FrontierThe story, in brief

Gemini Robotics On-Device brings AI to local robotic devices

Google just shipped a robotics model that runs locally. No cloud calls. No latency.

Illustration of a transparent lens revealing connected networks across layers of paper.
Exploring the next frontier of AI research.AI illustration by KeyNews
The KeyNews take

Why it matters

Google DeepMind is extending Gemini beyond text/vision into embodied AI with on-device inference—signaling a shift toward edge deployment of multimodal models and raising the stakes in the robotics AI race against OpenAI and Tesla.

The key facts

5 to know
  1. Gemini Robotics On-Device model announced

  2. On-device inference capability (no cloud dependency)

  3. General-purpose dexterity and fast task adaptation claimed

  4. Multimodal robotics model release

  5. Published June 24, 2025

Go to the source

Google DeepMind Blogdeepmind.google

Publisher excerpt: We’re introducing an efficient, on-device robotics model with general-purpose dexterity and fast task adaptation.
Read original report
Back to today's editionMore frontier news

Keep reading

Related stories

More from Frontier