Google DeepMind has released Gemini Robotics On-Device, a new AI model that enables robots to operate completely offline without requiring cloud connectivity. This standalone vision-language-action (VLA) model represents a significant step toward truly autonomous robots that can perform complex tasks like tying shoes or folding shirts while maintaining privacy and reliability in environments with poor internet connectivity.
What you should know: The new on-device model eliminates the cloud dependency that limited previous versions of Google’s robotics AI.
- Unlike the hybrid system that required both local and cloud-based processing, this standalone model runs entirely on the robot itself.
- The model maintains nearly the same accuracy as the cloud-hybrid version while providing full autonomy.
- Developers can now customize the AI for specific tasks with as few as 50 to 100 demonstrations through teleoperation.
How it works: The system leverages Gemini’s multimodal understanding to generate robot actions in real-time.
- “It’s drawing from Gemini’s multimodal world understanding in order to do a completely new task,” explains Carolina Parada, head of robotics at Google DeepMind.
- “In that same way Gemini can produce text, write poetry, just summarize an article, you can also write code, and you can also generate images. It also can generate robot actions.”
- The model processes visual data locally, eliminating the delays associated with cloud-based processing.
Key capabilities: The on-device model excels at straightforward tasks but has limitations with complex multi-step reasoning.
- It can handle traditionally difficult robotics tasks like shoe-tying and shirt-folding out of the box.
- More complex tasks requiring multi-step reasoning, such as sandwich-making, would likely need the more powerful cloud-hybrid model.
- “When we play with the robots, we see that they’re surprisingly capable of understanding a new situation,” Parada tells Ars Technica.
Why this matters: Offline operation addresses critical real-world deployment challenges for enterprise and healthcare robotics.
- Robots can now function reliably in environments with spotty or non-existent internet connectivity.
- Local processing enhances privacy protection, particularly valuable in healthcare settings where sensitive data is involved.
- The standalone approach makes robots more dependable in mission-critical situations where cloud outages could disable operations.
Safety considerations: Google implements multiple safety layers, though developers using the on-device model must build their own protections.
- The full Gemini Robotics system includes reasoning about safety, a VLA that produces options, and low-level controllers with safety-critical components.
- Google recommends developers connect to the Gemini Live API for safety layers and implement low-level controllers for critical safety checks.
- “With the full Gemini Robotics, you are connecting to a model that is reasoning about what is safe to do, period,” says Parada.
What’s next: The technology is currently available through Google’s trusted tester program, with broader implications for the robotics industry.
- The current release is based on Gemini 2.0, with the robotics team typically trailing one version behind the main Gemini development.
- Parada notes that Gemini 2.5 represents “a massive improvement in chatbot functionality,” suggesting similar advances may come to robotics.
- Developers interested in testing can apply for access to explore new applications and environments for AI-powered robots.
Recent Stories
DOE fusion roadmap targets 2030s commercial deployment as AI drives $9B investment
The Department of Energy has released a new roadmap targeting commercial-scale fusion power deployment by the mid-2030s, though the plan lacks specific funding commitments and relies on scientific breakthroughs that have eluded researchers for decades. The strategy emphasizes public-private partnerships and positions AI as both a research tool and motivation for developing fusion energy to meet data centers' growing electricity demands. The big picture: The DOE's roadmap aims to "deliver the public infrastructure that supports the fusion private sector scale up in the 2030s," but acknowledges it cannot commit to specific funding levels and remains subject to Congressional appropriations. Why...
Oct 17, 2025Tying it all together: Credo’s purple cables power the $4B AI data center boom
Credo, a Silicon Valley semiconductor company specializing in data center cables and chips, has seen its stock price more than double this year to $143.61, following a 245% surge in 2024. The company's signature purple cables, which cost between $300-$500 each, have become essential infrastructure for AI data centers, positioning Credo to capitalize on the trillion-dollar AI infrastructure expansion as hyperscalers like Amazon, Microsoft, and Elon Musk's xAI rapidly build out massive computing facilities. What you should know: Credo's active electrical cables (AECs) are becoming indispensable for connecting the massive GPU clusters required for AI training and inference. The company...
Oct 17, 2025Vatican launches Latin American AI network for human development
The Vatican hosted a two-day conference bringing together 50 global experts to explore how artificial intelligence can advance peace, social justice, and human development. The event launched the Latin American AI Network for Integral Human Development and established principles for ethical AI governance that prioritize human dignity over technological advancement. What you should know: The Pontifical Academy of Social Sciences, the Vatican's research body for social issues, organized the "Digital Rerum Novarum" conference on October 16-17, combining academic research with practical AI applications. Participants included leading experts from MIT, Microsoft, Columbia University, the UN, and major European institutions. The conference...