What changed
Google DeepMind unveiled Gemini Robotics 2.5, an enhanced version of its vision-language-action model designed for robotics. The update improves dexterity, reliability, and safety, enabling robots to operate in dynamic, real-world environments. Separately, the company introduced AlphaGenome 2, a next-generation AI model that predicts how DNA variations affect human cell types, aiding in disease research and drug development.
What this means in practice
Gemini Robotics 2.5 was tested across over 200 distinct tasks and achieved a 30% to 100% success rate depending on the task, completing tasks 20% to 30% faster than the previous version. The model works with various robots, including Apptronik’s Apollo 2 and Franka Duo platforms. AlphaGenome 2 provides detailed insights into how DNA changes affect gene activity in specific cell types, guiding researchers toward potential drug targets. It can predict DNA function readouts like gene expression and DNA methylation, which are critical for understanding epigenetic influences.
Limitations and availability
Gemini Robotics 2.5 is rolling out through DeepMind’s robotics partners, including Agile Robots SE, Apptronik, and Boston Dynamics Inc. AlphaGenome 2 is not approved for clinical or diagnostic use and is intended to guide research rather than replace experimental validation. The model adapts to new single-cell data without retraining, even from small samples. The supplied announcement does not specify the exact release timeline for AlphaGenome 2 beyond its announcement date.
