VideoVIBE - Theme and Legibility: leaderboard

Metric: Failure-diagnosis accuracy (%; questions of the Theme and Legibility subdimension of Visual and Motion Fidelity; zero-shot single-choice diagnosis from the interaction video and the page source code (Video+HTML), about 1.7K questions built from 6,338 verified failures of one-shot generated web applications recorded under human operation; free-form answers mapped deterministically to an option). Source: arxiv.org. Saturation forecast: Around December 2026. 13 models tracked.

Top models

#ModelScore
1Gemini 3.1 Pro (Preview)67.5
2Gemini 3 Flash61.25
3Gemini 2.5 Flash58.75
4Qwen 3.5 35B A3B47.5
5Gemini 3.1 Flash Lite35
6MiMo-V2.531.25
7Qwen 3 VL 8B (Thinking)28.75
8Qwen 3 VL 30B A3B (Thinking)27.5
9GLM-4.6V18.75
10Qwen 3 VL 30B A3B Instruct13.75
11Qwen 3 VL 8B Instruct10

Interactive version: theaggregate.ai/benchmark?slug=videovibe-theme-and-legibility · How It Works · Data refreshed daily, snapshot 2026-09-29.