ARC-AGI-2

Reasoning

Abstraction and Reasoning Corpus for AGI evaluation.

Metrics
Accuracy (%)
Paper Dataset Eval Code

How to Run

pip install arckit && Download tasks from arcprize.org && python evaluate.py

Leaderboard

Rank Model Provider Parameters Score
1 GPT-5.2 OpenAI Unknown 52.9%
2 Gemini 3 Deep Think Google Unknown 45.1%
3 Claude Opus 4.5 Anthropic Unknown 37.6%
4 Gemini 3 Pro Google Unknown 31.1%