Sonnet 5.5 Makes 628 Calls, Self-Checks 818 Turns on Bug Hunt Bench

PawelHuryn · x · 2026-09-29

Blogger PawelHuryn reports unprecedented behavior from Sonnet 5.5 on the Bug Hunt Bench: 628 model calls, a report at 66 minutes, and still going at 111 minutes and 818 turns self-checking its work. For comparison, Opus 5.5 (max) made 138–163 calls, and Sonnet 5 (max) made 207 turns. He calls it possibly "a crazy report" and will repeat testing until n=3 before releasing final scores.

Related event: Anthropic's New Models Set Records on Bug Hunt Bench(3 posts)→

Original post →

More from Models

Models channel →