Google launches its new AI model, Gemini 4 Carbon

Google has released its new AI model, Gemini 4 Carbon, which offers frontier performance in cybersecurity capabilities and long-term agentic workflows, providing access for developers, businesses, scientists, and consumers.

Google has released its new AI model, Gemini 4 Carbon, which delivers frontier performance in cybersecurity capabilities, agentic workflows, and complex information processing tasks, for developers, enterprises, scientists, and consumers.

Previously made available to cybersecurity defenders under the Fairwind Program, the Carbon model was developed following the Gemini 4 Argon model for tasks that push the boundaries of intelligence, such as cybersecurity defense, multi-step coding, debugging, and open-ended scientific research. While the Argon model remains ideal for general software engineering and information processing, the Carbon model focuses on the most demanding domains.

Gemini 4 Carbon is available to all paying developers and businesses via the API within Google AI Studio and through Gemini for enterprise, at the same introductory pricing as Gemini 4 Argon of $2 per million input tokens and $10 per million output tokens. The model is also rolling out gradually to all Google AI Ultra subscribers within Google Antigravity, the Gemini app, and AI Mode in Search.