Skip to content

Pull requests: vllm-project/vllm

New pull request New
3,389 Open 25,661 Closed
3,389 Open 25,661 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

[CI] Run pre-commit job on self-hosted runners ci/build
#45865 opened Jun 16, 2026 by khluu Member Loading…
2 tasks
Add extensible engine core outputs for OOT backends v1
#45864 opened Jun 16, 2026 by maxdebayser Contributor Loading…
[DSv4 Perf] DSv4 flashinfer sparse index cache for metadata, 2%~4% TTFT improvement nvidia ready ONLY add when PR is ready to merge/full CI is needed v1
#45863 opened Jun 16, 2026 by yewentao256 Member Loading…
1
[Bugfix] DeepSeek V4: recover tool calls within reasoning content bug Something isn't working deepseek Related to DeepSeek models tool-calling
#45862 opened Jun 16, 2026 by procr1337 Loading…
3 of 4 tasks
2
[Bugfix] Fix hetero TP handshake assertion for GQA-replicated KV heads bug Something isn't working kv-connector v1
#45860 opened Jun 16, 2026 by kannakAWS Loading…
3 of 4 tasks
1
3
[Log] Update deepgemm log ready ONLY add when PR is ready to merge/full CI is needed
#45857 opened Jun 16, 2026 by yewentao256 Member Loading…
[Kernel] Manual TP fusion via ResidualStream (AR+RMSNorm+quant) llama Related to Llama models mistral Related to Mistral models qwen Related to Qwen models speculative-decoding
#45855 opened Jun 16, 2026 by mgoin Member • Draft
[ROCm][Quant] Minimax-M3: Enable fp8_per_channel for bf16 weights on mi300x rocm Related to AMD ROCm
#45854 opened Jun 16, 2026 by hongxiayang Collaborator Loading…
4 tasks
[Bugfix][Gemma4] Pre-initialise streaming reasoning state when prompt ends inside an open <|channel> (fixes #45834) bug Something isn't working tool-calling
#45852 opened Jun 16, 2026 by nikhilesh-csa Loading…
3 of 4 tasks
1
10
[KV Offload] Use background thread for mmap / cpu_tensors pinning v1
#45850 opened Jun 16, 2026 by Acaciasama Loading…
3 of 4 tasks
2
[BUG] fix hidden states nan for hybrid attention models bug Something isn't working kv-connector ready ONLY add when PR is ready to merge/full CI is needed v1
#45849 opened Jun 16, 2026 by shanjiaz Contributor Loading…
3 of 4 tasks
1
1
[rust] EngineCoreSamplingParams: add serde defaults for omit_defaults fields rust
#45848 opened Jun 16, 2026 by wseaton Contributor Loading…
Skip stop string matching inside <think> blocks frontend v1
#45846 opened Jun 16, 2026 by elvircrn Contributor • Draft
4 tasks
2
[v1][kvcache] Honor prefix-cache retention interval for Mamba/linear attention ready ONLY add when PR is ready to merge/full CI is needed v1
#45845 opened Jun 16, 2026 by Dao007forever Contributor Loading…
3
[Bugfix] Fix CPU split-KV scratchpad sizing bug Something isn't working cpu Related to CPU backends
#45844 opened Jun 16, 2026 by gausah01 Contributor Loading…
[CI][NIXL] Pin NIXL to 1.2.0 ci/build kv-connector ready ONLY add when PR is ready to merge/full CI is needed
#45843 opened Jun 16, 2026 by itayalroy Contributor Loading…
2
Bump the minor-update group across 1 directory with 150 updates ci/build dependencies Pull requests that update a dependency file nvidia rocm Related to AMD ROCm
#45842 opened Jun 16, 2026 by dependabot Bot Loading…
add attention sinks to flex attention documentation Improvements or additions to documentation v1
#45841 opened Jun 16, 2026 by liangel-02 Contributor Loading…
1
[Perf] Skip/shrink all_token_ids copy in scheduler for non-async and V2 runner v1
#45840 opened Jun 16, 2026 by amanchugh89 Loading…
2
[Frontend] Support additional sampling parameters for translation API frontend
#45839 opened Jun 16, 2026 by guan404ming Contributor Loading…
3 of 4 tasks
Minimax m3 gfx950 mxfp4
#45838 opened Jun 16, 2026 by dllehr-amd Collaborator • Draft
4 tasks
[NVFP4 MoE/DSV4] Marlin: wire SwiGLU clamp + allow it for clamped models on non-Blackwell
#45836 opened Jun 16, 2026 by mikekg Contributor Loading…
[Bugfix] Restore unloaded FP8 scale params during layerwise reload bug Something isn't working
#45835 opened Jun 16, 2026 by aoshen02 Collaborator • Draft
3 tasks
[RFC][Core][Model] Voxtral realtime: unbounded-duration streaming via RoPE re-anchoring (experimental, default-off) documentation Improvements or additions to documentation mistral Related to Mistral models multi-modality Related to multi-modality (#4194) performance Performance-related issues v1
#45833 opened Jun 16, 2026 by damienlaine • Draft
2
[Bugfix][Gemma4] Fix parsing when thinking is disabled bug Something isn't working ready ONLY add when PR is ready to merge/full CI is needed tool-calling
#45832 opened Jun 16, 2026 by m4r1k Contributor Loading…
5
ProTip! Exclude everything labeled bug with -label:bug.