Physical AI Bench - Generation Overall: leaderboard
The "Generation Overall" subtask of Physical AI Bench. SHI Labs' benchmark for physical-world AI video models — scores physical understanding and (conditional) video generation for physical plausibility and correctness.
Metric: Overall Score (%). Source: huggingface.co. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Cosmos3-Super | 83.9 | |
| 2 | Cosmos3-Nano | 83.7 | |
| 3 | Veo-3 | 82.1 | |
| 4 | Cosmos3-Edge | 81.1 | |
| 5 | Cosmos-Predict2.5-14B | 81 | |
| 6 | Cosmos-Predict2.5-2B | 81 | |
| 7 | Wan2.2-I2V-A14B | 80.6 | |
| 8 | Wan2.2-TI2V-5B | 80.4 | |
| 9 | Cosmos-Predict2-14B-Video2World | 80 | |
| 10 | Wan2.1-I2V-14B-720P | 79.8 |
Interactive version: theaggregate.ai/benchmark?slug=physical-ai-bench-generation-overall · How It Works · Data refreshed daily, snapshot 2026-10-20.