Wafer says AMD's MI355X beats Nvidia B300 on Kimi K3 cost efficiency
Wafer reports 952 tokens per second from one eight-GPU MI355X node, arguing AMD's 288 GB-per-GPU memory can lower open-model inference costs.
By Ryan Merket · Published
Primary source: Wafer
Why it matters
Kimi K3's size turns GPU memory into a deployment constraint. Wafer's self-reported benchmark shows how software tuning and larger HBM capacity could make AMD economical for frontier-scale open models, even when Nvidia retains the raw-speed lead.
Wafer, founded by Emilio Andere (@gpuemi) and Steven Arellano (@gpusteve), says it ran Kimi K3 on eight AMD MI355X GPUs at better performance per dollar than an Nvidia B300 system, using memory capacity and a pair of software fixes to narrow AMD's inference gap.…
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.