sgl-project
2 Intel
sgl-project is the GitHub organization behind SGLang, an open-source serving framework for large language models and multimodal models. SGLang's documentation says the project is hosted under the non-profit open-source organization LMSYS. SGLang was introduced in a December 2023 arXiv paper by Lianmin Zheng and 11 co-authors, which describes a frontend language and a runtime that uses RadixAttention for KV cache reuse. The organization also maintains related repositories such as SpecForge, SGLang-Omni and Mini-SGLang.
Intel
SGLang v0.5.22 ships 950 PRs adding new LLM, VLM and diffusion model support. It reports up to 2.3x more per-GPU throughput for DeepSeek-V4.1 on H200, AMD MI350X and MI355X support, and faster decoding for Qwen and GLM variants on Blackwell GPUs.
SGLang v0.5.23 ships 1007 PRs adding cookbook support for new LLMs, VLMs and diffusion models, plus DeepSeek-V4.1 prefill and first-token latency gains on GB300 hardware.