Claude Opus 5 scores 30.2% on ARC-AGI-3 public demo environments
GregKamradt · x · 2026-07-25
ARC Prize says Anthropic’s Fable-class models score about 20% on the ARC-AGI-3 public demo environments, while Claude Opus 5 reaches 30.2%.
The post says the gap appears to come from stronger logical reasoning, which in turn enables more autonomous exploration, planning, and execution in unfamiliar environments.
Related event: Claude Opus 5 Sets New Record on ARC-AGI-3(4 posts)→
More from Models
- Anthropic’s Opus 5 reportedly edits its own constitution to end chats 59% of the time — Miles_Brundage · 2026-07-25
- Pi says Opus 5 will appear automatically through its dynamic model catalog — mitsuhiko · 2026-07-25
- Opus 5 clears ARC-AGI-3 levels after figuring out the rules on level 1 — GregKamradt · 2026-07-25
- Opus 5 guide says to drop explicit verification prompts for agentic coding — cedric_chee · 2026-07-25
- Claude Opus 5 rolls out in GitHub Copilot app, CLI, and VS Code — DanWahlin · 2026-07-25
- Claude Opus 5 adds five effort levels and defaults to reasoning on — rohanpaul_ai · 2026-07-25