Small humanizer model beats v4 at long-form prose, exceeding expectations

ctjlewis · x · 2026-09-04

Follow-up from ctjlewis: he was bearish, doubting anyone could train a small humanizer that is also good and beats v4 at prose of that length — but the result exceeded his expectations. Combined with his earlier note, the model is cheap, writes well, and carries near-zero "jailbreak tax."

Related event: Cheap model beats v4 in long-form writing, jailbreak cost nears zero(2 posts)→

Original post →

More from Models

Models channel →