Gemini Robotics On-Device 2
Our most-efficient vision-language-action model (VLA). Built specifically to handle network constraints, and optimized to run locally on robotic devices.
Gemini Robotics On-Device 2 adapts quickly to completely new robots – with fewer than 200 examples. It can also adapt to a variety of new objects and situations.
Capabilities
Gemini Robotics On-Device 2 is optimized to run locally, and quickly adapt to new robot embodiments.
Optimized to run locally
Many robots need to operate without network latency or internet connectivity. Gemini Robotics On-Device 2 is built specifically to handle these constraints.
Fast hardware adaptation
Natively multi-embodiment, enabling fast adaptation to completely new robot embodiments, shapes, and sensors – with just a few hours of training.
Model information
- Name
- Gemini Robotics On-Device 2
- Status
- Trusted testers
- Input
- Image
- Text
- Action
- Output
- Action
- Model card
- View model card