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
- [Bug] Non-positive n (n=0 or negative) crashes batch /generate with IndexError instead of a 400
- [Fix] io_struct: Reject non-positive parallel sampling count n in request normalization
- [Perf] diffusion: Extend NVENC video encoding to Wan2.1, LTX-2, SANA-WM, and MiniMax-H3
- [Doc] cookbook: Correct the GLM-5.3-Flash note on prefill CP with PD disaggregation
- [Fix] lang: Avoid mutating StreamExecutor.messages_ in Anthropic backend and pop(0) loop in VertexAI
- [Fix] function_call: stream multiple DeepSeek V3.1 tool calls correctly
- [Feat] engine: Accept positional_embed_overrides on generate and async_generate
- [Feat] EPD: per-item encoder prefix cache for batched multimodal requests
- [Bug] Breakable prefill CUDA graph: short prefills padded to a capture bucket give different logprobs than eager (Qwen3.8-27B-NVFP4, hybrid GDN, v0.5.23)
- [Feat] observability: report token usage on the request span under the current GenAI conventions
- Docs
- Python not yet supported