vllm - ✅(Solved) Fix [Bug]: Title: DeepSeek V4 intermittently leaks DSML fragments in auto + streaming mode, causing unstable tool-call parsing [2 pull requests, 1 participants]

Windswithyou · 2026-04-24T11:23:04Z

[vllm] PR 40760: New Model Support DeepseekV4 - Repository: vllm-project/vllm - Author: zyongye - State: open | merged: False - Link: https://github.com/vllm-p… # PR #40760: [New Model] Support DeepseekV4 - Repository: vllm-project/vllm - Author: zyongye - State: open | merged: False - Link: https://github.com/vllm-project/vllm/pull/40760 ## Description (problem / solution / changelog) Congratulations on Deepseek-ai to release the model. Thanks for all Inferact member's effort for support this. Note: This model implementation is highly optimized. All the component is coupled. Lot of manually fused kernel. Please consult @WoosukKwon @zyongye @ivanium before making any changes. Please see https://recipes.vllm.ai/deepseek-ai/DeepSeek-V4-Pro for recipes ## Changed files - `CMakeLists.txt` (modified, +4/-1) - `benchmarks/kernels/bench_compressor_fusion.py` (added, +416/-0) - `cmake/external_projects/deepgemm.cmake` (modified, +6/-1) - `cmake/external_projects/flashmla.cmake` (modified, +1/-1) - `csrc/cache_kernels.cu` (modified, +14/-7) - `csrc/fused_deepseek_v4_qnorm_rope_kv_insert_kernel.cu` (added, +477/-0) - `csrc/layernorm_kernels.cu` (modified, +15/-7) - `csrc/moe/moe_ops.h` (modified, +9/-0) - `csrc/moe/topk_softplus_sqrt_kernels.cu` (added, +715/-0) - `csrc/moe/torch_bindings.cpp` (modified, +6/-0) - `csrc/ops.h` (modified, +7/-1) - `csrc/persistent_topk.cuh` (modified, +17/-16) - `csrc/pos_encoding_kernels.cu` (modified, +39/-33) - `csrc/sampler.cu` (modified, +7/-1) - `csrc/topk.cu` (modified, +60/-35) - `csrc/torch_bindings.cpp` (modified, +13/-1) - `docker/Dockerfile` (modified, +50/-27) - `docker/Dockerfile.mooncake-layer` (added, +67/-0) - `docker/Dockerfile.nightly_torch` (modified, +2/-5) - `docker/versions.json` (modified, +1/-1) - `docs/design/attention_backends.md` (modified, +2/-2) - `requirements/cuda.txt` (modified, +4/-2) - `requirements/kv_connectors.txt` (modified, +12/-2) - `tests/kernels/attention/test_use_trtllm_attention.py` (modified, +1/-1) - `tests/kernels/core/test_fused_q_kv_rmsnorm.py` (added, +81/-0) - `tests/kernels/moe/test_deepgemm.py` (modified, +195/-1) - `tests/kernels/moe/test_ocp_mx_moe.py` (modified, +17/-15) - `tests/kernels/moe/test_topk_softplus_sqrt.py` (added, +159/-0) - `tests/kernels/test_compressor_kv_cache.py` (added, +406/-0) - `tests/kernels/test_fused_deepseek_v4_qnorm_rope_kv_insert.py` (added, +359/-0) - `tests/kernels/test_fused_indexer_q_rope_quant.py` (added, +98/-0) - `tests/kernels/test_fused_inv_rope_fp8_quant.py` (added, +998/-0) - `tests/kernels/test_top_k_per_row.py` (modified, +5/-10) - `tests/model_executor/test_routed_experts_capture.py` (modified, +3/-1) - `tests/models/registry.py` (modified, +1/-0) - `tests/reasoning/test_deepseekv3_reasoning_parser.py` (modified, +7/-0) - `tests/tokenizers_/fixtures/deepseek_v4/test_input_1.json` (added, +81/-0) - `tests/tokenizers_/fixtures/deepseek_v4/test_input_2.json` (added, +24/-0) - `tests/tokenizers_/fixtures/deepseek_v4/test_input_3.json` (added, +159/-0) - `tests/tokenizers_/fixtures/deepseek_v4/test_input_4.json` (added, +28/-0) - `tests/tokenizers_/fixtures/deepseek_v4/test_output_1.txt` (added, +36/-0) - `tests/tokenizers_/fixtures/deepseek_v4/test_output_2.txt` (added, +1/-0) - `tests/tokenizers_/fixtures/deepseek_v4/test_output_3.txt` (added, +38/-0) - `tests/tokenizers_/fixtures/deepseek_v4/test_output_4.txt` (added, +29/-0) - `tests/tokenizers_/test_deepseek_v4.py` (added, +147/-0) - `tests/tool_parsers/test_deepseekv4_tool_parser.py` (added, +123/-0) - `tests/v1/attention/test_indexer_deepseek_v4_slot_mapping.py` (added, +91/-0) - `tests/v1/core/test_kv_cache_utils.py` (modified, +3/-2) - `tests/v1/core/test_prefix_caching.py` (modified, +19/-20) - `tests/v1/core/test_scheduler.py` (modified, +2/-0) - `tests/v1/kv_connector/unit/test_mooncake_connector.py` (modified, +72/-23) - `tests/v1/kv_connector/unit/test_mooncake_connector_hma.py` (added, +400/-0) - `tests/v1/streaming_input/test_scheduler_streaming.py` (modified, +1/-0) - `tools/install_deepgemm.sh` (modified, +1/-1) - `vllm/_custom_ops.py` (modified, +34/-1) - `vllm/config/attention.py` (modified, +3/-0) - `vllm/config/cache.py` (modified, +14/-0) - `vllm/config/compilation.py` (modified, +1/-0) - `vllm/config/kernel.py` (modified, +3/-1) - `vllm/config/model.py` (modified, +5/-1) - `vllm/config/speculative.py` (modified, +11/-1) - `vllm/distributed/kv_transfer/kv_connector/v1/mooncake/mooncake_connector.py` (modified, +158/-42) - `vllm/model_executor/kernels/linear/scaled_mm/deep_gemm.py` (modified, +2/-0) - `vllm/model_executor/layers/attention/mla_attention.py` (modified, +29/-5) - `vllm/model_executor/layers/deepseek_compressor.py` (added, +436/-0) - `vllm/model_executor/layers/deepseek_v4_attention.py` (added, +1062/-0) - `vllm/model_executor/layers/fused_moe/config.py` (modified, +43/-1) - `vllm/model_executor/layers/fused_moe/experts/deep_gemm_moe.py` (modified, +234/-1) - `vllm/model_executor/layers/fused_moe/experts/trtl

vllm2026-04-24 11:23:04

ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

GitHub issue URL

Helpful · Quick feedback

GitHub stats

vllm-project/vllm#40801•Fetched 2026-04-25 06:03:59

View on GitHub

Comments

Participants

Timeline

Reactions

Author

Windswithyou

Participants

Windswithyou

Timeline (top)

subscribed ×3cross-referenced ×2labeled ×1

Root Cause

Initial Root Cause Hypothesis

Parser emits plain text too early before confirming whether current chunk tail is part of a tool-start marker.
DeepSeek DSML start tags are long and chunk-splitting makes this edge case likely.
required follows a different streaming branch, which explains the behavioral difference.

Fix Action

Fix / Workaround

============================== CPU Info

Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Address sizes: 46 bits physical, 57 bits virtual Byte Order: Little Endian CPU(s): 192 On-line CPU(s) list: 0-191 Vendor ID: GenuineIntel Model name: INTEL(R) XEON(R) PLATINUM 8558P CPU family: 6 Model: 207 Thread(s) per core: 2 Core(s) per socket: 48 Socket(s): 2 Stepping: 2 Frequency boost: enabled CPU max MHz: 2701.0000 CPU min MHz: 800.0000 BogoMIPS: 5400.00 Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cat_l2 cdp_l3 invpcid_single intel_ppin cdp_l2 ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local split_lock_detect avx_vnni avx512_bf16 wbnoinvd dtherm ida arat pln pts avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq la57 rdpid bus_lock_detect cldemote movdiri movdir64b enqcmd fsrm md_clear serialize tsxldtrk pconfig arch_lbr amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l1d arch_capabilities Virtualization: VT-x L1d cache: 4.5 MiB (96 instances) L1i cache: 3 MiB (96 instances) L2 cache: 192 MiB (96 instances) L3 cache: 520 MiB (2 instances) NUMA node(s): 2 NUMA node0 CPU(s): 0-47,96-143 NUMA node1 CPU(s): 48-95,144-191 Vulnerability Gather data sampling: Not affected Vulnerability Itlb multihit: Not affected Vulnerability L1tf: Not affected Vulnerability Mds: Not affected Vulnerability Meltdown: Not affected Vulnerability Mmio stale data: Not affected Vulnerability Reg file data sampling: Not affected Vulnerability Retbleed: Not affected Vulnerability Spec rstack overflow: Not affected Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl and seccomp Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization Vulnerability Spectre v2: Mitigation; Enhanced / Automatic IBRS; IBPB conditional; RSB filling; PBRSB-eIBRS SW sequence; BHI BHI_DIS_S Vulnerability Srbds: Not affected Vulnerability Tsx async abort: Not affected

PR fix notes

PR #40760: [New Model] Support DeepseekV4

Repository: vllm-project/vllm
Author: zyongye
State: open | merged: False
Link: https://github.com/vllm-project/vllm/pull/40760

Description (problem / solution / changelog)

Congratulations on Deepseek-ai to release the model. Thanks for all Inferact member's effort for support this.

Note: This model implementation is highly optimized. All the component is coupled. Lot of manually fused kernel. Please consult @WoosukKwon @zyongye @ivanium before making any changes.

Please see https://recipes.vllm.ai/deepseek-ai/DeepSeek-V4-Pro for recipes

Changed files

CMakeLists.txt (modified, +4/-1)
benchmarks/kernels/bench_compressor_fusion.py (added, +416/-0)
cmake/external_projects/deepgemm.cmake (modified, +6/-1)
cmake/external_projects/flashmla.cmake (modified, +1/-1)
csrc/cache_kernels.cu (modified, +14/-7)
csrc/fused_deepseek_v4_qnorm_rope_kv_insert_kernel.cu (added, +477/-0)
csrc/layernorm_kernels.cu (modified, +15/-7)
csrc/moe/moe_ops.h (modified, +9/-0)
csrc/moe/topk_softplus_sqrt_kernels.cu (added, +715/-0)
csrc/moe/torch_bindings.cpp (modified, +6/-0)
csrc/ops.h (modified, +7/-1)
csrc/persistent_topk.cuh (modified, +17/-16)
csrc/pos_encoding_kernels.cu (modified, +39/-33)
csrc/sampler.cu (modified, +7/-1)
csrc/topk.cu (modified, +60/-35)
csrc/torch_bindings.cpp (modified, +13/-1)
docker/Dockerfile (modified, +50/-27)
docker/Dockerfile.mooncake-layer (added, +67/-0)
docker/Dockerfile.nightly_torch (modified, +2/-5)
docker/versions.json (modified, +1/-1)
docs/design/attention_backends.md (modified, +2/-2)
requirements/cuda.txt (modified, +4/-2)
requirements/kv_connectors.txt (modified, +12/-2)
tests/kernels/attention/test_use_trtllm_attention.py (modified, +1/-1)
tests/kernels/core/test_fused_q_kv_rmsnorm.py (added, +81/-0)
tests/kernels/moe/test_deepgemm.py (modified, +195/-1)
tests/kernels/moe/test_ocp_mx_moe.py (modified, +17/-15)
tests/kernels/moe/test_topk_softplus_sqrt.py (added, +159/-0)
tests/kernels/test_compressor_kv_cache.py (added, +406/-0)
tests/kernels/test_fused_deepseek_v4_qnorm_rope_kv_insert.py (added, +359/-0)
tests/kernels/test_fused_indexer_q_rope_quant.py (added, +98/-0)
tests/kernels/test_fused_inv_rope_fp8_quant.py (added, +998/-0)
tests/kernels/test_top_k_per_row.py (modified, +5/-10)
tests/model_executor/test_routed_experts_capture.py (modified, +3/-1)
tests/models/registry.py (modified, +1/-0)
tests/reasoning/test_deepseekv3_reasoning_parser.py (modified, +7/-0)
tests/tokenizers_/fixtures/deepseek_v4/test_input_1.json (added, +81/-0)
tests/tokenizers_/fixtures/deepseek_v4/test_input_2.json (added, +24/-0)
tests/tokenizers_/fixtures/deepseek_v4/test_input_3.json (added, +159/-0)
tests/tokenizers_/fixtures/deepseek_v4/test_input_4.json (added, +28/-0)
tests/tokenizers_/fixtures/deepseek_v4/test_output_1.txt (added, +36/-0)
tests/tokenizers_/fixtures/deepseek_v4/test_output_2.txt (added, +1/-0)
tests/tokenizers_/fixtures/deepseek_v4/test_output_3.txt (added, +38/-0)
tests/tokenizers_/fixtures/deepseek_v4/test_output_4.txt (added, +29/-0)
tests/tokenizers_/test_deepseek_v4.py (added, +147/-0)
tests/tool_parsers/test_deepseekv4_tool_parser.py (added, +123/-0)
tests/v1/attention/test_indexer_deepseek_v4_slot_mapping.py (added, +91/-0)
tests/v1/core/test_kv_cache_utils.py (modified, +3/-2)
tests/v1/core/test_prefix_caching.py (modified, +19/-20)
tests/v1/core/test_scheduler.py (modified, +2/-0)
tests/v1/kv_connector/unit/test_mooncake_connector.py (modified, +72/-23)
tests/v1/kv_connector/unit/test_mooncake_connector_hma.py (added, +400/-0)
tests/v1/streaming_input/test_scheduler_streaming.py (modified, +1/-0)
tools/install_deepgemm.sh (modified, +1/-1)
vllm/_custom_ops.py (modified, +34/-1)
vllm/config/attention.py (modified, +3/-0)
vllm/config/cache.py (modified, +14/-0)
vllm/config/compilation.py (modified, +1/-0)
vllm/config/kernel.py (modified, +3/-1)
vllm/config/model.py (modified, +5/-1)
vllm/config/speculative.py (modified, +11/-1)
vllm/distributed/kv_transfer/kv_connector/v1/mooncake/mooncake_connector.py (modified, +158/-42)
vllm/model_executor/kernels/linear/scaled_mm/deep_gemm.py (modified, +2/-0)
vllm/model_executor/layers/attention/mla_attention.py (modified, +29/-5)
vllm/model_executor/layers/deepseek_compressor.py (added, +436/-0)
vllm/model_executor/layers/deepseek_v4_attention.py (added, +1062/-0)
vllm/model_executor/layers/fused_moe/config.py (modified, +43/-1)
vllm/model_executor/layers/fused_moe/experts/deep_gemm_moe.py (modified, +234/-1)
vllm/model_executor/layers/fused_moe/experts/trtllm_mxfp4_moe.py (modified, +84/-60)
vllm/model_executor/layers/fused_moe/fused_marlin_moe.py (modified, +19/-5)
vllm/model_executor/layers/fused_moe/fused_moe_method_base.py (modified, +1/-0)
vllm/model_executor/layers/fused_moe/gpt_oss_triton_kernels_moe.py (modified, +193/-2)
vllm/model_executor/layers/fused_moe/layer.py (modified, +9/-1)
vllm/model_executor/layers/fused_moe/oracle/mxfp4.py (modified, +409/-10)
vllm/model_executor/layers/fused_moe/router/base_router.py (modified, +5/-1)
vllm/model_executor/layers/fused_moe/router/custom_routing_router.py (modified, +2/-0)
vllm/model_executor/layers/fused_moe/router/fused_moe_router.py (modified, +2/-0)
vllm/model_executor/layers/fused_moe/router/fused_topk_bias_router.py (modified, +84/-16)
vllm/model_executor/layers/fused_moe/router/fused_topk_router.py (modified, +2/-0)
vllm/model_executor/layers/fused_moe/router/grouped_topk_router.py (modified, +3/-0)
vllm/model_executor/layers/fused_moe/router/router_factory.py (modified, +6/-3)
vllm/model_executor/layers/fused_moe/router/routing_simulator_router.py (modified, +2/-0)
vllm/model_executor/layers/fused_moe/runner/chunking_moe_runner.py (modified, +2/-0)
vllm/model_executor/layers/fused_moe/runner/default_moe_runner.py (modified, +4/-0)
vllm/model_executor/layers/fused_moe/runner/moe_runner.py (modified, +1/-0)
vllm/model_executor/layers/fused_moe/runner/moe_runner_base.py (modified, +14/-0)
vllm/model_executor/layers/fused_moe/shared_fused_moe.py (modified, +2/-0)
vllm/model_executor/layers/fused_moe/unquantized_fused_moe_method.py (modified, +1/-0)
vllm/model_executor/layers/fused_moe/utils.py (modified, +18/-0)
vllm/model_executor/layers/mhc.py (added, +436/-0)
vllm/model_executor/layers/quantization/__init__.py (modified, +3/-0)
vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors_moe/compressed_tensors_moe_w4a4_nvfp4.py (modified, +1/-0)
vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors_moe/compressed_tensors_moe_w4a8_int8.py (modified, +1/-0)
vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors_moe/compressed_tensors_moe_w8a8_fp8.py (modified, +1/-0)
vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors_moe/compressed_tensors_moe_w8a8_mxfp8.py (modified, +1/-0)
vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors_moe/compressed_tensors_moe_wna16_marlin.py (modified, +1/-0)
vllm/model_executor/layers/quantization/fp8.py (modified, +3/-0)
vllm/model_executor/layers/quantization/modelopt.py (modified, +3/-0)
vllm/model_executor/layers/quantization/mxfp4.py (modified, +333/-0)
vllm/model_executor/layers/quantization/online/fp8.py (modified, +1/-0)

PR #40805: fix/deepseekv4: prevent DSML marker leakage in streaming auto mode(v32 parser only)

Repository: vllm-project/vllm
Author: Windswithyou
State: open | merged: False
Link: https://github.com/vllm-project/vllm/pull/40805

Description (problem / solution / changelog)

Purpose

Fix #40801: DeepSeek V4 may leak DSML marker fragments in tool_choice=auto + stream=true mode due to split start-token handling in streaming parsing.

This PR provides a focused parser fix by updating the streaming logic to buffer potential start-token prefixes before emitting plain text, preventing DSML marker fragments from leaking into user-visible content.

Scope of change:

Update DeepSeek V3.2 tool parser streaming path (inherited by DeepSeek V4 parser path).
Prevent partial start-marker leakage in auto+stream tool-calling flows.
No model architecture / kernel / serving API behavior changes outside parser streaming handling.

Test Plan

Environment setup use vllm serve for deepseekv4

Test Result

Added/validated regression behavior for DeepSeek streaming parser path:
- No DSML start-marker fragment leakage in streamed plain-text content for auto mode.
- Tool-call extraction remains intact after buffering split prefixes.
Parser change is scoped and low-overhead (small bounded suffix/prefix matching per chunk).

(If maintainers need full logs, I can paste local pytest output in comments.)

<details> <summary> Essential Elements of an Effective PR Description Checklist </summary>

The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
The test plan, such as providing test command.
The test results, such as pasting the results comparison before and after, or e2e results.
(Optional) The necessary documentation update, such as updating supported_models.md and examples for a new model. (Not needed for this focused bugfix.)

</details>

Changed files

vllm/tool_parsers/deepseekv32_tool_parser.py (modified, +35/-17)

Code Example

Your output of `python collect_env.py` here
==============================
        System Info
==============================
OS                           : Ubuntu 22.04.5 LTS (x86_64)
GCC version                  : (Ubuntu 11.4.0-1ubuntu1~22.04.3) 11.4.0
Clang version                : Could not collect
CMake version                : Could not collect
Libc version                 : glibc-2.35

==============================
       PyTorch Info
==============================
PyTorch version              : 2.11.0+cu129
Is debug build               : False
CUDA used to build PyTorch   : 12.9
ROCM used to build PyTorch   : N/A
XPU used to build PyTorch    : N/A

==============================
      Python Environment
==============================
Python version               : 3.12.13 (main, Mar  4 2026, 09:23:07) [GCC 11.4.0] (64-bit runtime)
Python platform              : Linux-5.15.0-119-generic-x86_64-with-glibc2.35
    
==============================
       CUDA / GPU Info
==============================
Is CUDA available            : True
CUDA runtime version         : 12.9.86
CUDA_MODULE_LOADING set to   : 
GPU models and configuration : 
GPU 0: NVIDIA H20-3e
GPU 1: NVIDIA H20-3e
GPU 2: NVIDIA H20-3e
GPU 3: NVIDIA H20-3e
GPU 4: NVIDIA H20-3e
GPU 5: NVIDIA H20-3e
GPU 6: NVIDIA H20-3e
GPU 7: NVIDIA H20-3e

Nvidia driver version        : 575.57.08
cuDNN version                : Could not collect
HIP runtime version          : N/A
MIOpen runtime version       : N/A
Is XNNPACK available         : True

==============================
          CPU Info
==============================
Architecture:                         x86_64
CPU op-mode(s):                       32-bit, 64-bit
Address sizes:                        46 bits physical, 57 bits virtual
Byte Order:                           Little Endian
CPU(s):                               192
On-line CPU(s) list:                  0-191
Vendor ID:                            GenuineIntel
Model name:                           INTEL(R) XEON(R) PLATINUM 8558P
CPU family:                           6
Model:                                207
Thread(s) per core:                   2
Core(s) per socket:                   48
Socket(s):                            2
Stepping:                             2
Frequency boost:                      enabled
CPU max MHz:                          2701.0000
CPU min MHz:                          800.0000
BogoMIPS:                             5400.00
Flags:                                fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cat_l2 cdp_l3 invpcid_single intel_ppin cdp_l2 ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local split_lock_detect avx_vnni avx512_bf16 wbnoinvd dtherm ida arat pln pts avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq la57 rdpid bus_lock_detect cldemote movdiri movdir64b enqcmd fsrm md_clear serialize tsxldtrk pconfig arch_lbr amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l1d arch_capabilities
Virtualization:                       VT-x
L1d cache:                            4.5 MiB (96 instances)
L1i cache:                            3 MiB (96 instances)
L2 cache:                             192 MiB (96 instances)
L3 cache:                             520 MiB (2 instances)
NUMA node(s):                         2
NUMA node0 CPU(s):                    0-47,96-143
NUMA node1 CPU(s):                    48-95,144-191
Vulnerability Gather data sampling:   Not affected
Vulnerability Itlb multihit:          Not affected
Vulnerability L1tf:                   Not affected
Vulnerability Mds:                    Not affected
Vulnerability Meltdown:               Not affected
Vulnerability Mmio stale data:        Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed:               Not affected
Vulnerability Spec rstack overflow:   Not affected
Vulnerability Spec store bypass:      Mitigation; Speculative Store Bypass disabled via prctl and seccomp
Vulnerability Spectre v1:             Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2:             Mitigation; Enhanced / Automatic IBRS; IBPB conditional; RSB filling; PBRSB-eIBRS SW sequence; BHI BHI_DIS_S
Vulnerability Srbds:                  Not affected
Vulnerability Tsx async abort:        Not affected

==============================
Versions of relevant libraries
==============================
[pip3] flashinfer-python==0.6.8.post1
[pip3] numpy==2.2.6
[pip3] nvidia-cublas-cu12==12.9.1.4
[pip3] nvidia-cuda-cupti-cu12==12.9.79
[pip3] nvidia-cuda-nvrtc-cu12==12.9.86
[pip3] nvidia-cuda-runtime-cu12==12.9.79
[pip3] nvidia-cudnn-cu12==9.17.1.4
[pip3] nvidia-cudnn-frontend==1.18.0
[pip3] nvidia-cufft-cu12==11.4.1.4
[pip3] nvidia-cufile-cu12==1.14.1.1
[pip3] nvidia-curand-cu12==10.3.10.19
[pip3] nvidia-cusolver-cu12==11.7.5.82
[pip3] nvidia-cusparse-cu12==12.5.10.65
[pip3] nvidia-cusparselt-cu12==0.7.1
[pip3] nvidia-cutlass-dsl==4.4.2
[pip3] nvidia-cutlass-dsl-libs-base==4.4.2
[pip3] nvidia-ml-py==13.595.45
[pip3] nvidia-nccl-cu12==2.28.9
[pip3] nvidia-nvjitlink-cu12==12.9.86
[pip3] nvidia-nvshmem-cu12==3.4.5
[pip3] nvidia-nvtx-cu12==12.9.79
[pip3] pyzmq==27.1.0
[pip3] torch==2.11.0+cu129
[pip3] torch_c_dlpack_ext==0.1.5
[pip3] torchaudio==2.11.0+cu129
[pip3] torchvision==0.26.0+cu129
[pip3] transformers==4.57.6
[pip3] triton==3.6.0
[conda] Could not collect

==============================
         vLLM Info
==============================
ROCM Version                 : Could not collect
vLLM Version                 : 0.1.dev15830+g8d599d76a (git sha: 8d599d76a)
vLLM Build Flags:
  CUDA Archs: 7.0 7.5 8.0 8.9 9.0 10.0 12.0; ROCm: Disabled; XPU: Disabled
GPU Topology:
        GPU0    GPU1    GPU2    GPU3    GPU4    GPU5    GPU6    GPU7    NIC0    NIC1    NIC2    NIC3    NIC4    NIC5    CPU Affinity    NUMA Affinity   GPU NUMA ID
GPU0     X      NV18    NV18    NV18    NV18    NV18    NV18    NV18    PIX     PIX     NODE    NODE    NODE    SYS     0-47,96-143     0               N/A
GPU1    NV18     X      NV18    NV18    NV18    NV18    NV18    NV18    PIX     PIX     NODE    NODE    NODE    SYS     0-47,96-143     0               N/A
GPU2    NV18    NV18     X      NV18    NV18    NV18    NV18    NV18    NODE    NODE    NODE    NODE    NODE    SYS     0-47,96-143     0               N/A
GPU3    NV18    NV18    NV18     X      NV18    NV18    NV18    NV18    NODE    NODE    NODE    NODE    NODE    SYS     0-47,96-143     0               N/A
GPU4    NV18    NV18    NV18    NV18     X      NV18    NV18    NV18    SYS     SYS     SYS     SYS     SYS     NODE    48-95,144-191   1               N/A
GPU5    NV18    NV18    NV18    NV18    NV18     X      NV18    NV18    SYS     SYS     SYS     SYS     SYS     NODE    48-95,144-191   1               N/A
GPU6    NV18    NV18    NV18    NV18    NV18    NV18     X      NV18    SYS     SYS     SYS     SYS     SYS     PIX     48-95,144-191   1               N/A
GPU7    NV18    NV18    NV18    NV18    NV18    NV18    NV18     X      SYS     SYS     SYS     SYS     SYS     PIX     48-95,144-191   1               N/A
NIC0    PIX     PIX     NODE    NODE    SYS     SYS     SYS     SYS      X      PIX     NODE    NODE    NODE    SYS                             
NIC1    PIX     PIX     NODE    NODE    SYS     SYS     SYS     SYS     PIX      X      NODE    NODE    NODE    SYS                             
NIC2    NODE    NODE    NODE    NODE    SYS     SYS     SYS     SYS     NODE    NODE     X      PIX     NODE    SYS                             
NIC3    NODE    NODE    NODE    NODE    SYS     SYS     SYS     SYS     NODE    NODE    PIX      X      NODE    SYS                             
NIC4    NODE    NODE    NODE    NODE    SYS     SYS     SYS     SYS     NODE    NODE    NODE    NODE     X      SYS                             
NIC5    SYS     SYS     SYS     SYS     NODE    NODE    PIX     PIX     SYS     SYS     SYS     SYS     SYS      X                              

Legend:

  X    = Self
  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
  PIX  = Connection traversing at most a single PCIe bridge
  NV#  = Connection traversing a bonded set of # NVLinks

NIC Legend:

  NIC0: mlx5_0
  NIC1: mlx5_1
  NIC2: mlx5_2
  NIC3: mlx5_3
  NIC4: mlx5_4
  NIC5: mlx5_5

==============================
     Environment Variables
==============================
NVIDIA_VISIBLE_DEVICES=all
NVIDIA_REQUIRE_CUDA=cuda>=12.9 brand=unknown,driver>=535,driver<536 brand=grid,driver>=535,driver<536 brand=tesla,driver>=535,driver<536 brand=nvidia,driver>=535,driver<536 brand=quadro,driver>=535,driver<536 brand=quadrortx,driver>=535,driver<536 brand=nvidiartx,driver>=535,driver<536 brand=vapps,driver>=535,driver<536 brand=vpc,driver>=535,driver<536 brand=vcs,driver>=535,driver<536 brand=vws,driver>=535,driver<536 brand=cloudgaming,driver>=535,driver<536 brand=unknown,driver>=550,driver<551 brand=grid,driver>=550,driver<551 brand=tesla,driver>=550,driver<551 brand=nvidia,driver>=550,driver<551 brand=quadro,driver>=550,driver<551 brand=quadrortx,driver>=550,driver<551 brand=nvidiartx,driver>=550,driver<551 brand=vapps,driver>=550,driver<551 brand=vpc,driver>=550,driver<551 brand=vcs,driver>=550,driver<551 brand=vws,driver>=550,driver<551 brand=cloudgaming,driver>=550,driver<551 brand=unknown,driver>=560,driver<561 brand=grid,driver>=560,driver<561 brand=tesla,driver>=560,driver<561 brand=nvidia,driver>=560,driver<561 brand=quadro,driver>=560,driver<561 brand=quadrortx,driver>=560,driver<561 brand=nvidiartx,driver>=560,driver<561 brand=vapps,driver>=560,driver<561 brand=vpc,driver>=560,driver<561 brand=vcs,driver>=560,driver<561 brand=vws,driver>=560,driver<561 brand=cloudgaming,driver>=560,driver<561 brand=unknown,driver>=565,driver<566 brand=grid,driver>=565,driver<566 brand=tesla,driver>=565,driver<566 brand=nvidia,driver>=565,driver<566 brand=quadro,driver>=565,driver<566 brand=quadrortx,driver>=565,driver<566 brand=nvidiartx,driver>=565,driver<566 brand=vapps,driver>=565,driver<566 brand=vpc,driver>=565,driver<566 brand=vcs,driver>=565,driver<566 brand=vws,driver>=565,driver<566 brand=cloudgaming,driver>=565,driver<566 brand=unknown,driver>=570,driver<571 brand=grid,driver>=570,driver<571 brand=tesla,driver>=570,driver<571 brand=nvidia,driver>=570,driver<571 brand=quadro,driver>=570,driver<571 brand=quadrortx,driver>=570,driver<571 brand=nvidiartx,driver>=570,driver<571 brand=vapps,driver>=570,driver<571 brand=vpc,driver>=570,driver<571 brand=vcs,driver>=570,driver<571 brand=vws,driver>=570,driver<571 brand=cloudgaming,driver>=570,driver<571
TORCH_CUDA_ARCH_LIST=7.0 7.5 8.0 8.9 9.0 10.0 12.0
NVIDIA_DRIVER_CAPABILITIES=compute,utility
VLLM_USAGE_SOURCE=production-docker-image
CUDA_VERSION=12.9.1
VLLM_ENABLE_CUDA_COMPATIBILITY=0
LD_LIBRARY_PATH=/usr/local/nvidia/lib64:/usr/local/cuda/lib64:/usr/local/cuda/lib64
PYTORCH_NVML_BASED_CUDA_CHECK=1
TORCHINDUCTOR_COMPILE_THREADS=1
TORCHINDUCTOR_CACHE_DIR=/tmp/torchinductor_root

RAW_BUFFERClick to expand / collapse

Your current environment

<details> <summary>The output of <code>python collect_env.py</code></summary>

Your output of `python collect_env.py` here
==============================
        System Info
==============================
OS                           : Ubuntu 22.04.5 LTS (x86_64)
GCC version                  : (Ubuntu 11.4.0-1ubuntu1~22.04.3) 11.4.0
Clang version                : Could not collect
CMake version                : Could not collect
Libc version                 : glibc-2.35

==============================
       PyTorch Info
==============================
PyTorch version              : 2.11.0+cu129
Is debug build               : False
CUDA used to build PyTorch   : 12.9
ROCM used to build PyTorch   : N/A
XPU used to build PyTorch    : N/A

==============================
      Python Environment
==============================
Python version               : 3.12.13 (main, Mar  4 2026, 09:23:07) [GCC 11.4.0] (64-bit runtime)
Python platform              : Linux-5.15.0-119-generic-x86_64-with-glibc2.35
    
==============================
       CUDA / GPU Info
==============================
Is CUDA available            : True
CUDA runtime version         : 12.9.86
CUDA_MODULE_LOADING set to   : 
GPU models and configuration : 
GPU 0: NVIDIA H20-3e
GPU 1: NVIDIA H20-3e
GPU 2: NVIDIA H20-3e
GPU 3: NVIDIA H20-3e
GPU 4: NVIDIA H20-3e
GPU 5: NVIDIA H20-3e
GPU 6: NVIDIA H20-3e
GPU 7: NVIDIA H20-3e

Nvidia driver version        : 575.57.08
cuDNN version                : Could not collect
HIP runtime version          : N/A
MIOpen runtime version       : N/A
Is XNNPACK available         : True

==============================
          CPU Info
==============================
Architecture:                         x86_64
CPU op-mode(s):                       32-bit, 64-bit
Address sizes:                        46 bits physical, 57 bits virtual
Byte Order:                           Little Endian
CPU(s):                               192
On-line CPU(s) list:                  0-191
Vendor ID:                            GenuineIntel
Model name:                           INTEL(R) XEON(R) PLATINUM 8558P
CPU family:                           6
Model:                                207
Thread(s) per core:                   2
Core(s) per socket:                   48
Socket(s):                            2
Stepping:                             2
Frequency boost:                      enabled
CPU max MHz:                          2701.0000
CPU min MHz:                          800.0000
BogoMIPS:                             5400.00
Flags:                                fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cat_l2 cdp_l3 invpcid_single intel_ppin cdp_l2 ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local split_lock_detect avx_vnni avx512_bf16 wbnoinvd dtherm ida arat pln pts avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq la57 rdpid bus_lock_detect cldemote movdiri movdir64b enqcmd fsrm md_clear serialize tsxldtrk pconfig arch_lbr amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l1d arch_capabilities
Virtualization:                       VT-x
L1d cache:                            4.5 MiB (96 instances)
L1i cache:                            3 MiB (96 instances)
L2 cache:                             192 MiB (96 instances)
L3 cache:                             520 MiB (2 instances)
NUMA node(s):                         2
NUMA node0 CPU(s):                    0-47,96-143
NUMA node1 CPU(s):                    48-95,144-191
Vulnerability Gather data sampling:   Not affected
Vulnerability Itlb multihit:          Not affected
Vulnerability L1tf:                   Not affected
Vulnerability Mds:                    Not affected
Vulnerability Meltdown:               Not affected
Vulnerability Mmio stale data:        Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed:               Not affected
Vulnerability Spec rstack overflow:   Not affected
Vulnerability Spec store bypass:      Mitigation; Speculative Store Bypass disabled via prctl and seccomp
Vulnerability Spectre v1:             Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2:             Mitigation; Enhanced / Automatic IBRS; IBPB conditional; RSB filling; PBRSB-eIBRS SW sequence; BHI BHI_DIS_S
Vulnerability Srbds:                  Not affected
Vulnerability Tsx async abort:        Not affected

==============================
Versions of relevant libraries
==============================
[pip3] flashinfer-python==0.6.8.post1
[pip3] numpy==2.2.6
[pip3] nvidia-cublas-cu12==12.9.1.4
[pip3] nvidia-cuda-cupti-cu12==12.9.79
[pip3] nvidia-cuda-nvrtc-cu12==12.9.86
[pip3] nvidia-cuda-runtime-cu12==12.9.79
[pip3] nvidia-cudnn-cu12==9.17.1.4
[pip3] nvidia-cudnn-frontend==1.18.0
[pip3] nvidia-cufft-cu12==11.4.1.4
[pip3] nvidia-cufile-cu12==1.14.1.1
[pip3] nvidia-curand-cu12==10.3.10.19
[pip3] nvidia-cusolver-cu12==11.7.5.82
[pip3] nvidia-cusparse-cu12==12.5.10.65
[pip3] nvidia-cusparselt-cu12==0.7.1
[pip3] nvidia-cutlass-dsl==4.4.2
[pip3] nvidia-cutlass-dsl-libs-base==4.4.2
[pip3] nvidia-ml-py==13.595.45
[pip3] nvidia-nccl-cu12==2.28.9
[pip3] nvidia-nvjitlink-cu12==12.9.86
[pip3] nvidia-nvshmem-cu12==3.4.5
[pip3] nvidia-nvtx-cu12==12.9.79
[pip3] pyzmq==27.1.0
[pip3] torch==2.11.0+cu129
[pip3] torch_c_dlpack_ext==0.1.5
[pip3] torchaudio==2.11.0+cu129
[pip3] torchvision==0.26.0+cu129
[pip3] transformers==4.57.6
[pip3] triton==3.6.0
[conda] Could not collect

==============================
         vLLM Info
==============================
ROCM Version                 : Could not collect
vLLM Version                 : 0.1.dev15830+g8d599d76a (git sha: 8d599d76a)
vLLM Build Flags:
  CUDA Archs: 7.0 7.5 8.0 8.9 9.0 10.0 12.0; ROCm: Disabled; XPU: Disabled
GPU Topology:
        GPU0    GPU1    GPU2    GPU3    GPU4    GPU5    GPU6    GPU7    NIC0    NIC1    NIC2    NIC3    NIC4    NIC5    CPU Affinity    NUMA Affinity   GPU NUMA ID
GPU0     X      NV18    NV18    NV18    NV18    NV18    NV18    NV18    PIX     PIX     NODE    NODE    NODE    SYS     0-47,96-143     0               N/A
GPU1    NV18     X      NV18    NV18    NV18    NV18    NV18    NV18    PIX     PIX     NODE    NODE    NODE    SYS     0-47,96-143     0               N/A
GPU2    NV18    NV18     X      NV18    NV18    NV18    NV18    NV18    NODE    NODE    NODE    NODE    NODE    SYS     0-47,96-143     0               N/A
GPU3    NV18    NV18    NV18     X      NV18    NV18    NV18    NV18    NODE    NODE    NODE    NODE    NODE    SYS     0-47,96-143     0               N/A
GPU4    NV18    NV18    NV18    NV18     X      NV18    NV18    NV18    SYS     SYS     SYS     SYS     SYS     NODE    48-95,144-191   1               N/A
GPU5    NV18    NV18    NV18    NV18    NV18     X      NV18    NV18    SYS     SYS     SYS     SYS     SYS     NODE    48-95,144-191   1               N/A
GPU6    NV18    NV18    NV18    NV18    NV18    NV18     X      NV18    SYS     SYS     SYS     SYS     SYS     PIX     48-95,144-191   1               N/A
GPU7    NV18    NV18    NV18    NV18    NV18    NV18    NV18     X      SYS     SYS     SYS     SYS     SYS     PIX     48-95,144-191   1               N/A
NIC0    PIX     PIX     NODE    NODE    SYS     SYS     SYS     SYS      X      PIX     NODE    NODE    NODE    SYS                             
NIC1    PIX     PIX     NODE    NODE    SYS     SYS     SYS     SYS     PIX      X      NODE    NODE    NODE    SYS                             
NIC2    NODE    NODE    NODE    NODE    SYS     SYS     SYS     SYS     NODE    NODE     X      PIX     NODE    SYS                             
NIC3    NODE    NODE    NODE    NODE    SYS     SYS     SYS     SYS     NODE    NODE    PIX      X      NODE    SYS                             
NIC4    NODE    NODE    NODE    NODE    SYS     SYS     SYS     SYS     NODE    NODE    NODE    NODE     X      SYS                             
NIC5    SYS     SYS     SYS     SYS     NODE    NODE    PIX     PIX     SYS     SYS     SYS     SYS     SYS      X                              

Legend:

  X    = Self
  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
  PIX  = Connection traversing at most a single PCIe bridge
  NV#  = Connection traversing a bonded set of # NVLinks

NIC Legend:

  NIC0: mlx5_0
  NIC1: mlx5_1
  NIC2: mlx5_2
  NIC3: mlx5_3
  NIC4: mlx5_4
  NIC5: mlx5_5

==============================
     Environment Variables
==============================
NVIDIA_VISIBLE_DEVICES=all
NVIDIA_REQUIRE_CUDA=cuda>=12.9 brand=unknown,driver>=535,driver<536 brand=grid,driver>=535,driver<536 brand=tesla,driver>=535,driver<536 brand=nvidia,driver>=535,driver<536 brand=quadro,driver>=535,driver<536 brand=quadrortx,driver>=535,driver<536 brand=nvidiartx,driver>=535,driver<536 brand=vapps,driver>=535,driver<536 brand=vpc,driver>=535,driver<536 brand=vcs,driver>=535,driver<536 brand=vws,driver>=535,driver<536 brand=cloudgaming,driver>=535,driver<536 brand=unknown,driver>=550,driver<551 brand=grid,driver>=550,driver<551 brand=tesla,driver>=550,driver<551 brand=nvidia,driver>=550,driver<551 brand=quadro,driver>=550,driver<551 brand=quadrortx,driver>=550,driver<551 brand=nvidiartx,driver>=550,driver<551 brand=vapps,driver>=550,driver<551 brand=vpc,driver>=550,driver<551 brand=vcs,driver>=550,driver<551 brand=vws,driver>=550,driver<551 brand=cloudgaming,driver>=550,driver<551 brand=unknown,driver>=560,driver<561 brand=grid,driver>=560,driver<561 brand=tesla,driver>=560,driver<561 brand=nvidia,driver>=560,driver<561 brand=quadro,driver>=560,driver<561 brand=quadrortx,driver>=560,driver<561 brand=nvidiartx,driver>=560,driver<561 brand=vapps,driver>=560,driver<561 brand=vpc,driver>=560,driver<561 brand=vcs,driver>=560,driver<561 brand=vws,driver>=560,driver<561 brand=cloudgaming,driver>=560,driver<561 brand=unknown,driver>=565,driver<566 brand=grid,driver>=565,driver<566 brand=tesla,driver>=565,driver<566 brand=nvidia,driver>=565,driver<566 brand=quadro,driver>=565,driver<566 brand=quadrortx,driver>=565,driver<566 brand=nvidiartx,driver>=565,driver<566 brand=vapps,driver>=565,driver<566 brand=vpc,driver>=565,driver<566 brand=vcs,driver>=565,driver<566 brand=vws,driver>=565,driver<566 brand=cloudgaming,driver>=565,driver<566 brand=unknown,driver>=570,driver<571 brand=grid,driver>=570,driver<571 brand=tesla,driver>=570,driver<571 brand=nvidia,driver>=570,driver<571 brand=quadro,driver>=570,driver<571 brand=quadrortx,driver>=570,driver<571 brand=nvidiartx,driver>=570,driver<571 brand=vapps,driver>=570,driver<571 brand=vpc,driver>=570,driver<571 brand=vcs,driver>=570,driver<571 brand=vws,driver>=570,driver<571 brand=cloudgaming,driver>=570,driver<571
TORCH_CUDA_ARCH_LIST=7.0 7.5 8.0 8.9 9.0 10.0 12.0
NVIDIA_DRIVER_CAPABILITIES=compute,utility
VLLM_USAGE_SOURCE=production-docker-image
CUDA_VERSION=12.9.1
VLLM_ENABLE_CUDA_COMPATIBILITY=0
LD_LIBRARY_PATH=/usr/local/nvidia/lib64:/usr/local/cuda/lib64:/usr/local/cuda/lib64
PYTORCH_NVML_BASED_CUDA_CHECK=1
TORCHINDUCTOR_COMPILE_THREADS=1
TORCHINDUCTOR_CACHE_DIR=/tmp/torchinductor_root

</details>

🐛 Describe the bug

Title: DeepSeek V4 intermittently leaks DSML fragments in auto + streaming mode, causing unstable tool-call parsing

Description When serving DeepSeek V4 through vLLM OpenAI-compatible APIs, tool calling behaves inconsistently across modes:

tool_choice=required is generally stable and returns structured tool_calls.
tool_choice=auto with stream=true can intermittently leak DSML marker fragments into normal content.
After one leaked turn, subsequent turns are more likely to fail (conversation-state contamination risk).

Impact

Upstream agents may treat DSML fragments as normal assistant text.
Tool-state machines can desynchronize (tool branch expected, text branch taken).
The issue is intermittent in streaming mode, making it hard to detect and debug.

Reproduction Conditions

Model: DeepSeek V4 family
Server: vLLM OpenAI-compatible endpoint
Server flags include:
- --enable-auto-tool-choice
- --tool-call-parser deepseek_v4
Request parameters:
- non-empty tools
- tool_choice=auto
- stream=true
Repeating requests eventually shows DSML fragments in streamed content.

Observed Comparison

Switching to tool_choice=required significantly reduces or removes the issue.
Switching to stream=false significantly reduces or removes the issue.
This strongly suggests the problem is in the auto streaming parse path.

Expected Behavior

DSML markers should never be surfaced in streamed content, even when start markers span multiple chunks.
Parser should consistently output structured tool_calls.
One bad turn should not increase failure probability of later turns.

Actual Behavior

In auto + stream, if DSML start markers are split across chunks, partial marker text may be emitted as plain content before tool mode is recognized.
If clients append this leaked content into history, later turns become less stable.

Initial Root Cause Hypothesis

Parser emits plain text too early before confirming whether current chunk tail is part of a tool-start marker.
DeepSeek DSML start tags are long and chunk-splitting makes this edge case likely.
required follows a different streaming branch, which explains the behavioral difference.

Suggested Fix Direction

Add a prefix buffer before entering tool-call region.
Emit only content that is guaranteed not to be part of a split start marker.
As soon as a full start marker is recognized, switch to tool parsing immediately.
Reset buffer/state per request to prevent cross-request contamination.

Compatibility and Performance Notes

Scope: primarily affects auto + stream before tool region is confirmed.
Runtime overhead: small constant-time string prefix checks per chunk.
Memory overhead: tiny short-lived buffer (bounded by marker length).

Recommended Regression Tests

Start marker split across chunks at multiple boundaries.
Mixed plain-text prefix followed by tool call block.
Multi-turn requests to verify state reset.
Auto+stream vs required+stream comparison tests.

Environment

Date observed: 2026-04-24
Deployment: Linux server (vLLM), issue observed from agent/client side
API: OpenAI-compatible chat completions

Before submitting a new issue...

Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.

extent analysis

TL;DR

The issue can be fixed by implementing a prefix buffer to handle split DSML start markers in the auto streaming parse path.

Guidance

Review the current implementation of the tool-call-parser for deepseek_v4 to identify where the plain text is being emitted too early.
Consider adding a prefix buffer to store potential start marker text before confirming whether the current chunk tail is part of a tool-start marker.
Implement a mechanism to reset the buffer/state per request to prevent cross-request contamination.
Develop regression tests to cover different scenarios, such as start marker split across chunks, mixed plain-text prefix followed by tool call block, and multi-turn requests.

Example

No specific code example can be provided without more context about the current implementation. However, the fix direction suggests modifying the parser to include a buffer for handling split start markers.

Notes

The issue seems to be specific to the auto + stream mode, and the suggested fix direction provides a clear approach to addressing the problem. The implementation details will depend on the specific codebase and requirements.

Recommendation

Apply the suggested fix direction by adding a prefix buffer and modifying the parser to handle split start markers, as this approach addresses the root cause of the issue and provides a clear solution.

Vote matrix · Quick signals

Works

Did the solution work? Tap to confirm.

Easy Fix

Was it a quick fix?

Time Saver

Did it save you time?

Blocking

Was it severely blocking?

Common Issue

Are others likely hitting this too?

Flaky / Intermittent

Is it intermittent?

Verified / Reproducible

Can you reproduce it reliably?

#api #ssr #API rate limit #retriever error #environment variable

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Data

Security

Network

Code

UI/UX

Text

System

Multimedia

Protocol

API

Engineering

vllm - ✅(Solved) Fix [Bug]: Title: DeepSeek V4 intermittently leaks DSML fragments in auto + streaming mode, causing unstable tool-call parsing [2 pull requests, 1 participants]

Recommended Tools

GitHub issue graph ai analysis

Root Cause

Fix Action

Fix / Workaround

============================== CPU Info

PR fix notes

PR #40760: [New Model] Support DeepseekV4

Description (problem / solution / changelog)

Changed files

PR #40805: fix/deepseekv4: prevent DSML marker leakage in streaming auto mode(v32 parser only)

Description (problem / solution / changelog)

Purpose

Test Plan

Test Result

Changed files

Code Example

Your current environment

🐛 Describe the bug

Before submitting a new issue...

extent analysis

TL;DR

Guidance

Example

Notes

Recommendation

Still need to ship something?

TRENDING

vllm - ✅(Solved) Fix [Bug]: Title: DeepSeek V4 intermittently leaks DSML fragments in auto + streaming mode, causing unstable tool-call parsing [2 pull requests, 1 participants]

Recommended Tools

GitHub issue graph ai analysis

Root Cause

Fix Action

Fix / Workaround

============================== CPU Info

PR fix notes

PR #40760: [New Model] Support DeepseekV4

Description (problem / solution / changelog)

Changed files

PR #40805: fix/deepseekv4: prevent DSML marker leakage in streaming auto mode(v32 parser only)

Description (problem / solution / changelog)

Purpose

Test Plan

Test Result

Changed files

Code Example

Your current environment

🐛 Describe the bug

Before submitting a new issue...

extent analysis

TL;DR

Guidance

Example

Notes

Recommendation

Still need to ship something?

RELATED_DISCOVERY

TRENDING