MathSpatial — leaderboard
Metric: Overall Accuracy (%). Source: shuolucs.github.io. 17 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 | 58.5 |
| 2 | Gemini 2.5 Flash | 48.5 |
| 3 | Gemini 2.5 Pro | 44.9 |
| 4 | Claude Sonnet 4 | 26.7 |
| 5 | Llama 4 Maverick | 23.8 |
| 6 | GPT-4.1 | 22.6 |
| 7 | Claude 3.5 Sonnet | 22.3 |
| 8 | Claude 3.7 Sonnet | 21.5 |
| 9 | GLM-4.5V | 21 |
| 10 | GPT-4o | 19.6 |
| 11 | InternVL3-8B | 17.4 |
| 12 | Llama 3 8B | 15 |
Interactive version: theaggregate.ai/benchmark?slug=mathspatial · How the rankings work · Data refreshed daily, snapshot 2026-07-22.