Gemini 4 Argon
Google has released Gemini 4 Argon globally for complex, multi-step work across coding, professional documents, science, and cybersecurity. Google reports that the model can also produce interactive 3D experiences, including a browser racing game built from photos of a real desk.
The clearest practical example is inside Google itself. The company says Argon worked overnight on its open-source Agent Development Kit and found changes that cut preparation time for long sessions by 40% and accelerated cold imports by 21%.
Developers and enterprises can use the API now. Everyday access is available through the Gemini app and AI Mode in Search for Google AI Pro and Ultra subscribers.
Personal Android apps
Google Play now lets U.S. adults in selected markets build a custom Android app with Google AI Studio and run it privately on their own device. A standard Google account is enough, with no developer account required.
These personal apps remain separate from the public Play Store catalog. Sharing with a small circle of friends and family is planned for the coming months; publishing publicly will still require a professional developer account and the usual reviews.
Also
- Run a private agent on your own hardware Liquid AI released an open-weight reasoning model with 24 billion parameters but only 2 billion active per token, designed to call tools and handle multi-step tasks locally.
- Give different agents shared Notion context Notion 3.8 lets compatible agents read workspace context and update documents or task trackers, while admins can inspect and revoke individual connections.
- Detect scam conversations on the phone Google DeepMind and Singapore’s GovTech built a locally tuned model that warns about manipulative conversation patterns; deployment will be phased, consent-based, and processed on-device.
Pass it on