Google DeepMind’s new Gemini Robotics 2.5 vision-language-action model enhances robots’ dexterity and reliability. In a real-world factory pilot, Apptronik’s Apollo 2 androids used the model to sort automotive parts with a 96% success rate.


Google DeepMind’s new Gemini Robotics 2.5 vision-language-action model enhances robots’ dexterity and reliability. In a real-world factory pilot, Apptronik’s Apollo 2 androids used the model to sort automotive parts with a 96% success rate.