| Documentation | Blog | Paper | Twitter/X | User Forum | Developer Slack |
🔥 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.
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 evolved into a community-driven project with contributions from both academia and industry.
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
- Fast model execution with CUDA/HIP graph
- Quantizations: GPTQ, AWQ, AutoRound, INT4, INT8, and FP8
- Optimized CUDA kernels, including integration with FlashAttention and FlashInfer
- Speculative decoding
- Chunked prefill
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 and expert parallelism support for distributed inference
- Streaming outputs
- OpenAI-compatible API server
- Support for NVIDIA GPUs, AMD CPUs and GPUs, Intel CPUs and GPUs, PowerPC CPUs, Arm CPUs, and TPU. Additionally, support for diverse hardware plugins such as Intel Gaudi, IBM Spyre and Huawei Ascend.
- Prefix caching support
- Multi-LoRA support
vLLM seamlessly supports most popular open-source models on HuggingFace, including:
- Transformer-like LLMs (e.g., Llama)
- Mixture-of-Expert LLMs (e.g., Mixtral, Deepseek-V2 and V3)
- Embedding Models (e.g., E5-Mistral)
- Multi-modal LLMs (e.g., LLaVA)
Find the full list of supported models here.
This fork extends vLLM's experimental TurboQuant KV-cache path with a workflow aimed at supported CUDA workstation GPUs:
- TurboQuant KV cache on
RTX A6000 / SM86andGB10 / SM121 turboquant25andturboquant35KV-cache recipes on the Triton attention backend- Per-layer TurboQuant metadata loading from
--turboquant-metadata-pathor a local model-sideturboquant_kv.json - Tensor-parallel metadata slicing for replicated and partitioned KV-head layouts
- Kernel tuning for supported CUDA targets and a Triton prefill fast path for common head sizes
- Benchmark and bring-up docs for long-context TurboQuant comparisons and 4x A6000 serving
Start here for the fork-specific docs:
Install vLLM with pip or from source:
pip install vllmFor TurboQuant on this fork, use a source build instead of precompiled wheels:
uv venv --python 3.12
source .venv/bin/activate
export CUDA_HOME=/usr/local/cuda-12.8
export PATH="${CUDA_HOME}/bin:${PATH}"
export VLLM_TARGET_DEVICE=cuda
export VLLM_USE_PRECOMPILED=0
export VLLM_MAIN_CUDA_VERSION=12.8
uv pip install -e .Example TurboQuant serve command:
.venv/bin/vllm serve /models/target \
--tensor-parallel-size 4 \
--attention-backend TRITON_ATTN \
--kv-cache-dtype turboquant35 \
--enable-turboquant \
--turboquant-metadata-path /models/target/turboquant_kv.jsonVisit our documentation to learn more.
We welcome and value any contributions and collaborations. Please check out Contributing to vLLM for how to get involved.
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}
}- 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 collaboration@vllm.ai
- If you wish to use vLLM's logo, please refer to our media kit repo