SGLang: Fast inference for LLMs and multimodal models

Fields

id

sources/src5-sglang-github

URL

https://github.com/sgl-project/sglang

Publisher

sgl-project (LMSYS)

Kind

repo

Published

2024-01-01

Accessed

2026-10-04

notes

SGLang is an open-source inference framework for LLMs, vision-language, and diffusion models. Optimized for agentic workloads, RL rollouts, large-scale serving. Supports NVIDIA, AMD, Google TPU, Intel, Apple Silicon. 36.8k stars. Includes SGLang Diffusion for image/video generation. Learned from vLLM, FlashInfer, Guidance, Outlines, LMQL.

Notes

SGLang is an open-source inference framework for LLMs, vision-language, and diffusion models. Optimized for agentic workloads, RL rollouts, large-scale serving. Supports NVIDIA, AMD, Google TPU, Intel, Apple Silicon. 36.8k stars. Includes SGLang Diffusion for image/video generation. Learned from vLLM, FlashInfer, Guidance, Outlines, LMQL.

Cross-references

This record references no other records.