🤖 Powering the Next Generation of Smart Robots: Introducing Gemini Robotics 1.5

This is an exciting announcement about the progress in AI and robotics!

Here is the user-friendly, online version of your text, broken down for clarity and easier reading:

🤖 Powering the Next Generation of Smart Robots: Introducing Gemini Robotics 1.5

We are entering a new era of Physical Agents—robots that can perceive, plan, think, use tools, and act to solve complex, real-world tasks.

Building on the success of the original Gemini Robotics models, we're taking a major leap forward to create truly general-purpose intelligent robots.

🌟 Two New Models for Advanced Robotic Intelligence

We're introducing two powerful new models that unlock advanced "agentic" (self-planning) experiences for physical tasks:

Model NameCore CapabilityKey FeaturesGemini Robotics 1.5Vision-Language-Action (VLA): The robot's motor control and executor.Thinks Before Acting: Transparently shows its thought process before performing a step. Learns Across Robots: Skills learned on one robot can be quickly applied to others.Gemini Robotics-ER 1.5Vision-Language Model (VLM) for Embodied Reasoning (ER): The robot's high-level brain and planner.Advanced Planning: Creates detailed, multi-step plans for complex missions. Uses Tools: Can natively use digital tools like Google Search and other third-party functions. State-of-the-Art Spatial Understanding.

Export to Sheets

🧠 How the Models Work Together

These models work as an Agentic Framework to solve tough, multi-step problems:

1. High-Level Brain (Gemini Robotics-ER 1.5): Acts as the orchestrator.

   • It receives a complex task (e.g., "Sort these objects into the correct bins based on my local rules.").

   • It reasons about the physical world, uses tools (like searching the web for local recycling guidelines), and creates a detailed, step-by-step plan.

   • It then gives the plan as simple, natural language instructions to the action model.

2. Action & Execution (Gemini Robotics 1.5):

   • It takes the natural language instruction for a single step.

   • It uses its vision to understand the environment and executes the required motor commands.

   • It can even explain its thinking process for greater transparency.

🚀 What This Means for Developers

These advances will help developers build robots that are:

• More Capable: Can handle complex, multi-step instructions.

• More Versatile: Can adapt and understand their environment to solve general tasks.

GET STARTED TODAY

• Gemini Robotics-ER 1.5 is available now to developers via the Gemini API in Google AI Studio.

• Gemini Robotics 1.5 is currently available to select partners.

👉 Read more about building with the next generation of physical agents on the Developer Blog!