sglang

SGLang is a fast serving framework for large language models and vision language models.

sglang Permissive license Apache License Version 2.0, January 2004 http://www.apache.org/licenses/ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION 1. Definitions. "License" shall mean the terms and conditions for… (full text in the JSON record) Active ★ 31,752 v0.5.17 released 2026-08-08

Install

sglang on PyPI

pip

pip install sglang

uv

uv add sglang

poetry

poetry add sglang

Package facts

License	Apache License Version 2.0, January 2004 http://www.apache.org/licenses/ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION 1. Definitions. "License" shall mean the terms and conditions for… (full text in the JSON record) (permissive)
Python support	supports the current Python release (>=3.10)
Install friction	medium — platform-specific wheel
Runtime dependencies	73 — aiohttp, anthropic, apache-tvm-ffi, av, blobfile, build, compressed-tensors, cuda-python, datasets, decord2, distro, easydict, einops, fastapi, flash-attn-4, flashinfer_python, gguf, helion, humming-kernels, interegular, IPython, kernels, llguidance, mistral_common, modelscope, msgspec, ninja, numba, numpy, nvidia-cutlass-dsl
Maintenance	actively maintained — 5 days since the last release
Last repo commit	2026-08-13
First released	2024-01-16
Popularity	one of the top 1,000 most-downloaded packages on PyPI (30-day window, as of 2026-08-13)
Known vulnerabilities	none known (OSV.dev, checked 2026-08-13)

Evidence: sglang-0.5.17-cp310-cp310-manylinux_2_34_aarch64.whl; sglang-0.5.17-cp310-cp310-manylinux_2_34_x86_64.whl; sglang-0.5.17-cp311-cp311-manylinux_2_34_aarch64.whl; sglang-0.5.17-cp311-cp311-manylinux_2_34_x86_64.whl; sglang-0.5.17-cp312-cp312-manylinux_2_34_aarch64.whl; sglang-0.5.17-cp312-cp312-manylinux_2_34_x86_64.whl; sglang-0.5.17-cp313-cp313-manylinux_2_34_aarch64.whl; sglang-0.5.17-cp313-cp313-manylinux_2_34_x86_64.whl

License :: OSI Approved :: Apache Software LicenseProgramming Language :: Python :: 3

About sglang

from the package's own PyPI description — quoted content, verbatim

PyPI (image) PyPI - Downloads (image) license (image) issue resolution (image) open issues (image) Ask DeepWiki (image)

</div>

<a href="https://lmsys.org/blog/">Blog</a> | <a href="https://docs.sglang.io/">Documentation</a> | <a href="https://roadmap.sglang.io/">Roadmap</a> | <a href="https://slack.sglang.io/">Join Slack</a> | <a href="https://meet.sglang.io/">Weekly Dev...

Read as markdown · JSON record · Source repository · Homepage

AI interpretation — verify before relying

AI-generated interpretation of the package facts above; every digit, version, license, or vulnerability id it cites is grounded in the facts already shown on this page

SGLang is a high-performance serving framework for large language models and multimodal models, delivering low-latency and high-throughput inference with features like RadixAttention prefix caching, speculative decoding, and structured outputs.

Medium install friction: 73 runtime dependencies including compiled components (cuda-python, apache-tvm-ffi, flash-attn-4, flashinfer_python) and specialized kernels. Prebuilt wheels available for Python 3.10–3.13 on x86_64 and aarch64. Active maintenance with latest release 5 days old.

Apache License 2.0 (permissive): you may use, modify, and distribute sglang freely in commercial and private projects, provided you include the license and retain attribution notices in derivative works.

Usage

pip install sglang
import sglang as sgl
# Requires GPU and model weights; see documentation for serving setup

Requires NVIDIA CUDA-capable GPU, cuda-python, and compiled dependencies (flash-attn-4, flashinfer_python); Python ≥3.10

Verdict: SGLang is an actively maintained framework for production LLM serving with strong community engagement. Medium install friction due to 73 dependencies and compiled components is typical for high-performance inference. Apache 2.0 license is permissive. No known vulnerabilities. Best suited for teams deploying large models at scale on GPU infrastructure.

Needs verification

Whether all 73 runtime dependencies are required for basic usage or if many are optional for specific features
Whether the framework supports AMD or other non-NVIDIA GPUs despite cuda-python being listed
Performance benchmarks against other serving frameworks to validate claimed speedups
Actual community adoption metrics beyond maintenance status

llm serving frameworkhigh-throughput inference enginelanguage model deploymentgpu-accelerated model servingdistributed llm inferencefast batch schedulingmultimodal model serving