Google has introduced Gemini 4 Argon, a new frontier model aimed at software engineering, science, robotics and cyber defense. It is available now to Google AI Pro and Ultra subscribers.
Google says Gemini 4 Argon brings gains in reasoning, visual perception, spatial understanding and writing. For developers, a headline feature is a 1M-token output window, which Google says allows full multi-file applications, large refactors and end-to-end translation of legacy files or database schemas in a single pass.
Google describes one engineering use case at Chrome. Argon helped the team separate core webpage handling from a legacy folder of about 6 million lines inside Chromium's 40 million-line codebase. Work that would have taken months was finished in weeks, and Google says this will let live webpages move between tabs, side panels and floating windows without reloading.
In security, Argon ties for first place on CWE-Bench at 68% for vulnerability remediation. Google calls it its most resilient model against indirect prompt injection, and says it is expanding to hundreds more Fairwind Program partners. Wiz is using the model in its Scan for Good initiative.
Google also lists safeguards, including refusal of harmful requests, monitoring of the model's chain-of-thought and actions, and hardened sandboxes. Partners such as Box, Roboflow and Kepler gave positive early evaluations, according to Google. Academic scientists can apply for six months of free top-tier access in Google Antigravity, subject to eligibility and capacity, while defenders can apply to the Fairwind Program.
Source: Hacker News (front page)
