Uncensored 26.4B Inkling 2 Mini runs locally on a 12GB laptop, keeps 98% MMLU

udmrzn · x · 2026-11-22

AtomicChat released uncensored GGUF quantizations of Inkling 2 Mini (26.4B), in three sizes from 8.67 to 16.53GB, runnable locally on a 12GB laptop.

In their tests refusals dropped from 100% to 0% while retaining about 98% of the original MMLU accuracy. Quants are available on Hugging Face.

Related event: Uncensored 26.4B Inkling 2 Mini runs locally on 12GB laptops(2 posts)→

Original post →

More from Infra

Infra channel →