AI news ·
Gemini Robotics 2.5 Reaches 96% Success on Mercedes-Benz Lines
Google’s Gemini Robotics 2.5 model reached a 96% success rate in automotive parts handling tests, improving task success by 30–100% over prior results.

Google DeepMind’s Gemini Robotics 2.5 model achieved a 96% success rate on automotive parts handling tasks in testing with Apptronik and Mercedes-Benz. Across 200 tests, the system improved task success by 30% to 100% compared to prior performance.
The tests also incorporated new safety guidance, which improved completion rates when the robot operated around obstacles and nearby people. The results were measured on production-representative tasks involving parts handling.