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
- [Perf] sampling: Draw final samples in vocabulary blocks
- [Scheduler] Add opt-in shared schedule/forward stream for the overlap scheduler
- [DCP] HiCache L2 for DSA models with MTP under decode context parallelism
- [Fix] grpc: Add option to use remaining context
- Skip empty scheduler time stats in streamed outputs
- [Fix] grpc: Preserve max_new_tokens three-state semantics
- [Fix] deepseek-v4.1: Guard bounded replay in the breakable prefill graph off CUDA and under prefill CP
- [Feature] Carry the next-token embedding row in the distributed sampling exchange (no separate embedding collective)
- [Bug][Speculative] EAGLE/MTP causes 90% KV cache starvation on models with untied embeddings (Fix Included)
- [Frontend] Expose JSON and SSE completion schemas in OpenAPI
- Docs
- Python not yet supported