Skip to content

Pull requests: vllm-project/vllm

New pull request New
5,221 Open 31,444 Closed
5,221 Open 31,444 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

[Bugfix] Rank NVFP4 CuteDSL W4A16 kernel below the native W4A4 kernels bug Something isn't working
#55405 opened Sep 4, 2026 by iharshlalakiya Loading…
3 of 4 tasks
1
5
[Perf][GDN] Build cudagraph-capture metadata without a device sync k3 kimi nvidia
#55404 opened Sep 4, 2026 by njhill Member Loading…
4
[Hybrid] Block priorities enabling opportunistic APC hits with sparse caching kv-cache-manager
#55403 opened Sep 4, 2026 by s3woz Contributor • Draft
1
[Docs] Remove dead snippet includes left behind by refactors documentation Improvements or additions to documentation
#55402 opened Sep 4, 2026 by simpleqt Loading…
2 tasks done
5
[Bugfix][Core] Fix logging calls whose arguments do not match their placeholders bug Something isn't working kv-connector
#55401 opened Sep 4, 2026 by simpleqt Loading…
4
[Feature][Core][Rust Frontend] Add a Rejected finish reason for engine-stated request rejections rust
#55399 opened Sep 4, 2026 by FeathBow Contributor Loading…
1
[5/N] HiSparse: support direct GPU landing for P/D transfers kv-cache-manager kv-connector scheduler
#55398 opened Sep 4, 2026 by MatthewBonanni Member • 4/6 • Draft
1
[Bugfix][Parser] Hide incomplete Muse Glimmer ATEM markers bug Something isn't working tool-calling
#55396 opened Sep 4, 2026 by lming2001 • Draft
3 of 4 tasks
1
2
[Bugfix] do not treat weight filenames as custom proposer paths bug Something isn't working
#55393 opened Sep 4, 2026 by Chessing234 Loading…
1 task done
5
Revert "[CI] Remove deleted nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16 and its arch aliases" nvidia ready ONLY add when PR is ready to merge/full CI is needed
#55392 opened Sep 4, 2026 by mgoin Member Loading…
3
[Bugfix] reject empty prompt lists in CompletionRequest bug Something isn't working frontend
#55391 opened Sep 4, 2026 by Chessing234 Loading…
1 task done
2
[Bugfix] Annotate MTP draft KV cache groups positionally on the hybrid grouping path bug Something isn't working kv-cache-manager verified Run pre-commit for new contributors without triggering other tests
#55390 opened Sep 4, 2026 by Navjot10 Loading…
2
[Model] Extend device-side mm normalization to GLM4V/GLM5Next glm
#55389 opened Sep 4, 2026 by cjackal Contributor Loading…
4 tasks done
1
[Bugfix] error cleanly when vllm serve --model has no value bug Something isn't working
#55388 opened Sep 4, 2026 by Chessing234 Loading…
1 task done
7
[CI/Build] Move the default CUDA build from 13.0 to 13.2 ci/build documentation Improvements or additions to documentation nvidia
#55387 opened Sep 4, 2026 by atalman Contributor Loading…
6
[Bugfix][Compile] Fold LoRA wrap state into the AOT cache key bug Something isn't working torch.compile
#55386 opened Sep 4, 2026 by he-yufeng Contributor Loading…
1
1
[perf] wire FA and FlashMLA for sm90 GLM5Next NoPE SparseMLA glm
#55385 opened Sep 4, 2026 by JaredforReal Contributor • Draft
1
[Bugfix] Share resolved KV cache layout across config copies bug Something isn't working kv-cache-manager kv-connector nvidia ready ONLY add when PR is ready to merge/full CI is needed
#55384 opened Sep 4, 2026 by LucasWilkinson Collaborator Loading…
3
[Turing] Restore flashinfer support for SM75 nvidia
#55380 opened Sep 4, 2026 by ir1ka Contributor Loading…
6 of 7 tasks
4
[CI] Disable Ascend NPU test ci/build
#55379 opened Sep 4, 2026 by khluu Member Loading…
1
[ROCm] Hold the router weight in fp32 when fp32 routing is requested deepseek Related to DeepSeek models rocm Related to AMD ROCm
#55378 opened Sep 4, 2026 by stefanskiasan Loading…
1
2
[Bugfix] Autotune FlashInfer deferred MoE decode kernels before CUDA graph capture bug Something isn't working nvidia
#55377 opened Sep 4, 2026 by mingg26 Contributor Loading…
2
[Build] Remove obsolete TPU Dockerfile ci/build ready ONLY add when PR is ready to merge/full CI is needed
#55376 opened Sep 4, 2026 by WoosukKwon Collaborator Loading…
4 tasks done
3
[Bugfix][Qwen4Exp] fix state index strides in fused PLE conv bug Something isn't working qwen Related to Qwen models
#55375 opened Sep 4, 2026 by peakcrosser7 Contributor Loading…
3 of 4 tasks
3
[KV Connector] Support piecewise prefix loading with NIXL connectors kv-connector
#55374 opened Sep 4, 2026 by z-zanez Contributor Loading…
3 tasks done
4
ProTip! Add no:assignee to see everything that’s not assigned.