Google unveils Gemini 4 Carbon AI model aimed at cybersecurity
Google introduced Gemini 4 Carbon, an AI model designed to improve vulnerability detection and secure code generation, and made it initially available to select cybersecurity teams.
Alphabet's Google announced the Gemini 4 Carbon model, the latest addition to its Gemini AI family, as competition intensifies among firms developing advanced coding and cybersecurity tools. The new model is reported to achieve state-of-the-state performance on tasks such as vulnerability discovery, exploitation and software patching, and to generate secure code at more than twice the success rate of the earlier Gemini 4 Argon in internal tests. Google said Carbon is built to handle complex, long-running engineering and coding assignments and will first be offered to a limited group of cybersecurity defenders through the company's Fairwind Program while further safeguards are evaluated. Pricing is set at $2 per million input tokens and $10 per million output tokens, matching the introductory rates for Gemini 4 Argon.