OpenVLM MathVista - STA: leaderboard
OpenCompass OpenVLM MathVista statistical reasoning subcategory score.
Metric: Accuracy (%). Source: huggingface.co. Source: huggingface.co. Saturation forecast: Estimated already saturated. 284 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 89.4 | #98 |
| 2 | Gemini 2.5 Pro | 88.7 | #152 |
| 3 | GPT-5 Mini (2025-08-07) | 87.4 | #172 |
| 4 | InternVL3-78B | 87 | #352 |
| 5 | QVQ-72B-Preview | 86.7 | #440 |
| 6 | GPT-4.1 Mini | 86.4 | #355 |
| 7 | Taichu-VLR-7B | 86.4 | #526 |
| 8 | InternVL3-38B | 85 | #405 |
| 9 | InternVL3-14B | 84.7 | #505 |
| 10 | Doubao-1.5-Pro | 84.4 | #287 |
| 11 | VLAA-Thinker-Qwen2.5VL-7B | 84.4 | #555 |
| 12 | InternVL2.5-78B-CoT | 84.1 | #469 |
| 13 | Qwen 2 VL 72B | 83.4 | #460 |
| 14 | GPT-4o ChatGPT | 83.1 | |
| 15 | Qwen 2.5 VL 7B | 82.1 | #643 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mathvista-sta · How It Works · Data refreshed daily, snapshot 2026-10-20.