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: preserve text before DeepSeek V3.1 streaming tool calls
- [sgl-router] Serve the Anthropic Messages API on /v1/messages
- [Bug] DeepSeek V3 tool-call parser streams arguments "{}" when the closing fence and end tokens arrive in one stream event (MTP / --stream-interval > 1)
- fix(function_call): stream DeepSeek V3 tool arguments that arrive with the name
- [diffusion] fix: report the port each scheduler actually bound to its clients
- fix(attention): dispatch hybrid-model full-attn layers by call form in FA backend
- [Fix] Preserve scalar EOS token 0 from generation config
- [Perf] Reuse opaque metadata during multi-worker output fanout
- [sgl-router] Add API profiles: a vendor API contract as one YAML file
- [RL] Pass return_indexer_topk through chat completions and sessions
- Docs
- Python not yet supported