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] Hunyuan3D: Skip redundant OBJ exports when generating GLB
- [diffusion] Add encode-only disagg serving with POST /encode
- [diffusion] Return FLUX.3 Image training targets in encode-only mode
- [Bug] bench_serving "Accept length" is the server's lifetime average, not this run's
- [Fix] bench_serving: report per-run accept length (bench-only, no server change)
- [Feat] rust-server: Report the scheduler's per-request stats in native /generate meta_info
- [Feat] diffusion: Support Hy-Nomas INT8 DiT and FP8 VAE checkpoints
- [Feat] [NVIDIA] flash-attn: FA4 SM90 FP8 e4m3 KV cache for head_dim 512 (Gemma-4)
- [Feat] rust-server: Serve /start_profile and /stop_profile
- [Fix] piecewise: make layer-boundary runtime paths dynamo-traceable
- Docs
- Python not yet supported