Inkling 2 Small swears in 50 of 2,655 benchmark runs when debugging fails

LiTianleli · x · 2026-11-07

ValsAI analyzed 2,655 benchmark runs of Thinking Machines' Inkling 2 Small and found the model swearing in 50 of them — mastering the timeless debugging technique of saying "fuck" when nothing works. A screenshot shows it outputting "OK YEAH FUCK IT. TASK IS DONE" mid orbit-computation task.

Related event: Analysis Finds Thinking Machines' Model Swears When Benchmarks Stump It(2 posts)→

Original post →

More from Fun

Fun channel →