Skip to content

Quickstart

Currently the recommanded way to start vLLM-MetaX is via docker.

Belows is version mapping to released plugin and maca:

plugin version maca version docker distribution tag
v0.8.5 maca2.33.1.13 vllm:maca.ai2.33.1.13-torch2.6-py310-ubuntu22.04-amd64
v0.9.1 maca3.0.0.5 vllm:maca.ai3.0.0.5-torch2.6-py310-ubuntu22.04-amd64
v0.10.1.1 (dev) maca3.1.0.7 not released
v0.10.2 maca3.2.0.7 not released
v0.11.0 maca3.2.x.x not released
master maca3.2.x.x. not released

Note: All the vllm tests are based on the related maca version. Using incorresponding version of maca for vllm may cause unexpected bugs or errors. This is not garanteed.

vLLM-MetaX is out of box via these docker images.

Offline Batched Inference

OpenAI-Compatible Server

On Attention Backends