At a Glance · AI-generated

Google launched Gemini 4 Argon on September 30, its most advanced AI model to date, featuring one million output tokens designed for coding, knowledge work, and cybersecurity applications. The company restricted initial access to "trusted cyber defenders" and security professionals, citing safety concerns amid ongoing industry debates over AI risks. Google said the model can identify and patch critical software vulnerabilities, positioning it competitively with offerings from OpenAI and Anthropic, though analysts noted it does not establish clear technical superiority.