Google DeepMind is introducing two novel models, derived from Gemini 2.0, specifically designed for applications in robotics.

• Gemini Robotics represents the most sophisticated vision-language-action model to date, empowering robots to execute a broader range of tasks in real-world environments than previously possible.
• Gemini Robotics-ER allows roboticists to develop and implement their own programs utilizing Gemini’s embodied reasoning capabilities.

These models are currently being tested in collaboration with partners and select testers. For more information, visit the Google DeepMind blog.