# vLLM MetaX Plugin

A community-maintained hardware plugin that lets the vLLM inference engine run on MetaX GPUs, using MetaX's CUDA-like MACA backend for near-native CUDA performance.

vLLM MetaX Plugin is a hardware plugin, implemented following vLLM's hardware-pluggable RFCs, that enables vLLM to run on MetaX GPUs by targeting MetaX's MACA backend, which provides a CUDA-like execution environment on that hardware. It is presented as the recommended way to support MetaX hardware within the vLLM community.

The project tracks vLLM's own release cadence closely, publishing an aligned release for most vLLM versions across roughly a year of releases, adding support for more models and improving performance with each release, and the maintainers have hosted vLLM community meetups focused on distributed inference and diverse hardware accelerator support.

Website: https://vllm-metax.readthedocs.io/en/latest/