During his CES 2025 keynote, Nvidia CEO Jensen Huang explained that the platform includes advanced generative models, improved tokenizers, safety features, and a faster video processing system, all designed to boost the development of physical AI systems like self-driving cars and robots, reports venturebeat.com.
Developing physical AI models is expensive and needs a lot of real-world data and testing. Cosmos World Foundation Models (WFMs) make it easier for developers to create large amounts of realistic, physics-based synthetic data to train and test their models. Developers can also customize these models by adjusting the Cosmos WFMs to their needs.
Cosmos models will be available with an open license to help speed up progress in the robotics and self-driving car communities. Developers can preview the first models in the Nvidia API catalog or download the models and fine-tuning tools from the Nvidia NGCTM catalog or Hugging Face.
“It is trained on 20 million hours of video,” Huang said. “Nvidia Cosmos. It’s about teaching the AI to understand the physical world.”
Leading robotics and automotive companies, including 1X, Agile Robots, Agility, Figure AI, Foretellix, Fourier, Galbot, Hillbot, IntBot, Neura Robotics, Skild AI, Virtual Incision, Waabi, and XPENG, along with ridesharing giant Uber are among the first to adopt Cosmos.
Jensen Huang, founder and CEO of Nvidia, in a statement said, “The ChatGPT moment for robotics is coming. Like large language models, world foundation models are fundamental to advancing robot and AV development, yet not all developers have the expertise and resources to train their own,”.
CEO added, “We created Cosmos to democratize physical AI and put general robotics in reach of every developer.”
Bd-pratidin English/ Afia
