Google DeepMind Details Gemini Robotics 1.5 System
A company blog post outlined a dual-model architecture designed to give physical robots advanced reasoning and multi-step task completion abilities.
Dual-Model Architecture
Google DeepMind on Sep 26, 2025, published technical details on Gemini Robotics 1.5, a new system designed to give robots the ability to perform long, multi-step tasks in the physical world. The system employs a dual-model architecture that separates high-level reasoning from low-level execution.
The first component, Gemini Robotics-ER 1.5, is an "embodied reasoning" model that acts as a high-level planner. It can use external tools like Google Search to gather information needed to complete a task, such as checking local recycling rules. The second component, Gemini Robotics 1.5, is a vision-language-action (VLA) model that translates the …
Archive Access
This article is older than 24 hours. Create a free account to access our 7-day archive.