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
- Bind streaming abort cleanup to request lifecycle
- [Test] Add unit test coverage for RadixCache: bigram mode, priority propagation, evict with protected nodes
- [Bug] Abort can commit a delayed final chunked-prefill token on latest main
- [Bug] Qwen3-VL + EAGLE3: draft-extend emits too few mRoPE positions, crashing on any batched traffic
- docs: fix model test suite link
- [MoE] Write CuTeDSL output directly into FlashInfer A2A workspace
- [Kimi-K3] Fuse pending AttnRes add into direct all-gather
- Add FastSafetensors sharded-state loader
- [Feature] Add AdaFlash adaptive verification for DFlash speculative decoding
- W4A4 MegaMoE FP4-acts on DeepSeek-V4-Flash-0731 (8x B200, SGLang v0.5.17): end-to-end TTFT unchanged on two serving shapes (scheduling-bound and GEMM-bound), plus bounded quality observations
- Docs
- Python not yet supported