From the source
Google DeepMind introduced Gemini Robotics 2.5, a vision-language-action model it says is its most capable to date, widening the range of skills robots can learn and how fast they acquire them.
- The company evaluated the model across more than 200 distinct tasks, including serving food with tongs, tying a bow on a gift box and pipetting liquid, and says it reaches a 30% to 100% higher success rate than previous baselines while finishing tasks 20% to 30% faster.
- In a proof-of-concept on Mercedes-Benz production lines in Kecskemét, Hungary, and Sindelfingen, Germany, Apptronik's Apollo 2 humanoid used the model to pick and pack automotive parts with a 96% success rate.
- DeepMind says inference-time reasoning guidance steers the robot around physical hazards in real time, lifting safe task completion by 33% around obstacles and 43% around nearby people.
- The company is also releasing ASIMOV-Jailbreak, an open-source benchmark of hundreds of prompts that disguise unsafe requests across dozens of scenes, on which Gemini 4 Argon reaches 99.5% jailbreak resistance against an 82% baseline for Gemini Robotics ER 2.
- Gemini Robotics 2.5 is rolling out to the company's robotics partners.





