Skip to content

Pull requests: vllm-project/vllm

New pull request New
4,834 Open 30,575 Closed
4,834 Open 30,575 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

[Revert] MXFP4 + block-FP8 Humming support after H100 correctness failure quantization
#53805 opened Aug 25, 2026 by khluu Member • Draft
1
[XPU] Add env overrides for DiffKV attention 2D launch and tile size intel-gpu Related to Intel GPU
#53804 opened Aug 25, 2026 by slokesha Loading…
4 tasks done
1
Fix mamba align retention checkpoints kv-cache-manager mrv2 Model Runner V2 specific
#53803 opened Aug 25, 2026 by ptorsten Loading…
1
[Bugfix] Align hybrid prefix-cache hit boundaries bug Something isn't working kv-cache-manager scheduler
#53802 opened Aug 25, 2026 by ptorsten Loading…
1
[Bugfix] Close the socket test_loopback_bind creates when bind fails bug Something isn't working
#53801 opened Aug 25, 2026 by eyad6789 Loading…
1
fix MI455 customed paged attn rocm Related to AMD ROCm
#53800 opened Aug 25, 2026 by Concurrensee Contributor • Draft
[Bugfix][Scheduler] Handle missing req_id in update_from_output gracefully bug Something isn't working scheduler
#53799 opened Aug 25, 2026 by dhakshin32 Loading…
2
[Bugfix] Seed align-mode mamba state_idx in mamba block units bug Something isn't working mrv2 Model Runner V2 specific
#53798 opened Aug 25, 2026 by ptorsten Loading…
1
Add support for loading dflash2 model in speculators format dflash
#53797 opened Aug 25, 2026 by fynnsu Contributor Loading…
3 of 4 tasks
[Core][Frontend] Add per-request prefix cache telemetry documentation Improvements or additions to documentation frontend kv-cache-manager scheduler
#53795 opened Aug 25, 2026 by cook1e-0707 Loading…
6 tasks done
5
[Perf] Add tuning configs for SM103 Kimi k3 nvfp4 DEP projection GEMM k3 kimi
#53794 opened Aug 25, 2026 by wzhao18 Contributor • Draft
4 tasks
Fuse ReLU2 with static FP8 activation quantization quantization
#53793 opened Aug 25, 2026 by samnordmann Contributor • Draft
[ROCm] Resolve the indexer fp8 cache dtype once at import rocm Related to AMD ROCm
#53792 opened Aug 25, 2026 by amd-sriram Loading…
1
[RFC] Fuse Nemotron shared ReLU2 and static FP8 quantization quantization
#53791 opened Aug 25, 2026 by samnordmann Contributor • Draft
[Bugfix] NemotronHMTP: add hf_to_vllm_mapper so quant exclusions reach the MTP draft bug Something isn't working quantization
#53790 opened Aug 25, 2026 by juhi10071998 Contributor Loading…
[ROCm][Perf] Bound the decode paged-MQA-logits sanitize to the top-k read region rocm Related to AMD ROCm
#53789 opened Aug 25, 2026 by amd-sriram Loading…
1
[Do Not Merge][Attention] Enable masked MHA for GLM-5 head dimensions ci/build glm performance Performance-related issues
#53785 opened Aug 25, 2026 by MatthewBonanni Member Loading…
[Distributed] Support pre-shared ncclUniqueId rendezvous for weight transfer
#53784 opened Aug 25, 2026 by dharak-cohere • Draft
1
[4/N] Expose HiSparse cache metrics kv-connector mrv2 Model Runner V2 specific scheduler
#53782 opened Aug 25, 2026 by MatthewBonanni Member • 4/6 Loading…
[3/N] HiSparse: host-resident sparse-MLA decode hot-buffering ci/build cpu Related to CPU backends deepseek Related to DeepSeek models documentation Improvements or additions to documentation DSv4 glm kv-cache-manager kv-connector minimax mrv2 Model Runner V2 specific nvidia ready ONLY add when PR is ready to merge/full CI is needed scheduler v1
#53781 opened Aug 25, 2026 by MatthewBonanni Member • 3/6 Loading…
4
[2/N][KV Connector][NIXL] Support per-region transfer geometry kv-connector ready ONLY add when PR is ready to merge/full CI is needed
#53780 opened Aug 25, 2026 by MatthewBonanni Member • 2/6 Loading…
2
[1/N][KV Connector] Identify externally transferable KV cache groups kv-cache-manager kv-connector ready ONLY add when PR is ready to merge/full CI is needed
#53779 opened Aug 25, 2026 by MatthewBonanni Member • 1/6 Loading…
2
[Build] Restore generic wheel for no_device targets ci/build
#53778 opened Aug 25, 2026 by Silv3S Loading…
3 of 4 tasks
1
[Bugfix][XPU] Fix async output deadlock from blocking cross-stream event bug Something isn't working intel-gpu Related to Intel GPU
#53776 opened Aug 25, 2026 by yeonsily Loading…
3 of 4 tasks
1
[Revert] Restore per-group K3 Mamba metadata preparation k3 kimi mrv2 Model Runner V2 specific
#53774 opened Aug 25, 2026 by khluu Member • Draft
6
ProTip! Filter pull requests by the default branch with base:main.