# MATH — AI model rankings

accuracy on the original competition mathematics test set

21 tracked models have a published MATH 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 | Gemini 2.0 Pro | Google | 91.8% | Lab | Feb 5 2025 |
| 2 | Gemma 3 | Google | 89% | Lab | Mar 12 2025 |
| 3 | Gemini 1.5 Pro-002 | Google | 86.5% | Lab | Sep 24 2024 |
| 4 | Gemini 1.5 Flash-002 | Google | 77.9% | Lab | Sep 24 2024 |
| 5 | LLaMA 3.3 | Meta | 77% | Lab | Dec 6 2024 |
| 6 | GPT-4o | OpenAI | 76.6% | Lab | May 13 2024 |
| 7 | LLaMA 3.1 | Meta | 73.8% | Lab | Jul 23 2024 |
| 8 | GPT-4o mini | OpenAI | 70.2% | Lab | Jul 18 2024 |
| 9 | LLaMA 3.2 | Meta | 68% | Lab | Sep 25 2024 |
| 10 | GPT-4 Turbo | OpenAI | 64.3% | Lab | Nov 6 2023 |
| 11 | Gemini 1.5 Pro | Google | 58.5% | Lab | Feb 15 2024 |
| 12 | Gemini 1.5 Flash | Google | 54.9% | Lab | May 14 2024 |
| 13 | Gemini 1.0 Ultra | Google | 53.2% | Lab | Feb 8 2024 |
| 14 | LLaMA 3 (8B/70B) | Meta | 50.4% | Lab | Apr 18 2024 |
| 15 | Gemma 2 | Google | 42.3% | Lab | Jun 27 2024 |
| 16 | Gemini 1.5 Flash-8B | Google | 35.9% | Lab | Oct 3 2024 |
| 17 | Gemini 1.0 Pro | Google | 32.6% | Lab | Dec 13 2023 |
| 18 | Gemma 1 | Google | 24.3% | Lab | Feb 21 2024 |
| 19 | Gemini 1.0 Nano | Google | 22.8% | Lab | Dec 6 2023 |
| 20 | LLaMA 2 | Meta | 13.5% | Lab | Jul 18 2023 |
| 21 | LLaMA 1 | Meta | 10.6% | Lab | Feb 24 2023 |


---

Canonical page: https://aireleasetracker.com/benchmark/math-full
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.