WHAT NEW IN X10.28 WED

AI collections from X · 22 collections · 58,925 posts

minimax-m3.1IMAGE

Unsloth releases UD3 quants for MiniMax M3.1 Flash

Unsloth shipped UD3 quantizations of MiniMax-M3.1-Flash, with NVFP4 MoE layers by default and UD-IQ1_M retaining about 89.9% top-1; runs in Unsloth Desktop.

Daniel Han
@danielhanchenArchived 1 day agoOriginally posted 2026.10.26
Unsloth releases UD3 quants for MiniMax M3.1 Flash
Daniel Han @danielhanchenOriginally posted 2026.10.26
Open original on X

We made Unsloth UD3 MiniMax-M3.1-Flash quants! We tested across multiple domains to ensure our quants work great for them as well! By default, MiniMax-M3.1-Flash uses NVFP4 for MoE layers and BF16 / MXFP8 for the rest. UD-IQ1_M retains ~89.9% top-1%. Works in Unsloth Desktop!

This post quotes another post: “Unsloth releases 2-bit GGUF for MiniMax M3.1 Flash: runs on 150GB RAM”

x.com/danielhanchen/status/2114733721396080945Archived 2026.10.27 12:04