AI for Education Visual Maths: leaderboard
Fab AI's AI for Education initiative asks models 237 early grade maths questions that each depend on an image, a subset it selected from the Math-Vision (Math-V) dataset. The set covers five strands: algebra, geometry, measurement, number and operations, and statistics and probability. This board is the overall percent correct, higher is better. Source: benchmarks.ai-for-education.org/leaderboards/maths.
Metric: Accuracy (%). Source: www.fab-ai.org. Saturation forecast: Around January 2027. 76 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.6 Sol | 91.14 | #19 |
| 2 | GPT-5.5 | 89.87 | #31 |
| 3 | Gemini 3.7 Flash | 89.45 | #23 |
| 4 | Claude Fable 5 | 86.92 | #16 |
| 5 | GPT-5.4 (Medium) | 86.5 | #79 (GPT-5.4) |
| 6 | Gemini 3.1 Pro (Preview) | 82.7 | #58 |
| 7 | Gemini 3 Flash | 80.59 | #99 |
| 8 | Gemini 3 Pro | 80.59 | #80 |
| 9 | Qwen 3.6 Plus | 80.59 | #120 |
| 10 | Claude Opus 5 | 80.59 | #12 |
| 11 | Qwen 3.5 Plus | 80.17 | #125 |
| 12 | Gemini 3.6 Flash | 78.48 | #51 |
| 13 | GPT-5.2 (Medium) | 78.48 | #110 (GPT-5.2) |
| 14 | Grok 4.5 | 77.64 | #48 |
| 15 | Qwen 3.5 Flash | 76.79 | #232 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=ai-for-education-visual-maths · How It Works · Data refreshed daily, snapshot 2026-10-20.