Skip to content

Pull requests: vllm-project/vllm

New pull request New
3,743 Open 27,387 Closed
3,743 Open 27,387 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

[KV Connector] Fragment-major canonical CPU layout for KV offload kv-connector v1
#48414 opened Jul 12, 2026 by Etelis Contributor • Draft
[Bugfix][MM] Fix MiniCPM-V placeholder replacement and image processor loading on Transformers v5 bug Something isn't working
#48413 opened Jul 12, 2026 by YunzhuLu Contributor Loading…
4 tasks
1
[Bugfix] Include inline per-token-head scales in offloaded page transfer width bug Something isn't working kv-connector v1
#48411 opened Jul 12, 2026 by Etelis Contributor Loading…
[Model] Apertus 1.5 multimodality new-model Requests to new models
#48410 opened Jul 12, 2026 by blancsw Contributor Loading…
3 of 4 tasks
3
Add runtime LoRA weight updates frontend v1
#48409 opened Jul 12, 2026 by aoshen02 Collaborator • Draft
[KV Connector] Add per-layer KV head region schema for TP-agnostic offload kv-connector v1
#48408 opened Jul 12, 2026 by Etelis Contributor • Draft
[Attention] Skip sparse indexer for short sequences nvidia v1
#48407 opened Jul 12, 2026 by qianlihuang Contributor • Draft
[Bugfix] Size the sparse-indexer expanded block table from the runner's block-table width bug Something isn't working v1
#48404 opened Jul 12, 2026 by drakosha Contributor Loading…
1
[Model] Add EAGLE-3 support for SeedOssForCausalLM
#48403 opened Jul 12, 2026 by notwitcheer Loading…
1
[Bugfix] Warn when prefix caching is inert on hybrid models (block_size > max_model_len) bug Something isn't working
#48402 opened Jul 12, 2026 by Majid-Taheri Contributor Loading…
1
[Core] Simplify KVBlockZeroer index tensor handling ready ONLY add when PR is ready to merge/full CI is needed v1
#48399 opened Jul 12, 2026 by njhill Member Loading…
[Bugfix] Fix int32 address overflow in DeepGEMM ep_gather Triton kernel bug Something isn't working
#48398 opened Jul 12, 2026 by raskin1 Loading…
1
[Spec Decode][V1] Warm _mtp_shared_head_rmsnorm_kernel at startup (stacked on #41481) speculative-decoding v1
#48393 opened Jul 12, 2026 by leihuang-sketch Loading…
1
[Spec Decode] DFlash/DSpark draft support under decode context parallelism qwen Related to Qwen models speculative-decoding v1
#48392 opened Jul 12, 2026 by drakosha Contributor • Draft
Fix/fused add rms norm batch invariant block size v1
#48391 opened Jul 12, 2026 by oops-oom Loading…
4
[Core] Support fp32 lm_head for generation models via head_dtype (RFC #48305 §3.6) v1 verified Run pre-commit for new contributors without triggering other tests
#48390 opened Jul 12, 2026 by KKothuri • Draft
1
12
[V1] Add device-neutral KVConnector host-contract mixins kv-connector v1
#48389 opened Jul 12, 2026 by lokic233 Loading…
1
[CI][AMD] Configure MI300 tests for native execution without DinD ci/build rocm Related to AMD ROCm
#48387 opened Jul 12, 2026 by AndreasKaratzas Member • Draft
fix(reasoning): recompute ernie45 response-end index after stripping prefix tool-calling
#48386 opened Jul 12, 2026 by chuenchen309 Loading…
1
add pad-aware reduce path
#48385 opened Jul 12, 2026 by gnovack Contributor • Draft
[chore] remove kernel_block_sizes check in get_preferred_block_size v1
#48384 opened Jul 12, 2026 by andyxning Contributor Loading…
4 tasks
[Perf][Model Loader] Reload layout-preserving weights directly
#48382 opened Jul 12, 2026 by aoshen02 Collaborator • Draft
[Bugfix][Spec Decode] DSpark: store draft KV cache in model dtype for MLA-only cache layouts; fail fast on DCP bug Something isn't working v1
#48381 opened Jul 12, 2026 by drakosha Contributor Loading…
1
[Bugfix] Set kv_quant_mode on the generic MLA KV-cache spec bug Something isn't working
#48379 opened Jul 12, 2026 by drakosha Contributor Loading…
1
Revert "[2/N][KV-Cache Layout Refactor] Pack K/V into the content dim across attention backends" (#44455) kv-connector nvidia rocm Related to AMD ROCm v1
#48377 opened Jul 12, 2026 by vllm-agent Contributor • Draft
ProTip! Type g i on any issue or pull request to go back to the issue listing page.