vllm
https://github.com/vllm-project/vllm
Python
A high-throughput and memory-efficient inference and serving engine for LLMs
Triage Issues!
When you volunteer to triage issues, you'll receive an email each day with a link to an open issue that needs help in this project. You'll also receive instructions on how to triage issues.
Triage Docs!
Receive a documented method or class from your favorite GitHub repos in your inbox every day. If you're really pro, receive undocumented methods or classes and supercharge your commit history.
Python not yet supported51 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [ModelOpt] Support fused MoE weight reload
- [Test Coverage] Add tests for MoE dispatch and Triton qzeros reshape
- [Bugfix] Avoid MCP events for Harmony browser calls
- [CI][Docs] Fix decode/prefill consistency test prefix construction; add GLM-4-9B and Phi-4 to batch-invariant tested models
- [Bugfix][Parser] Terminate streamed tool arguments the flush cannot extend
- [Kernel][Perf] Route non-spec GDN decode through the fused CUDA kernel
- [DCP] Add DCP qrep to GLM-5.2 flat model definition
- [Bug]: structured outputs never enforced for Muse Glimmer with --reasoning-parser muse_glimmer (json_object too; duplicate of #52594)
- [Performance]: v0.23→v0.26: ~20% decode ITL regression on non-gated (relu2) BF16 MoE (SM100) — TRTLLM-gen backend swap (#43853) adds per-call host cost; --moe-backend flashinfer_cutlass recovers ~90%
- [KV Offload] Size node-local shared regions by local_world_size
- Docs
- Python not yet supported