WHAT NEW IN X10.28 WED

AI collections from X · 22 collections · 58,924 posts

New todayminimax-m3.1IMAGE

MiniMax M3.1 Flash 428B quantized to 143.9GB for single-machine local runs

Unsloth's 2-bit dynamic GGUF shrinks the 428B M3.1 Flash to 143.9GB, runnable on a 150GB RAM machine with 91.5% top-1 agreement versus the original, plus official DeepSWE/OSWorld/CyberGym scores.

Landscape Designer
@czw88888Archived 6 hours agoOriginally posted 2026.10.28
MiniMax M3.1 Flash 428B quantized to 143.9GB for single-machine local runs (image 1 of 3)
MiniMax M3.1 Flash 428B quantized to 143.9GB for single-machine local runs (image 2 of 3)
MiniMax M3.1 Flash 428B quantized to 143.9GB for single-machine local runs (image 3 of 3)
Landscape Designer @czw88888Originally posted 2026.10.28
Open original on X

428B参数的模型,现在一台150GB内存的机器就能跑。 MiniMax M3.1 Flash总参数428B,每个token只激活23B。Unsloth的2-bit动态GGUF压到143.9GB,和原版的top-1输出一致率91.5%。 能力没怎么缩水。官方数据:DeepSWE v1.1 73.7%,OSWorld V2 61.7,CyberGym 82.2。 MoE加激进量化,让前沿级agent模型第一次进了高配工作站的范围。按token收费的闭源API,护城河又薄了一层。 #MiniMax #开源模型 #Unsloth #本地部署 #MoE https://x.com/MiniMax_AI/status/2114815625310933277

This post quotes another post: “MiniMax confirms Unsloth 2-bit GGUF: 143.9GB footprint, 91.5% top-1 agreement”

x.com/czw88888/status/2115218246290886697Archived 2026.10.28 12:02