# MATH-500 — AI model rankings

accuracy on the 500-problem MATH subset

11 tracked models have a published MATH-500 score. Higher is better. Scores come from published lab reports and benchmark sources; recorded sources appear beside the scores.

## Ranking

| Rank | Model | Developer | Score | Source | Released |
| --- | --- | --- | --- | --- | --- |
| 1 | o4-mini-high | OpenAI | 98.2% | Lab | Apr 16 2025 |
| 2 | o3-mini-high | OpenAI | 97.9% | Lab | Jan 31 2025 |
| 3 | o3 | OpenAI | 97.8% | Lab | Apr 16 2025 |
| 4 | o4-mini | OpenAI | 97.5% | Lab | Apr 16 2025 |
| 5 | o3-mini | OpenAI | 97.3% | Lab | Jan 31 2025 |
| 6 | o1-mini | OpenAI | 90% | Lab | Sep 12 2024 |
| 7 | GPT-4.5 | OpenAI | 87.1% | Lab | Feb 27 2025 |
| 8 | o1-preview | OpenAI | 85.5% | Lab | Sep 12 2024 |
| 9 | GPT-4.1 | OpenAI | 82.1% | Lab | Apr 14 2025 |
| 10 | GPT-4.1 mini | OpenAI | 81.4% | Lab | Apr 14 2025 |
| 11 | GPT-4.1 nano | OpenAI | 62.3% | Lab | Apr 14 2025 |


---

Canonical page: https://aireleasetracker.com/benchmark/math-500
Site index: https://aireleasetracker.com/llms.txt
Source: AI Release Tracker (https://aireleasetracker.com). Most benchmark scores come from lab launch material; gathered results identify the leaderboard that published them.