Skip to content

Benchmark

VBench-Long aesthetic quality

Scores are percentages.

Models measured
9
Scores recorded
9
Sources
1
IQ weight
0.43

Score distribution

62.3%
Mean
63%
Median
2.9%
Std dev
59.8%
10th percentile
64.8%
90th percentile
56.9%
Min
56.9%67.3%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.43 for video-generation models, measured over 8 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Vidu Q12025-04-17· VBench data docs67.3%
2Kling V1.6· VBench data docs64.8%
3Veo 3· VBench data docs63.8%
4Sora· VBench data docs63.5%
5Video 01· VBench data docs63%
6Vidu Q4· VBench data docs60.9%
7Hunyuan Video2025-05-22· VBench data docs60.3%
8LTX Video5s 768×512· VBench data docs59.8%
9Mochi 1· VBench data docs56.9%

Data credits: VBench (Vchitect), MIT (licence)