sglang
https://github.com/sgl-project/sglang
Python
SGLang is a high-performance serving framework for large language models and multimodal models.
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 supported1 Subscribers
Add a CodeTriage badge to sglang
Help out
- Issues
- Fix requested-token logprob handoff in P/D serving
- fix(speculative): prevent multi-gpu deadlock by rejecting grammar requests with DSpark + TP>1
- Fix reasoning token usage past speculative finish boundaries
- [Bug] DSpark + TP: grammar-constrained request batched with any other request deadlocks all ranks (overlap and non-overlap), GPUs spin in a collective
- [AMD][gfx1250] Retune sparse paged-prefill launch config for DSV4-Flash
- [CPU] Part 2: Add DeepSeek3.x/GLM5.x support and optimizations on CPU
- Support token ids in chat completion streams
- [AMD] Pick the MoRI dispatch format from the MoE's input contract
- [Bug] NVFP4 online MoE loading can use overwritten RunAI streamer tensors
- [Bug] GLM-5.3-Flash RunAI streaming can corrupt fused MLA weights
- Docs
- Python not yet supported