Gemini Robotics

Gemini Robotics is a family of AI models for robots that Google DeepMind first introduced in 2025. It consists of vision-language-action models, which take images and language instructions and output robot actions, and embodied reasoning models (Gemini Robotics ER), which understand space and plan tasks.

An On-Device variant is compressed to run on the robot itself. The models are typically combined so that the reasoning model plans and the action model produces the physical motion.

This entry is based on AIPOST articles and widely known facts. If something is wrong, please send us a correction request.