Gemini Robotics 2 brings whole body intelligence to robots
deepmind ·
Google DeepMind introduced Gemini Robotics 2, an AI system for robots that provides whole-body control, advanced dexterity, and multi-robot collaboration. It includes three models: Gemini Robotics 2, a vision-language-action model for motor control; Gemini Robotics ER 2, an embodied reasoning model for planning and human communication; and Gemini Robotics On-Device 2, an efficient model that runs locally and adapts to new robot bodies in hours. The system can control humanoid robots like Apptronik's Apollo 2 to perform tasks such as walking, reaching, and manipulating objects, and it enables delicate actions like tying knots with five-fingered hands or gripping with standard grippers. Gemini Robotics ER 2 also supports longer multi-step tasks and teamwork between different robots. The company emphasizes safety, introducing the ASIMOV-Agentic benchmark and improving constraint following and human-proximity detection. Gemini Robotics ER 2 is available on Google AI Studio and in private preview, while the VLA and on-device models are available to early-access partners.
deepmind