Nvidia debuts Groot N1, a major model for humanoid robotics

Nvidia launches what it calls an AI Foundation model for a humanoid robotics.

Announced in GTC 2025 in San JoseThe model, called Groot N1, is a model of “general”, trained in both synthetic and real data. In a video presenting Groot N1, Nvidia says it has a “double architecture” for “thinking fast and slow” inspired by human cognitive processes.

Groot N1 is an evolution of NVIDIA Project Groot, which the company started at its GTC conference last year. Project Groot focused on cases of industrial use, but Groot N1 expands the focus on humanoid robots in a number of various form factors.

Groot N1’s slow thinking allows a robot to also perceive reason for his environment and instructions, and then plan the right action to take, according to Nvidia. As for the fast -thinking system, it translates the aforementioned plan into robotic actions, including actions that include manipulating objects in multiple steps.

Groot N1 comes in open code. Along with the model, Nvidia launches simulation frames and drawings to generate synthetic training data.

“The age of the generalistic robotics is here,” says Nvidia CEO Jenson Huang in a statement.

In recent years, humanoid robots have attracted a lot of publicity. Companies like X1 and Figure try to create general-purpose robots that move more or less as humans. The challenges are great, but these companies claim that the technology has reached the point in which the massively manufactured humanoid robotic systems are a realistic short -term target.

Thehe very much disappointment In the latest history of robotics, they suggest that it will be easier to say than achieved.