23 @@ def test_video_loader_type_doesnt_exist():
         pytest.param(
             "zai-org/GLM-4.1V-9B-Thinking",
             DynamicVideoBackend,
-            None,
             id="glm4v",
         ),
         pytest.param(
             "zai-org/GLM-4.6V-Flash",
             GLM46VVideoBackend,
-            None,
             id="glm46v",
         ),
         pytest.param(
             "Qwen/Qwen3-VL-4B-Instruct",
             Qwen3VLVideoBackend,
-            None,
             id="qwen3vl",
         ),
-        # Qwen2-VL/Qwen2.5-VL ship no ``video_processor_type`` in their
-        # preprocessor config, so resolution relies on the model_type ->
-        # video processor fallback in get_video_processor_cls_name_from_config.
-        # They also ship no default fps/num_frames, so the HF sampler needs an
-        # explicit target rate; pass fps=2 to match the loader default.
-        pytest.param(
-            "Qwen/Qwen2-VL-7B-Instruct",
-            Qwen2VLVideoBackend,
-            {"fps": 2},
-            id="qwen2vl",
-        ),
-        pytest.param(
-            "Qwen/Qwen2.5-VL-7B-Instruct",
-            Qwen2VLVideoBackend,
-            {"fps": 2},
-            id="qwen2_5_vl",
-        ),
     ],
 )
 def test_video_processor_from_model_repo(
     model_repo: str,
     expected_loader_cls: type,
-    hf_sample_kwargs: dict[str, int | float] | None,
 ):
     """Test that a model repo resolves to the correct video loader backend.
 
@@ -166,7 +143,7 @@ def test_video_processor_from_model_repo(
             fps=vllm_meta["fps"],
             duration=vllm_meta["duration"],
         )
-        hf_indices = processor.sample_frames(hf_metadata, **(hf_sample_kwargs or {}))
+        hf_indices = processor.sample_frames(hf_metadata)
         vllm_indices = np.array(vllm_meta["frames_indices"])
         np.testing.assert_array_equal(
             hf_indices,

tests/quantization/test_auto_round.py

@@ -43,8 +43,8 @@ MODELS = [
     pytest.param(
         "Intel/Qwen2-0.5B-Instruct-int4-sym-AutoRound",
         marks=pytest.mark.skipif(
-            not (current_platform.is_cuda() or current_platform.is_xpu()),
-            reason="AWQ AutoRound model only supports CUDA/XPU backend for now.",
+            not current_platform.is_cuda(),
+            reason="AWQ AutoRound model only supports CUDA backend for now.",
         ),
         id="auto_round:auto_awq",
     ),

tests/quantization/test_per_token_kv_cache.py

@@ -61,8 +61,8 @@ class QuantConfig:
     quant_max: float
     quant_min: float
     kv_quant_mode: KVQuantMode
-    # INT8 rounds explicitly; FP8 relies on dtype cast rounding.
-    rounds_before_store: bool
+    # INT8 Triton stores truncate; FP8 hardware casts round.
+    uses_trunc: bool
 
 
 INT8_CONFIG = QuantConfig(
@@ -71,7 +71,7 @@ INT8_CONFIG = QuantConfig(
     quant_max=127.0,
     quant_min=-128.0,
     kv_quant_mode=KVQuantMode.INT8_PER_TOKEN_HEAD,
-    rounds_before_store=True,
+    uses_trunc=True,
 )
 FP8_CONFIG = QuantConfig(
     cache_dtype=FP8_DTYPE,
@@ -79,7 +79,7 @@ FP8_CONFIG = QuantConfig(
     quant_max=FP8_MAX,
     quant_min=FP8_MIN,
     kv_quant_mode=KVQuantMode.FP8_PER_TOKEN_HEAD,
-    rounds_before_store=False,
+    uses_trunc=False,
 )
 
 QUANT_CONFIGS = [INT8_CONFIG, FP8_CONFIG]
@@ -104,7 +104,7 @@ def _quantize_per_token_head_ref(
     absmax = data.float().abs().amax(dim=2)  # [num_tokens, num_heads]
     scales = (absmax / cfg.quant_max).clamp(min=1e-6)
     scaled = data.float() * (1.0 / scales[:, :, None])
-    if cfg.rounds_before_store:
+    if cfg.uses_trunc:
         q = scaled.round().clamp(cfg.quant_min, cfg.quant_max).to(cfg.cache_dtype)
     else:
         q = scaled.clamp(cfg.quant_min, cfg.quant_max).to(cfg.cache_dtype)
@@ -255,7 +255,7 @@ def test_per_token_head_round_trip_accuracy(
 ):
     """Verify per-token-head round-trip: kernel dequant matches reference.
 
-    INT8: round-to-nearest before int8 store.
+    INT8: Triton truncates on float->int8 store.
     FP8: hardware cast (clamp then cast).
     """
     from vllm.v1.attention.ops.triton_reshape_and_cache_flash import (
@@ -315,52 +315,6 @@ def test_per_token_head_round_trip_accuracy(
                 )
 
 
-@torch.inference_mode()
-def test_int8_per_token_head_raw_cache_matches_round_reference():
-    """INT8 cache writes should match round-to-nearest quantization exactly."""
-    from vllm.v1.attention.ops.triton_reshape_and_cache_flash import (
-        triton_reshape_and_cache_flash_per_token_head_quant,
-    )
-
-    torch.set_default_device(DEVICE_TYPE)
-
-    head_size = 8
-    block_size = 4
-
-    key = torch.tensor(
-        [[[-127.0, -2.6, -2.4, -1.6, -1.4, -0.6, -0.4, 127.0]]],
-        dtype=torch.bfloat16,
-    )
-    value = -key
-
-    key_cache = torch.zeros(1, block_size, 1, head_size, dtype=torch.int8)
-    value_cache = torch.zeros_like(key_cache)
-    k_scale_cache = torch.ones(1, block_size, 1, dtype=torch.float32)
-    v_scale_cache = torch.ones_like(k_scale_cache)
-    slot_mapping = torch.tensor([2], dtype=torch.long)
-
-    triton_reshape_and_cache_flash_per_token_head_quant(
-        key,
-        value,
-        key_cache,
-        value_cache,
-        k_scale_cache,
-        v_scale_cache,
-        slot_mapping,
-    )
-
-    ref_k_quant, ref_k_scales = _quantize_per_token_head_ref(key, INT8_CONFIG)
-    ref_v_quant, ref_v_scales = _quantize_per_token_head_ref(value, INT8_CONFIG)
-
-    slot = slot_mapping.item()
-    blk = slot // block_size
-    off = slot % block_size
-    assert torch.equal(key_cache[blk, off], ref_k_quant[0])
-    assert torch.equal(value_cache[blk, off], ref_v_quant[0])
-    torch.testing.assert_close(k_scale_cache[blk, off], ref_k_scales[0])
-    torch.testing.assert_close(v_scale_cache[blk, off], ref_v_scales[0])
-
-
 # ===========================================================================
 # 4. Negative slot mapping (padding tokens should be skipped)
 # ===========================================================================
@@ -507,7 +461,7 @@ def test_triton_unified_attention_per_token_head_scale(
 
     scaled_k = key_cache_bf16.float() / k_scale_cache[:, :, :, None]
     scaled_v = value_cache_bf16.float() / v_scale_cache[:, :, :, None]
-    if qcfg.rounds_before_store:
+    if qcfg.uses_trunc:
         key_cache_q = (
             scaled_k.round().clamp(qcfg.quant_min, qcfg.quant_max).to(qcfg.cache_dtype)
         )

tests/test_pooling_params.py

@@ -74,29 +74,6 @@ def test_embed_dimensions(model_info: EmbedModelInfo):
         pooling_params.verify(model_config)
 
 
-@dataclass()
-class MockMatryoshkaModelConfig:
-    pooler_config: PoolerConfig
-    is_matryoshka: bool = True
-    matryoshka_dimensions: list[int] | None = None
-    served_model_name: str = "mock-matryoshka-model"
-    embedding_size: int = 32
-
-
-def test_embed_dimensions_matryoshka_without_list_upper_bound():
-    task = "embed"
-    model_config = MockMatryoshkaModelConfig(
-        pooler_config=PoolerConfig(seq_pooling_type="CLS"),
-        matryoshka_dimensions=None,
-        embedding_size=32,
-    )
-
-    PoolingParams(task=task, dimensions=16).verify(model_config)
-
-    with pytest.raises(ValueError):
-        PoolingParams(task=task, dimensions=64).verify(model_config)
-
-
 @pytest.mark.parametrize("task", ["classify"])
 def test_classify(task):
     model_config = MockModelConfig(pooler_config=PoolerConfig(seq_pooling_type="CLS"))

tests/v1/attention/test_attention_backends_selection.py

@@ -6,12 +6,10 @@ from types import SimpleNamespace
 
 import pytest
 
-from vllm.model_executor.layers.mamba.linear.minimax_linear_attn import (
-    MiniMaxText01LinearAttention,
-)
 from vllm.model_executor.layers.mamba.mamba_mixer import MambaMixer
 from vllm.model_executor.layers.mamba.mamba_mixer2 import MambaMixer2
 from vllm.model_executor.layers.mamba.short_conv import ShortConv
+from vllm.model_executor.models.minimax_text_01 import MiniMaxText01LinearAttention
 from vllm.v1.attention.backends.linear_attn import LinearAttentionBackend
 from vllm.v1.attention.backends.mamba1_attn import Mamba1AttentionBackend
 from vllm.v1.attention.backends.mamba2_attn import Mamba2AttentionBackend