Updated

Google DeepMind revealed Gemini Robotics 2.5, a new robotics model, along with an open safety benchmark called ASIMOV-jailbreak, Ars Technica reported on November 10, 2026. The company said the new model is up to twice as successful at some tasks. The benchmark is released as open material for testing robot safety. The excerpt does not give the specific tasks or results behind the success rates.