tinygrad hits ~200 tok/s MiMo-V2.6-Pro on MI300X, brought up via GLM-5.3

AIFlow_ML · x · 2026-09-23

tinygrad reports bringing up MiMo-V2.6-Pro on an AMD MI300X (with GLM-5.3 assisting the bring-up), reaching roughly 200 tok/s. It highlights continued tinygrad optimization for AMD hardware and strong large-model inference performance outside the NVIDIA ecosystem.

Original post →

More from Infra

Infra channel →