Open LMM Reasoning - MathVista - STA — leaderboard

Metric: Accuracy (%). Source: huggingface.co. 100 models tracked.

Top models

#ModelScore
1InternVL3-78B87
2QVQ-72B-Preview86.7
3GPT-4.1 Mini86.4
4InternVL3-38B85
5InternVL3-14B84.7
6Doubao-1.5-Pro84.4
7Qwen 2 VL 72B83.4
8GPT-4o ChatGPT83.1
9GPT-4.1 (2025-04-14)81.7
10InternVL2.5-78B81.4
11Ovis2-8B80.7
12Gemini 2.0 Flash80.1
13Claude 3.5 Sonnet (20241022)79.1
14InternVL3-8B78.4
15Gemini 1.5 Pro (002)77.7

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