AIME 2025

Math

American Invitational Mathematics Examination 2025.

Metrics
Accuracy (%)
Dataset Eval Code

How to Run

Use OpenAI simple-evals framework or manually evaluate against AIME 2025 problems

Leaderboard

Rank Model Provider Parameters Score
1 GPT-5.2 Thinking OpenAI Unknown 100.0%
2 Gemini 3 Pro Google Unknown 95.0%
3 Claude Opus 4.5 Anthropic Unknown 93.0%
4 DeepSeek-R1 DeepSeek 671B MoE 79.2%