Physical AI Bench - Conditional Generation Overall: leaderboard
The "Conditional 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: Mean Quality Score. Source: huggingface.co. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Cosmos3-Nano | 10.35 | |
| 2 | Cosmos3-Super | 10.33 | |
| 3 | Wan2.2-Fun-5B-Control | 9.06 | |
| 4 | Wan2.2-Fun-A14B-Control | 8.71 | |
| 5 | Cosmos-Transfer2.5-2B | 8.24 |
Interactive version: theaggregate.ai/benchmark?slug=physical-ai-bench-conditional-generation-overall · How It Works · Data refreshed daily, snapshot 2026-10-20.