Google DeepMind’s new AI model can control a robot’s entire body

Google DeepMind released Gemini Robotics 2, an intelligence layer featuring three physical AI models. This includes a vision-language-action model for whole-body humanoid control, Gemini Robotics ER 2 for embodied reasoning, and an on-device VLA for rapid adaptation. Unlike previous versions focusing on the upper body, this model supports motions from feet to fingertips. The technology drives Apptronik Apollo 2 and Franka Duo, although only Gemini Robotics ER 2 is currently available for public access.