Google DeepMind releases Gemini Robotics 2.5 vision-language-action model for real-world robot tasks

According to Google DeepMind, Gemini Robotics 2.5 is a vision-language-action model tested across over 200 distinct tasks to increase robot dexterity, reliability, and safety in unstructured environments. The model expands the breadth of skills robots can learn and accelerates acquisition speed, building on Gemini 4 Argon's robotics research capabilities. Google is rolling the model out to robotics partners.