# MGSM — AI model rankings

solving grade-school word problems translated into multiple languages

30 tracked models have a published MGSM 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 | OpenAI | 93.7% | Lab | Apr 16 2025 |
| 2 | o4-mini-high | OpenAI | 93.5% | Lab | Apr 16 2025 |
| 3 | Claude 3.5 Sonnet (upgraded) | Anthropic | 92.5% | Lab | Oct 22 2024 |
| 4 | LLaMA 4 Maverick | Meta | 92.3% | Lab | Apr 5 2025 |
| 4 | o3 | OpenAI | 92.3% | Lab | Apr 16 2025 |
| 6 | o3-mini-high | OpenAI | 92% | Lab | Jan 31 2025 |
| 7 | Claude 3.5 Sonnet | Anthropic | 91.6% | Lab | Jun 20 2024 |
| 7 | LLaMA 3.1 | Meta | 91.6% | Lab | Jul 23 2024 |
| 9 | LLaMA 3.3 | Meta | 91.1% | Lab | Dec 6 2024 |
| 10 | o1-preview | OpenAI | 90.8% | Lab | Sep 12 2024 |
| 10 | o3-mini | OpenAI | 90.8% | Lab | Jan 31 2025 |
| 12 | Claude 3 Opus | Anthropic | 90.7% | Lab | Mar 4 2024 |
| 13 | LLaMA 4 Scout | Meta | 90.6% | Lab | Apr 5 2025 |
| 14 | GPT-4o | OpenAI | 89.9% | Lab | May 13 2024 |
| 14 | o1-mini | OpenAI | 89.9% | Lab | Sep 12 2024 |
| 16 | Gemini 1.5 Pro | Google | 88.73% | Lab | Feb 15 2024 |
| 17 | GPT-4.1 mini | OpenAI | 88.2% | Lab | Apr 14 2025 |
| 18 | GPT-4 Turbo | OpenAI | 87.1% | Lab | Nov 6 2023 |
| 19 | GPT-4o mini | OpenAI | 87% | Lab | Jul 18 2024 |
| 20 | LLaMA 3.2 | Meta | 86.9% | Lab | Sep 25 2024 |
| 20 | GPT-4.5 | OpenAI | 86.9% | Lab | Feb 27 2025 |
| 20 | GPT-4.1 | OpenAI | 86.9% | Lab | Apr 14 2025 |
| 23 | Claude 3.5 Haiku | Anthropic | 85.6% | Lab | Oct 22 2024 |
| 24 | Claude 3 Sonnet | Anthropic | 83.5% | Lab | Mar 4 2024 |
| 25 | Gemini 1.5 Flash | Google | 82.6% | Lab | May 14 2024 |
| 26 | Gemini 1.0 Ultra | Google | 79% | Lab | Feb 8 2024 |
| 27 | Claude 3 Haiku | Anthropic | 75.1% | Lab | Mar 4 2024 |
| 28 | GPT-4.1 nano | OpenAI | 73% | Lab | Apr 14 2025 |
| 29 | Gemini 1.5 Flash-8B | Google | 70.5% | Lab | Oct 3 2024 |
| 30 | Gemini 1.0 Pro | Google | 63.5% | Lab | Dec 13 2023 |


---

Canonical page: https://aireleasetracker.com/benchmark/mgsm
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.