Skip to content

Benchmark

VBench-Long temporal style

Scores are percentages.

Models measured
9
Scores recorded
9
Sources
1
IQ weight
0.49

Score distribution

24.7%
Mean
24.5%
Median
1.2%
Std dev
23.7%
10th percentile
26%
90th percentile
22.6%
Min
22.6%26.5%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.49 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 docs26.5%
2Veo 3· VBench data docs26%
3Video 01· VBench data docs25.6%
4Sora· VBench data docs25%
5Hunyuan Video2025-05-22· VBench data docs24.5%
6Kling V1.6· VBench data docs24.5%
7Vidu Q4· VBench data docs23.8%
8Mochi 1· VBench data docs23.7%
9LTX Video5s 768×512· VBench data docs22.6%

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