Claude reportedly scores 30.2% on ARC-AGI-3 and gets a novel-problem-solving prompt pattern

socialwithaayan · x · 2026-07-27

The thread says Claude scores 30.2% on ARC-AGI-3, roughly triple the next best model, and interprets that as evidence of stronger novel problem solving.

It then suggests a prompt pattern for open-ended research strategy: restate the problem in three ways, import mechanisms from unrelated fields, translate them into the target problem, and stress-test the proposals.

Related event: Claude Opus 5 Tests: Unlocking Monster Potential via Multi-Agent Workflows(10 posts)→

Original post →

More from coding & agent

coding & agent channel →