Google has launched a new frontier AI model called Gemini 4 Argon, and early access is going first to a small group of trusted cybersecurity defenders rather than the public. The model is built to handle long, complex tasks in one go, spanning software engineering, finance, legal work and cyber defense, and Google says it is already reshaping how its own engineers build products.
Argon is rolling out through a program called Fairwind, even as Google works alongside the US government's voluntary review process for new frontier models. Wider access for developers, businesses and everyday users will follow only once more safety testing is done, the company said.
