Gemini Robotics 1.5

Gemini Robotics 1.5 is Google DeepMind's latest Vision-Language-Action (VLA) model, building on the original Gemini Robotics. It processes visual input (robot camera images) and text instructions and outputs motor commands to control robot joints. This is a key distinction from VLM/LLM models: the model does not describe what it sees, but directly controls physical motion.

The model generalizes across new instructions, actions, and visual contexts, and a single model can operate across diverse robotic platforms (ALOHA, Bi-arm Franka, humanoid Apptronik Apollo). Paired with Gemini Robotics-ER 1.6, it forms a complete system for physical robot control.