Benchmark
Physical AI Bench Generation Subject Consistency
Scores are percentages.
- Models measured
- 4
- Scores recorded
- 4
- Sources
- 1
- IQ weight
- No weight published
Score distribution
- 91.6%
- Mean
- 91.6%
- Median
- 0.75%
- Std dev
- 90.6%
- 10th percentile
- 92.7%
- 90th percentile
- 90.6%
- Min
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Cosmos 3 Super· Physical AI Bench data | 92.7% |
| 2 | Wan 2.2 I2v A14b· Physical AI Bench data | 91.6% |
| 3 | Veo 3· Physical AI Bench data | 91.4% |
| 4 | LTX-Video-13B· Physical AI Bench data | 90.6% |
Data credits: Physical AI Bench (SHI Labs), MIT (licence)