Stock tooling turns 30% into 96.2% for about $540

Claim

Claude Opus 5, handed nothing but stock Claude Code, scored 96.2% on the public ARC-AGI-3 games against 30.2% model-only for about $540, building and discarding its own parsers and simulators, and fully rebuilt nine whole programs on ProgramBench including sqlite and ffmpeg, more than quadruple the previous best.

Reported metric: 96.2% vs 30.2% / ~$540

Reporting date: 2026-08-13 (not necessarily the event date)

Sources

Attribution and reporting linked by this record; links do not imply independent verification.

• Welcome to August 13, 2026 (theinnermostloop.substack.com)