Nvidia unveils Cosmos 3, an open physical AI foundation model, to help robots and autonomous cars better understand the real world with limited training data
Nvidia unveiled Cosmos 3, an open AI world model designed to help robots, autonomous vehicles and other physical systems better understand and predict real-world environments.
Lead Source
How this story grew
More
Discussion
TechSnif Coverage
Nvidia Drops Cosmos 3 to Teach Robots About the Real World
Nvidia's new open AI model helps robots and self-driving cars understand physical environments with less training data.
Nvidia just pulled the curtain back on Cosmos 3, an open physical AI foundation model built to help machines make sense of the messy real world.
The model targets robots, autonomous vehicles, and other physical AI systems that need to understand and predict real-world environments. The big selling point: it works with limited training data. That's a significant deal for robotics and AV companies that can't easily collect millions of real-world scenarios.
By making Cosmos 3 open, Nvidia is positioning itself as the infrastructure layer for physical AI — not just selling GPUs, but providing the foundational intelligence that runs on them.
The move signals Nvidia's deepening push beyond chips and into the software stack that powers autonomous systems. Robots and self-driving cars need world models to operate safely. Nvidia wants to be the one supplying them.