reasoning benchmark
ARC-AGI-3
ARC-AGI-3 is the third-generation Abstraction and Reasoning Corpus benchmark, an interactive-reasoning evaluation designed to measure fluid, novel problem-solving ability that remains far from saturated for frontier models.
Updated Aug 7, 2026
Published models4
Registry coverage4
MetricScore
EvidenceB
ARC-AGI-3 leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | AN | 30.2% | 100.0% | 4 | C | |
| 2 | OP | 7.8% | 66.7% | 4 | C | |
| 3 | OP | 0.8% | 33.3% | 4 | C | |
| 4 | OP | 0.2% | 0.0% | 4 | C |
Score distribution
Top published rows on the benchmark's original scale.
ARC-AGI-3 highlights
The top published results on this benchmark's own scale.
What is ARC-AGI-3?
Definition and scoring fields from the benchmark registry.
ARC-AGI-3 is the third-generation Abstraction and Reasoning Corpus benchmark, an interactive-reasoning evaluation designed to measure fluid, novel problem-solving ability that remains far from saturated for frontier models.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- ARC-AGI-3
- Modality
- multimodal
- Primary category
- reasoning
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- arc-agi-3|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about ARC-AGI-3.
Which model scores highest on ARC-AGI-3?
Claude Opus 5 is currently ranked first with 30.2%.
What does ARC-AGI-3 measure?
ARC-AGI-3 is the third-generation Abstraction and Reasoning Corpus benchmark, an interactive-reasoning evaluation designed to measure fluid, novel problem-solving ability that remains far from saturated for frontier models.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
4 unique published model results are currently shown.
Does this benchmark affect the overall score?
This benchmark is preserved as source-native evidence but is not eligible for the current overall score.