Skip to content

Benchmark

VBench-Long background consistency

Scores are percentages.

Models measured
9
Scores recorded
9
Sources
1
IQ weight
0.03

Score distribution

97%
Mean
97%
Median
0.358%
Std dev
96.6%
10th percentile
97.3%
90th percentile
96.3%
Min
96.3%97.6%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.03 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
1Hunyuan Video2025-05-22· VBench data docs97.6%
2Mochi 1· VBench data docs97.3%
3LTX Video5s 768×512· VBench data docs97.2%
4Vidu Q12025-04-17· VBench data docs97.1%
5Video 01· VBench data docs97%
6Veo 3· VBench data docs96.9%
7Kling V1.6· VBench data docs96.8%
8Vidu Q4· VBench data docs96.6%
9Sora· VBench data docs96.3%

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