Google announces Gemini 4 Argon with restricted initial access
Google has announced Gemini 4 Argon, describing it as its next frontier artificial intelligence model. The company says the system is designed to deliver frontier performance in complex workflows involving real-world software engineering, enterprise knowledge work and cybersecurity defense. Legal and finance work are among the enterprise applications mentioned by Google.
Koray Kavukcuoglu, Google’s chief AI architect and a senior vice president at Google DeepMind, said the model is initially available only to a set of “trusted cyber defenders.” The restriction means access is being expanded selectively rather than offered broadly at launch.
Kavukcuoglu also said Google is actively participating in the U.S. government’s voluntary process for pre-release model access. According to him, the company plans to widen availability gradually. Google says Gemini 4 Argon is already powering some of its own operations, although the announcement excerpt does not specify which systems or services are involved. No broader release schedule was provided in the available announcement.
Written by Sakhanda Wire AI from reporting by The Verge. Facts, quotes and full context are in the original article below.