Skip to content

Benchmark

VBench-Long appearance style

Scores are percentages.

Models measured
9
Scores recorded
9
Sources
1
IQ weight
0.21

Score distribution

21.9%
Mean
21.5%
Median
1.4%
Std dev
20.3%
10th percentile
23.6%
90th percentile
20.1%
Min
20.1%24.8%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.21 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
1Sora· VBench data docs24.8%
2Veo 3· VBench data docs23.6%
3Vidu Q12025-04-17· VBench data docs22.4%
4Hunyuan Video2025-05-22· VBench data docs22.2%
5Vidu Q4· VBench data docs21.5%
6LTX Video5s 768×512· VBench data docs21.5%
7Kling V1.6· VBench data docs20.8%
8Mochi 1· VBench data docs20.3%
9Video 01· VBench data docs20.1%

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