Claude Opus 5 from @AnthropicAI is the new SOTA on ARC-AGI-3: 30.2%...
ARC Prize@arcprize
36 views
Jul 24, 2026
~2 min read
Advertisement
1
Claude Opus 5 from @AnthropicAI is the new SOTA on ARC-AGI-3: 30.2%
The previous high score (7.8%) was set by GPT-5.6 Sol (Max)
Throughout our analysis, we observed novel behavior that allows Opus 5 to solve previously unbeaten environments, outperforming Fable
The previous high score (7.8%) was set by GPT-5.6 Sol (Max)
Throughout our analysis, we observed novel behavior that allows Opus 5 to solve previously unbeaten environments, outperforming Fable
2
In our testing to date, Anthropic’s Fable-class models score approximately 20% on the ARC-AGI-3 Public Demo environments
Claude Opus 5 reaches 30.2%, materially outperforming Fable
Our analysis suggests the gain comes from stronger logical reasoning, which enables more autonomous exploration, planning, and execution across unfamiliar environments
Claude Opus 5 reaches 30.2%, materially outperforming Fable
Our analysis suggests the gain comes from stronger logical reasoning, which enables more autonomous exploration, planning, and execution across unfamiliar environments
3
Claude Opus 5 was able to score 100% on 5 previously unbeaten environments
Of these, it was able to beat 4 of them matching or surpassing human level efficiency
Newly beaten environments: ar25, ft09, lp85, r11l, s5i5
6 of the 25 public demo environments have now been solved
See Opus 5 play s5i5: arcprize.org/replay/cb8fbc9…
Of these, it was able to beat 4 of them matching or surpassing human level efficiency
Newly beaten environments: ar25, ft09, lp85, r11l, s5i5
6 of the 25 public demo environments have now been solved
See Opus 5 play s5i5: arcprize.org/replay/cb8fbc9…
4
During our analysis of Opus 5, we observed a new capability previously unseen from frontier models
Opus 5 used advanced logical reasoning to turn ARC-AGI-3 layouts into algebraic notation. On action 23 it described the scene as "4_center = 2×axis − 5_center"
This is the first explicit reflection equation by a model we've analyzed
By action 248 the model went further and generalizes this to two dimensions: “ghost = reflection about axis strips: br' = 2·br_axis − br, bc' = 2·bc_axis − bc”
ar25 was previously unbeaten. View replay: arcprize.org/replay/b71b579…
Opus 5 used advanced logical reasoning to turn ARC-AGI-3 layouts into algebraic notation. On action 23 it described the scene as "4_center = 2×axis − 5_center"
This is the first explicit reflection equation by a model we've analyzed
By action 248 the model went further and generalizes this to two dimensions: “ghost = reflection about axis strips: br' = 2·br_axis − br, bc' = 2·bc_axis − bc”
ar25 was previously unbeaten. View replay: arcprize.org/replay/b71b579…
5
ARC-AGI-2
Claude Opus 5 scores 90.4% for $2.06/task
This is competitive with previous SOTA performance for slightly higher cost
Claude Opus 5 scores 90.4% for $2.06/task
This is competitive with previous SOTA performance for slightly higher cost
6
ARC-AGI-1
Claude Opus 5 scores 97.5% for $0.70/task
This is competitive with previous SOTA performance for slightly higher cost
Claude Opus 5 scores 97.5% for $0.70/task
This is competitive with previous SOTA performance for slightly higher cost
7
- Leaderboard: arcprize.org/leaderboard
- Reproduce the results: github.com/arcprize/arc-a…, github.com/arcprize/arc-a…
- Testing policy: arcprize.org/policy
- Full Claude Opus 5 Results: arcprize.org/results/anthro…
- Reproduce the results: github.com/arcprize/arc-a…, github.com/arcprize/arc-a…
- Testing policy: arcprize.org/policy
- Full Claude Opus 5 Results: arcprize.org/results/anthro…
8
If you'd like to download the ARC-AGI-3 replay logs (which include the model's reasoning & thought process) for the public demo games, view the Opus 5 Scorecard
arcprize.org/scorecards/mod…
arcprize.org/scorecards/mod…