MiniMax M3 Runs on NVIDIA Vera Rubin NVL72 via vLLM, Claiming 7.8x Throughput Over GB200

MiniMax (official) 1 hour ago

Key Info

MiniMax M3 is now supported on NVIDIA Vera Rubin NVL72 through vLLM, with collaboration from Inferact, NVIDIA AI, and Red Hat AI.

Early vLLM results show more than 7.8x the throughput of GB200 on MiniMax M3 for AgentX workloads
vLLM has been updated to support NVIDIA Vera Rubin since its announcement
Multiple partners contributed to bringing the model up on the new hardware

Compiled from 1 post

#MiniMax M3#vLLM#NVIDIA Vera Rubin