OpenVLM MathVista - NUM — leaderboard

Metric: Accuracy (%). Source: huggingface.co. 284 models tracked.

Top models

#ModelScore
1GPT-5 (2025-08-07)61.8
2Gemini 2.5 Pro60.4
3GPT-4o ChatGPT59.7
4InternVL3-78B55.6
5GPT-5 Mini (2025-08-07)54.9
6GPT-4.1 (2025-04-14)52.8
7Qwen 2 VL 72B51.4
8GPT-4.551.4
9InternVL3-38B50.7
10Gemini 2.0 Flash50
11InternVL3-14B48.6
12Gemini 1.5 Pro (002)47.2
13Pixtral-12B47.2
14Llama 3.2 90B Vision Instruct47.2
15InternVL3-8B46.5

Interactive version: theaggregate.ai/benchmark?slug=openvlm-mathvista-num · How the rankings work · Data refreshed daily, snapshot 2026-07-22.