Vibeleaderboard
← Back to Vibers
vLLM
Company

vLLM

github.com/vllm-project

1 Intel

vLLM is an open-source library for fast LLM inference and serving, built around PagedAttention memory management and continuous batching. It was originally developed in the Sky Computing Lab at UC Berkeley and is now maintained by a community of academic institutions and companies with more than 2,000 contributors. The PyTorch Foundation lists vLLM among its projects, and it runs on NVIDIA and AMD GPUs, Google TPUs, AWS Trainium and Intel CPUs.

Intel

vllm v0.33.0rc3 — v0.33.0(github.com)

vLLM 0.33.0 adds support for many new models, upgrades to PyTorch 2.14, and drops CUDA 12.9 wheels and images. It also ships an AWS EFA image for KV transfer and multi-node serving, plus speculative decoding changes.

Developer Toolsbuilt by vllm-project