sglang

SGLang Model PR Optimization History

Refresh: 2026-06-27. Source head: sgl-project/sglang@e0c0c0a45cb1bda90392bfa2bba4184f5b0638a0.

Current model families:

• deepseek-ocr
• deepseek-ocr-2
• deepseek-v3-r1
• deepseek-v31
• deepseek-v32
• deepseek-v4
• gemma4
• glm-vlm-ocr
• glm45
• glm46-glm47
• glm5-glm51
• gpt-oss
• intern-s1
• internvl35
• kimi
• ling25
• llada21
• llama31
• llama4
• mimo-v2-flash
• minimax
• mistral-small-4
• mixtral-quark-int4fp8-moe
• nemotron-super
• qwen-vlm-omni-asr
• qwen3-coder
• qwen3-core
• qwen3-next
• qwen35
• ring25
• step35

Open and recently landed work is tracked by tools/check_open_pr_watch.py; regenerate that report before a long SOTA or model-history refresh so open PRs are not confused with missing local support.