SciVQR - Mathematics: leaderboard
The mathematics questions of SciVQR (2026), 3,254 image-based scientific questions from competitions, entrance exams and graduate-level sources in English and Chinese, answered with zero-shot chain-of-thought. Accuracy in percent; higher is better.
Metric: Mathematics accuracy (%) with zero-shot chain-of-thought prompting on SciVQR (3,254 image-based scientific questions from competitions, entrance exams and graduate-level sources, 78 percent multiple-choice and 22 percent free-form, in English and Chinese, high-school to graduate level); higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 15 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | O4 Mini | 84.5 | #368 | |
| 2 | Gemini 2.5 Flash (Preview 04-17) | 83.2 | #453 | |
| 3 | O1 | 79.2 | #403 | |
| 4 | MiMo-VL-7B-RL-2508 | 65.5 | #823 | |
| 5 | Keye-VL-1.5-8B | Thinking | 61.1 | #1720 |
| 6 | Qwen 2.5 VL 72B Instruct | 54.3 | #841 | |
| 7 | InternVL3-78B-Instruct | 47.7 | #1415 | |
| 8 | GPT-4o | 47.4 | #751 | |
| 9 | InternVL3-14B-Instruct | 36.9 | #1801 | |
| 10 | InternVL3-8B-Instruct | 36.4 | #2872 | |
| 11 | Qwen 2.5 VL 7B Instruct | 35.4 | #1902 | |
| 12 | MiniCPM-o-2.6 | 26.8 | #2423 | |
| 13 | LLaVA-OneVision-7B | 21.9 | #2981 | |
| 14 | Molmo-7B-D-0924 | 20.3 | #7207 | |
| 15 | LLaVA-NeXT-7B | 8 | #5032 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=scivqr-mathematics · How It Works · Data refreshed daily, snapshot 2026-11-21.