Skip to content

Benchmark

VBench-Long multiple objects

Scores are percentages.

Models measured
9
Scores recorded
9
Sources
1
IQ weight
0.75

Score distribution

67.9%
Mean
66.7%
Median
14.2%
Std dev
50.5%
10th percentile
82.2%
90th percentile
45.4%
Min
45.4%93.9%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.75 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 docs93.9%
2Veo 3· VBench data docs82.2%
3Video 01· VBench data docs76%
4Sora· VBench data docs70.8%
5Hunyuan Video2025-05-22· VBench data docs66.7%
6Kling V1.6· VBench data docs64%
7Vidu Q4· VBench data docs61.7%
8Mochi 1· VBench data docs50.5%
9LTX Video5s 768×512· VBench data docs45.4%

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