DeepMind Technologies Limited, a Google LLC subsidiary, unveiled on Thursday the launch of Gemini Robotics 1.5, its most advanced vision-language-action (VLA) model available to select partners. A companion model, Gemini Robotics-ER 1.5, will also be available to developers through the Gemini API, a programmatic interface, in Google AI Studio.
The Gemini Robotics 1.5 model is designed to bring agentic AI into the physical world by enabling robots to perceive, plan, and execute multi-step tasks, such as sorting laundry, tidying, recycling, and trash separation. Meanwhile, Gemini Robotics-ER 1.5 acts as a high-level reasoning system, creating detailed, multi-step plans and achieving cutting-edge results in spatial reasoning benchmarks.
Google said the new models mark a step toward building general-purpose robots, enabling them to “reason, plan, actively use tools and generalize,” noting the models were designed with safety and alignment principles in mind,
Administrator and Writer
