Commits

Branch selector

main

User selector

All users

All time

Commit history

Commits on Jan 20, 2026

• [Metrics] Complete removal of deprecated vllm:time_per_output_token_seconds metric (#32661)

  Show description for bb91720
  

  carlory

  and

  claude

  authored
  bb91720

  Copy full SHA for bb91720

• [Bugfix] Fix the fp8_mqa_logits dim mismatch (#32652)

  Show description for c4e5bdf
  chaunceyjiang

  authored
  c4e5bdf

  Copy full SHA for c4e5bdf

• [3/N] Initialize MM components in context managers (I-L) (#32650)

  Show description for 7f1bcd1
  DarkLight1337

  authored
  7f1bcd1

  Copy full SHA for 7f1bcd1

• [Core] Cleanup shm based object store on engine shutdown (#32429)

  Show description for 8be263c
  walterbm

  authored
  8be263c

  Copy full SHA for 8be263c

• [2/N] Initialize MM components in context managers (E-H) (#32641)

  Show description for e1a34c3
  DarkLight1337

  authored
  e1a34c3

  Copy full SHA for e1a34c3

• [Refactor] Make FP8 Linear Ops use kernel abstraction (#27814)

  Show description for 148117e
  vllmellm

  authored
  148117e

  Copy full SHA for 148117e

• [Model Runner V2] Skip kernel launch for penalties & logit_bias (#32634)

  Show description for e9c83cd
  WoosukKwon

  authored
  e9c83cd

  Copy full SHA for e9c83cd

• [1/N] Initialize MM components in context managers (A-D) (#32632)

  Show description for b75e85d
  DarkLight1337

  authored
  b75e85d

  Copy full SHA for b75e85d

• [Model] Use context managers for encoder- and LM-only mode (#32605)

  Show description for 4753f3b
  DarkLight1337

  authored
  4753f3b

  Copy full SHA for 4753f3b

• [Model Runner V2] Decouple temperature from penalties (#32629)

  Show description for 6c01ffb
  WoosukKwon

  authored
  6c01ffb

  Copy full SHA for 6c01ffb

• [Model Runner V2] Refactor get_cudagraph_and_dp_padding (#32625)

  Show description for 7b7cdce
  WoosukKwon

  authored
  7b7cdce

  Copy full SHA for 7b7cdce

• [Feat] allow inplace loading lora (#31326)

  Show description for 12dab78
  

  Jackmin801

  and

  jeejeelee

  authored
  12dab78

  Copy full SHA for 12dab78

• [Model Runner V2] Initialized communication buffer for DP (#32624)

  Show description for 05dc4bf
  WoosukKwon

  authored
  05dc4bf

  Copy full SHA for 05dc4bf

Commits on Jan 19, 2026

• [Attention][MLA] Make FLASHINFER_MLA the default MLA backend on Blackwell, and TRTLLM the default prefill (#32615)

  Show description for 1a1fc3b
  

  MatthewBonanni

  and

  yewentao256

  authored
  1a1fc3b

  Copy full SHA for 1a1fc3b

• [Model Runner V2] Refactor `dummy_run` (#32533)

  Show description for 43fada5
  WoosukKwon

  authored
  43fada5

  Copy full SHA for 43fada5

• feat: spec decode with draft models (#24322)

  Show description for 4a5299c
  tomasruizt

  authored
  4a5299c

  Copy full SHA for 4a5299c

• docs: prefix caching seems quite outdated (#28784)

  Show description for 73f2a81
  

  3 peopleauthored
  73f2a81

  Copy full SHA for 73f2a81

• [BugFix] Fix TRT-LLM NVFP4 DP/EP (#32349)

  Show description for 7350331
  

  jiahanc

  and

  Robert Shaw

  authored
  7350331

  Copy full SHA for 7350331

• [CI] Add Helion as an optional dependency (#32482)

  Show description for 9d1e611
  gmagogsfm

  authored
  9d1e611

  Copy full SHA for 9d1e611

• [BUGFIX] Fix `test_mla_backends.py`. Scale MLA projection weights to prevent numerical instability (#32529)

  Show description for 0727cc9
  vadiklyutiy

  authored
  0727cc9

  Copy full SHA for 0727cc9

• [CI][amd] Revert NIXL connector change to avoid crash (#32570)

  Show description for a0490be
  qli88

  authored
  a0490be

  Copy full SHA for a0490be

• support dynamic resolution image encoding for Nemotron Nano VL (#32121)

  Show description for cd3ac5b
  netanel-haber

  authored
  cd3ac5b

  Copy full SHA for cd3ac5b

• [Misc] Remove unused ModelKeys (#32608)

  Show description for 2636d76
  jeejeelee

  authored
  2636d76

  Copy full SHA for 2636d76

• Add support for LoRA adapters in Nemotron-H models (#30802)

  Show description for aa7f37c
  danisereb

  authored
  aa7f37c

  Copy full SHA for aa7f37c

• [Frontend] Score entrypoint support data_1 & data_2 and queries & documents as inputs (#32577)

  Show description for c88860d
  noooop

  authored
  c88860d

  Copy full SHA for c88860d

• [NIXL][Metrics] Track `nixl_num_kv_expired_reqs` metric in Prometheus (#32340)

  Show description for 758df5a
  NickLucche

  authored
  758df5a

  Copy full SHA for 758df5a

• [CI/Build] Fix dependency conflict between model-hosting-container-standards and starlette (#32560)

  Show description for cdd03d2
  

  DanielMe

  and

  Daniel Mescheder

  authored
  cdd03d2

  Copy full SHA for cdd03d2

• [Core] Whisper support `torch.compile` (#30385)

  Show description for 74c583b
  NickLucche

  authored
  74c583b

  Copy full SHA for 74c583b

• [ROCm][CI] Add ROCm attention backend support for EAGLE DP tests (#32363)

  Show description for c0a350c
  AndreasKaratzas

  authored
  c0a350c

  Copy full SHA for c0a350c

• [GLM-4.7] GLM Model support for GLM-Lite (#31386)

  Show description for 71832ba
  zRzRzRzRzRzRzR

  authored
  71832ba

  Copy full SHA for 71832ba

• [CI][Hardware][AMD] Fix test_rotary_embedding_mla_cache_fused (#32408)

  Show description for 11bbf86
  mawong-amd

  authored
  11bbf86

  Copy full SHA for 11bbf86

• [Frontend] Add render endpoints for prompt preprocessing (#32473)

  Show description for 3c8740a
  

  hyeongyun0916

  and

  gemini-code-assist[bot]

  authored
  3c8740a

  Copy full SHA for 3c8740a

• [CI/Build] Use Common Event Map Fixture in Harmony / MCP Server Tests (#32531)

  Show description for 7518a3d
  alex-jw-brooks

  authored
  7518a3d

  Copy full SHA for 7518a3d

• [BugFix] Fix embed_input_ids argument error of QwenVLForConditionalGeneration (#32462)

  honglyua-il

  authored
  976af2f

  Copy full SHA for 976af2f

• [Model Runner V2] Refactor `update_states` (#32562)

  Show description for 9a1f16d
  WoosukKwon

  authored
  9a1f16d

  Copy full SHA for 9a1f16d