Google DeepMind has unveiled Gemini Robotics 2, a version of its Gemini AI that can control humanoid robots for the first time. The system provides “intelligent whole-body control,” enabling humanoids to perform tasks autonomously, as demonstrated in a video showing robots screwing in light bulbs and tying up bin bags. Gemini Robotics 2 combines a vision language model (VLM) with two vision language action (VLA) models, which facilitates communication with humans and environmental understanding. Google DeepMind described the release as a “milestone” toward achieving physical AGI (artificial general intelligence), where robots possess human-level abilities. Carolina Parada, head of robotics at Google DeepMind, wrote in a blog post: “For the first time, our model can now control entire humanoid robots, translating intent into intelligent whole-body control.”