Google Releases Gemini 4 Argon, Frontier AI Model Built for Autonomous Vulnerability Detection and Patching

Google has launched Gemini 4 Argon, its most capable frontier AI model to date, specifically engineered to identify and remediate critical software vulnerabilities without human intervention. The model, which expanded context handling to 1 million tokens from the previous 64,000-token limit, is currently available only to trusted cybersecurity organizations through Google's Fairwind Program, which counts more than 650 participating partners including government agencies and Cloud customers.