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
Add a CodeTriage badge to vllm
Help out
- Issues
- [Attention Backend] Select split-K per shape in HPC Attention Decode Backend, add integration tests
- [Rust Frontend]: Added olmo3 reasoning parser
- fix(security): bound MRoPE temporal positions to prevent engine crash
- [Bug]: Deploying the GLM5.2-nvfp4 model using the 0.24.0 image , the model occasionally outputs "!!!!!!!!!!!!!!!!!" in the thinking phase,
- [Bug]: Qwen3.6 hybrid model with prefix caching + MTP3 causes tool-call leakage and needle recall failure, while no-MTP path is correct[错误]：
- [Performance]: Qwen3.5 native MTP can be slower than no-MTP CUDA graph baseline despite good acceptance
- [EC Transfer] Add MooncakeStoreECConnector for multimodal hidden-state transfer
- [Model] Add support for OpenPangu V2 model
- [RFC]: Qwen3-ASR pseudo-streaming can repeat or miss segments
- [Bug] SimpleCPUOffload: residual output garbling under high concurrency — load path lacks cross-stream sync with compute (WAR), symmetric gap left by #46278
- Docs
- Python not yet supported