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] Inkling multimodal returns HTTP 500 instead of 400 for invalid image/audio input
- feat(bench): add --gsp-input-ids to skip server tokenization
- [Bug]: abort chunk re-sends the full accumulated text under --incremental-streaming-output
- [Bug] DeepSeek chunked-prefix prefill merges base-2 LSE into a natural-log merge_state_v2 (silent accuracy loss on every prefix-cache hit)
- Declare hybrid Mamba indexer host pool
- feat(function_call): add kimi_k2_raw_id parser keeping model-emitted tool call ids
- fix: preserve CUDA error enums during HiCache cleanup
- [Bug] Anthropic `/v1/messages`: prior-turn thinking is spliced into content, which breaks the Qwen3.8 chat template (empty `<think>` + inline `<think>…</think>`); model starts emitting `</think>` as visible text
- [DSV4] Use the shared memory budget for dense prefill
- [Bugfix] Flush pending tool calls at end of stream in the Llama 3.2 detector
- Docs
- Python not yet supported