Google unveiled Gemini 4 Carbon, its latest frontier AI model, focused on complex coding and cybersecurity

Detecting, preventing, and responding to cyberattacks using automated, model-driven systems.

Google DeepMind's frontier AI model, released on October 28, 2026, two weeks after Gemini 4 Argon's wide rollout, first only to selected cybersecurity defenders through Google's Fairwind Program; it ranked first on the Artificial Analysis Intelligence Index v5 with a score of 59.