Gemini Robotics-ER 1.6

Gemini Robotics-ER 1.6 (Embodied Reasoning) is a Vision-Language Model (VLM) developed by Google DeepMind, built on the Gemini 3.0 Flash architecture. It specializes in spatial and physical reasoning for robotics — including precision pointing, task planning, success detection, and industrial instrument reading.

The model processes text, image, audio, and video inputs (up to 128K token context) and generates text outputs. It can natively call external tools (Google Search, VLA models, user-defined functions) and combine visual reasoning with code execution (agentic vision). It serves as a high-level reasoning module in robotic systems and does not directly generate motor control commands.