Updated daily at 05:00 UTC
Benchmark
ARC-AGI-3
Measures: Hundreds of hand-made, turn-based game environments with no instructions or stated goals; an agent must explore, infer the rules and win, and is scored on action efficiency relative to human baselines (100% = beats every game as efficiently as humans).
- Maintainer
- ARC Prize Foundation
- Introduced
- 2026
- Status
- Active
- Saturation
- Near-ceiling
- Typical provenance
- Independent
- In scoring set
- Yes
Leaderboard
Every result we hold, whoever ran it. Only independent and third-party rows count towards a score.
No results to show.