Who's who · Open-source projects

vLLM

Open-source engine for serving large language models quickly and cheaply at scale

At a glance

Founded
2023
Based in
Berkeley, United States
People
Woosuk Kwon, Core maintainer; co-founder, Inferact; Simon Mo, Core maintainer; co-founder, Inferact; Kaichao You, Core maintainer; co-founder, Inferact; Roger Wang, Core maintainer; co-founder, Inferact
Known for
vLLM, PagedAttention
Licence
Apache 2.0
Funding
Open-source project of the PyTorch Foundation
GitHub stars
93,000+ (Oct 2026)
Governance
PyTorch Foundation hosted project (May 2025)
Maintainers' company
Inferact, $150M raised at $800M (Jan 2026)
Website
https://github.com/vllm-project/vllm

vLLM is an open-source inference and serving engine for large language models, used to run models efficiently on servers. Inference is the work of running a trained model to produce answers. vLLM began in the Sky Computing Lab at the University of California, Berkeley in 2023 around PagedAttention, a method for managing model memory more efficiently. It has grown into a full serving stack with features such as continuous batching, speculative decoding, prefix caching, quantisation and distributed serving.

It runs on NVIDIA, AMD and Intel GPUs and on ordinary CPUs, with plug-ins for Google TPUs, Huawei Ascend and other accelerators, and supports more than 200 model architectures. Production users named by its backers include Meta, Google and Character.ai, and a16z said in January 2026 that vLLM was running on more than 400,000 GPUs at once. The repository has more than 90,000 GitHub stars and over 2,000 contributors, and is licensed under Apache 2.0.

Since May 2025 vLLM has been a hosted project of the PyTorch Foundation. In January 2026 core maintainers Simon Mo, Woosuk Kwon, Kaichao You and Roger Wang launched Inferact, a company building on vLLM, which raised $150 million at an $800 million valuation and says vLLM stays open source.

Sources

  1. PyTorch Foundation, PyTorch Foundation welcomes vLLM as a hosted project, 2025-05-07
  2. Inferact, Inferact, 2026-10-02
  3. Andreessen Horowitz, Investing in Inferact, 2026-01-22
  4. GitHub, vllm-project/vllm repository, 2026-10-02

Facts last checked 2026-10-02