MathNet - Algebra: leaderboard
Metric: Problem-solving accuracy (%) on the algebra problems on MathNet-Solve-Test, Olympiad problems from official national competition booklets of 47 countries; a GPT-5 grader scores each solution 0-7 against the official solution and 6 or more counts as correct; image-capable models get the figures, text-only models a text description of them; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 14 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 83.7 |
| 2 | GPT-5 | 80.3 |
| 3 | Gemini 3 Flash (Preview) | 77.7 |
| 4 | GPT-5 Mini | 67.6 |
| 5 | GPT-5 Nano | 53.9 |
| 6 | Claude Opus 4.6 (Thinking) | 53.2 |
| 7 | DeepSeek V3.2 (Non-reasoning) | 51.6 |
| 8 | Gemini 2.5 Flash | 50.5 |
| 9 | Grok 3 | 37.7 |
| 10 | GPT-4.1 | 29.4 |
| 11 | Llama 4 Maverick Instruct FP8 | 22.5 |
| 12 | GPT-4o | 10.9 |
| 13 | Ministral 3B | 6.4 |
Interactive version: theaggregate.ai/benchmark?slug=mathnet-algebra · How It Works · Data refreshed daily, snapshot 2026-10-07.