OpenVLM MathVista - NUM — leaderboard
Metric: Accuracy (%). Source: huggingface.co. 284 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 (2025-08-07) | 61.8 |
| 2 | Gemini 2.5 Pro | 60.4 |
| 3 | GPT-4o ChatGPT | 59.7 |
| 4 | InternVL3-78B | 55.6 |
| 5 | GPT-5 Mini (2025-08-07) | 54.9 |
| 6 | GPT-4.1 (2025-04-14) | 52.8 |
| 7 | Qwen 2 VL 72B | 51.4 |
| 8 | GPT-4.5 | 51.4 |
| 9 | InternVL3-38B | 50.7 |
| 10 | Gemini 2.0 Flash | 50 |
| 11 | InternVL3-14B | 48.6 |
| 12 | Gemini 1.5 Pro (002) | 47.2 |
| 13 | Pixtral-12B | 47.2 |
| 14 | Llama 3.2 90B Vision Instruct | 47.2 |
| 15 | InternVL3-8B | 46.5 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mathvista-num · How the rankings work · Data refreshed daily, snapshot 2026-07-22.