Nvidia has unveiled a new AI model specifically designed for humanoid robots. The model, called Groot N1, was introduced at the GTC 2025 event in San Jose. It is described as a “generalist” model, trained on both synthetic and real data. According to Nvidia, Groot N1 features a “dual-system architecture” inspired by human cognitive processes, allowing for both quick thinking and deliberative decision-making.

Groot N1 is an advancement of Nvidia’s Project Groot, which was originally focused on industrial applications. However, Groot N1 expands its scope to humanoid robots of various shapes and sizes. The slow-thinking system of Groot N1 enables robots to perceive their surroundings, make decisions, and plan their actions accordingly. On the other hand, the fast-thinking system translates these plans into physical actions, such as manipulating objects through multiple steps.

Nvidia has made Groot N1 available as an open-source model. In addition, they are providing simulation frameworks and blueprints for generating synthetic training data. Nvidia’s CEO, Jensen Huang, stated, “The age of generalist robotics is here.”

Humanoid robots have garnered significant attention in recent years, with companies like X1 and Figure working on creating versatile robots that mimic human movements. While the challenges are substantial, these companies believe that mass-produced humanoid robotic systems are within reach in the near future.