Google DeepMind releases Gemini Robotics 2.5 for robotics partners
Article timeline
The articles in this story. Times are ET.
🚨 AI News | TestingCatalogXGoogle announces Gemini Robotics 2.5 for physical AIAIGoogle announces Gemini Robotics 2.5, a new model for physical AI, built with support from its Gemini 4 Argon model. In a proof of concept with Apptronik and Mercedes-Benz, Gemini Robotics 2.5 helped the Apollo 2 robot pick and organize car parts for assembly with 96% accuracy.
Google DeepMindOfficialGoogle adds safety measures to Gemini Robotics 2.5 for hazard detectionAIGoogle DeepMind says it has built new safety measures into Gemini Robotics 2.5, giving robots ways to spot hazards, steer away from humans, and test against jailbreaks. The post links to a page with more details on the model.
- Google DeepMindOfficialGoogle DeepMind's Gemini Robotics 2.5 hits 96% accuracy on car-parts assembly
AIGoogle DeepMind reports that Gemini Robotics 2.5 helped Apptronik's Apollo 2 robot pick and organize car parts for assembly with 96% accuracy in a proof of concept with Mercedes-Benz.
- Google DeepMindOfficialGemini 4 Argon helps build Gemini Robotics 2.5, speeding up training
AIGoogle DeepMind says Gemini 4 Argon helped build Gemini Robotics 2.5, speeding up months of training. The model creates virtual practice spaces from a single camera view and reviews replays of its own mistakes to fix its technique. Google DeepMind says this drives stronger performance on the real robot.
- Google DeepMindOfficialGoogle DeepMind introduces Gemini Robotics 2.5, a physical AI for robots
AIGoogle DeepMind introduces Gemini Robotics 2.5, its latest physical AI, which it says can learn hundreds of everyday skills. The skills include folding laundry, serving pastries, and autonomously sorting car parts on a factory floor. The post promises a snapshot of its capabilities in a thread.
- Ars Technica · AINewsGoogle DeepMind reveals Gemini Robotics 2.5 and an open safety benchmark
AIGoogle DeepMind has revealed Gemini Robotics 2.5, a new version of its robotics platform. Google says the model is much better at dexterous tasks, evaluated across more than 200 activities such as serving food with tongs and using a pipette. Google also showed Apptronik's Apollo 2 humanoid, running Gemini Robotics 2.5, tying a bow with five-fingered hands.
- Google DeepMindOfficialPickGoogle DeepMind releases Gemini Robotics 2.5 for real-world robot dexterity
AIGoogle DeepMind introduces Gemini Robotics 2.5, a vision-language-action model that it says improves dexterity, reliability, and safety for robots. Across 200 tasks, the company reports 30% to 100% higher success rates and 20% to 30% faster task completion than previous baselines. In a Mercedes-Benz production pilot, Apptronik's Apollo 2 humanoid reached a 96% success rate on automotive parts kitting, and the model is rolling out to robotics partners.