MATH-500

accuracy on the 500-problem MATH subset

o4-mini-high by OpenAI holds the best MATH-500 result among tracked models, at 98.2% (released Apr 16 2025). Higher scores are better on this benchmark.

1. o4-mini-high (OpenAI) — 98.2%; 2. o3-mini-high (OpenAI) — 97.9%; 3. o3 (OpenAI) — 97.8%; 4. o4-mini (OpenAI) — 97.5%; 5. o3-mini (OpenAI) — 97.3%.

11 tracked models have a published MATH-500 score. Scores are collected from published lab reports and benchmark sources, with recorded sources credited alongside the results. We do not re-run the evaluations.