MWS Vision Bench — leaderboard

Metric: Overall Score (%). Source: huggingface.co. 38 models tracked.

Top models

#ModelScore
1Qwen 3.7 Plus70.8
2Claude Opus 4.670.4
3Qwen 3.6 35B A3B69.4
4Qwen 3.6 27B69.1
5Gemini 2.5 Pro69
6Kimi K268.6
7GPT-5.468.3
8Claude Sonnet 4.668.2
9Gemini 3 Flash (Preview)68.1
10Gemini 3.1 Flash Lite (Preview)67.4
11Gemini 3.5 Flash67.4
12Gemini 2.5 Flash67.2
13Claude Opus 4.567
14Qwen 3.5 35B A3B67
15Claude Sonnet 4.566.9

Interactive version: aibenchmarks.dev/benchmark?slug=mws-vision-bench · How the rankings work · Data refreshed daily, snapshot 2026-07-20.