Gemini 4 Argon available to everyone
Gemini 4 Argon is designed primarily for demanding tasks such as refactoring large software projects, analyzing large amounts of documentation, detecting security vulnerabilities, and supporting scientific research. It supports a context window with up to one million shared tokens via an API, and Google promises up to 95% lower costs for stored input data. According to engineers, the model also delivers more direct responses and improved understanding of visual content.
Google reports notable progress in the use of autonomous agents. The number of proposed solutions is expected to increase by more than 35 %, while the number of canceled or failed fixes is expected to decrease. Among the more interesting use cases, Google highlights the overhaul of part of the Chrome browser source code, where the model helped reorganize a large set of historically created modules. The process, which involved rearranging several hundred files, was completed in a few weeks, according to the company, and is expected to contribute to greater browser stability and faster switching between tabs in the long term.



















