Claude Opus 5 Scores Three Times Higher Than Next Best Model on ARC-AGI-3

claudeai · x · 2026-07-25

Anthropic announced Claude Opus 5's performance on the ARC-AGI-3 evaluation, which tests AI models on solving novel problems. Opus 5 scored three times as high as the next best model, demonstrating exceptional reasoning and generalization capabilities.

Related event: Anthropic Unveils Claude Opus 5 with Top Benchmark Scores(28 posts)→

Original post →

More from Models

Models channel →