Google DeepMind introduces Gemini Robotics 2.5 for real-world robot tasks

Google DeepMind says it is introducing Gemini Robotics 2.5, its latest vision-language-action model for robots, and is rolling it out to robotics partners. In tests across more than 200 tasks, the model improved success rates by 30% to 100% over prior baselines and finished work 20% to 30% faster. A pilot with Apptronik and Mercedes-Benz on automotive production lines in Hungary and Germany used Apollo 2 for kitting and reached a 96% success rate.