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] ngram: Wire SGLANG_NGRAM_FORCE_GREEDY_VERIFY back into NGRAM verify
- [Fix] multimodal: Cap local media reads at the size limit
- [Docs] Use rope_parameters for the YaRN overrides in the Qwen3-Next and Ling cookbooks
- [AMD] GLM-5.2: use custom AR + add-RMSNorm (non-fused) in decode/verify
- [Feat] [AMD] dsa: Allow breakable prefill CUDA graph on HIP
- [Feature] Accept per-item precomputed media with explicit offsets and hashes
- [Spec] Abort only the request when grammar traversal rejects a draft token
- [Perf] [NVIDIA] sampling: Enable Cake softmax on SM100
- [Fix] memory: Reserve the symmetric-memory prealloc out of the KV budget
- [RFC] Preserve multimodal identity end to end: precomputed inputs, collision-safe KV reuse, and KV-event parity
- Docs
- Python not yet supported