vllm on GitHub (vllm-project/vllm)

vllm README

Easy, fast, and cheap LLM serving for everyone

🔥 We have built a vLLM website to help you get started with vLLM. Please visit vllm.ai to learn more. For events, please visit vllm.ai/events to join us.

About

vLLM is a fast and easy-to-use library for LLM inference and serving.

Originally developed in the Sky Computing Lab at UC Berkeley, vLLM has grown into one of the most active open-source AI projects built and maintained by a diverse community of many dozens of academic institutions and companies from over 2000 contributors.

vLLM is fast with:

State-of-the-art serving throughput
Efficient management of attention key and value memory with PagedAttention
Continuous batching of incoming requests, chunked prefill, prefix caching
Fast and flexible model execution with piecewise and full CUDA/HIP graphs
Quantization: FP8, MXFP8/MXFP4, NVFP4, INT8, INT4, GPTQ/AWQ, GGUF, compressed-tensors, ModelOpt, TorchAO, and more
Optimized attention kernels including FlashAttention, FlashInfer, TRTLLM-GEN, FlashMLA, and Triton
Optimized GEMM/MoE kernels for various precisions using CUTLASS, TRTLLM-GEN, CuTeDSL
Speculative decoding including n-gram, suffix, EAGLE, DFlash
Automatic kernel generation and graph-level transformations using torch.compile
Disaggregated prefill, decode, and encode

vLLM is flexible and easy to use with:

Seamless integration with popular Hugging Face models
High-throughput serving with various decoding algorithms, including parallel sampling, beam search, and more
Tensor, pipeline, data, expert, and context parallelism for distributed inference
Streaming outputs
Generation of structured outputs using xgrammar or guidance
Tool calling and reasoning parsers
OpenAI-compatible API server, plus Anthropic Messages API and gRPC support
Efficient multi-LoRA support for dense and MoE layers
Support for NVIDIA GPUs, AMD GPUs, Intel GPUs, and x86/ARM/PowerPC CPUs. Additionally, diverse hardware plugins such as Google TPUs, Intel Gaudi, IBM Spyre, Huawei Ascend, Rebellions NPU, Apple Silicon, MetaX GPU, and more.

vLLM seamlessly supports 200+ model architectures on Hugging Face, including:

Decoder-only LLMs (e.g., Llama, Qwen, Gemma)
Mixture-of-Expert LLMs (e.g., Mixtral, DeepSeek-V3, Qwen-MoE, GPT-OSS)
Hybrid attention and state-space models (e.g., Mamba, Qwen3.5)
Multi-modal models (e.g., LLaVA, Qwen-VL, Pixtral)
Embedding and retrieval models (e.g., E5-Mistral, GTE, ColBERT)
Reward and classification models (e.g., Qwen-Math)

Find the full list of supported models here.

Getting Started

Install vLLM with uv (recommended) or pip:

uv pip install vllm

Or build from source for development.

Visit our documentation to learn more.

Contributing

We welcome and value any contributions and collaborations. Please check out Contributing to vLLM for how to get involved.

Citation

If you use vLLM for your research, please cite our paper:

@inproceedings{kwon2023efficient,
  title={Efficient Memory Management for Large Language Model Serving with PagedAttention},
  author={Woosuk Kwon and Zhuohan Li and Siyuan Zhuang and Ying Sheng and Lianmin Zheng and Cody Hao Yu and Joseph E. Gonzalez and Hao Zhang and Ion Stoica},
  booktitle={Proceedings of the ACM SIGOPS 29th Symposium on Operating Systems Principles},
  year={2023}
}

Contact Us

For technical questions and feature requests, please use GitHub Issues
For discussing with fellow users, please use the vLLM Forum
For coordinating contributions and development, please use Slack
For security disclosures, please use GitHub's Security Advisories feature
For collaborations and partnerships, please contact us at [email protected]

Media Kit

If you wish to use vLLM's logo, please refer to our media kit repo

94K ·GitHub

Vllm

What is this?

vLLM is a library for fast LLM inference and serving, built around PagedAttention for efficient memory management of attention keys and values.

Is this a library I import, or a standalone server?

Both: it's a Python library you can use directly for inference, and it also ships an OpenAI-compatible API server plus Anthropic Messages API and gRPC support for serving models.

How do I install and get it running?

Install vLLM with uv or pip, or build it from source for development; the docs have a full Quickstart guide.

What hardware do I need to run it?

It needs a supported accelerator: NVIDIA, AMD, or Intel GPUs, x86/ARM/PowerPC CPUs, or specialized hardware plugins like Google TPUs, Intel Gaudi, Huawei Ascend, or Apple Silicon.

Does it support any model I throw at it?

It supports 200+ model architectures from Hugging Face, but models outside that list (or very new architectures) may not work until support is added.

What kinds of models can vLLM serve besides plain text LLMs?

It supports Mixture-of-Expert models (DeepSeek-V3, Mixtral, GPT-OSS), hybrid attention/state-space models (Mamba, Qwen3.5), multi-modal models (LLaVA, Qwen-VL, Pixtral), and embedding/retrieval or reward/classification models.

What quantization formats does it support?

FP8, MXFP8/MXFP4, NVFP4, INT8, INT4, GPTQ/AWQ, GGUF, compressed-tensors, ModelOpt, and TorchAO, among others.

How does vLLM get such high throughput?

It combines PagedAttention for efficient KV-cache memory, continuous batching, chunked prefill, prefix caching, CUDA/HIP graphs, optimized attention and GEMM/MoE kernels, speculative decoding, and torch.compile-based kernel generation.

Can I run multiple LoRA adapters at once?

Yes, vLLM has efficient multi-LoRA support for both dense and MoE model layers.

What are similar projects?

sglang, text-generation-inference, Qwen3.8-Flash-Next, ktransformers, TensorRT-LLM, jeeves, project-maya, verl

Which topics is it in?

blackwell, tpu, gpt-oss, qwen3, llm-serving, model, rocm, kimi, speculative-decoding, mixture-of-experts, model-serving, amd

Creator: vllm-project
Size: 314.1 MB
Stars: 94K
Updated: CommitOct 17 Releasev0.31.0Oct 05
Age: 3y
Commits · 90 Days