1. [Publications](/publications)
2. World Simulation With Video Foundation Models for Physical AI
 
 # World Simulation With Video Foundation Models for Physical AI

  ![](/sites/default/files/styles/wide/public/publications/Transfer%20Static%20frame.png?itok=_a1BSLXK)

 World Simulation is the core to scalable physical AI development. At CoRL 2025, NVIDIA announced major updates to [Cosmos World Foundation Models (WFMs)](https://www.nvidia.com/en-us/ai/cosmos/) that let developers generate diverse data for accelerating training physical AI models at scale using text, image and video prompts.

[**Cosmos Predict 2.5**](https://research.nvidia.com/labs/dir/cosmos-predict2.5/) combines three WFMs into one, reducing complexity while enabling longer (up to 30s) video generation and multi-view outputs for richer simulations.

[**Cosmos Transfer 2.5**](https://research.nvidia.com/labs/dir/cosmos-transfer2.5/) is 3.5x smaller yet faster and higher quality, producing photorealistic data from spatial inputs and ground truth simulations.

All Cosmos models are openly available under the NVIDIA Open Model License <https://github.com/nvidia-cosmos>



 ## Authors



[Ming-Yu Liu](/person/ming-yu-liu)

 (View all contributors on page 35)

 

 

 ## Publication Date



Monday, September 29, 2025

 

 ## Research Area



[Autonomous Vehicles](/research-area/autonomous-vehicles)

[Physical AI](/research-area/physical-ai)

[Robotics](/research-area/robotics)

[World Simulation](/research-area/world-simulation)

 

 

 ## Uploaded Files



[World\_Simulation\_with\_Video\_Foundation\_Models\_for\_Physical\_AI.pdf](https://d1qx31qr3h6wln.cloudfront.net/publications/World_Simulation_with_Video_Foundation_Models_for_Physical_AI_0.pdf "Open file in new window")33.19 MB