Key facts
- Gemini 4 Carbon shows significant gains in cybersecurity applications, including vulnerability detection, validation, and remediation.
- The model reportedly generates code that is free of vulnerabilities more than twice as often as Gemini 4 Argon.
- Google claims Carbon is 10 times better at generating high-quality patches compared to non-Google models.
- Carbon ranks fourth on the WebDev Arena leaderboard with an 89-point score increase.
- Gemini 4 Carbon resisted all attempted prompt-injection attacks in the Grey Swan evaluation.
- The model is being rolled out through Google's Fairwinds cyber defense program.
Alphabet's Google announced Gemini 4 Carbon, its newest frontier AI model, on Wednesday, aiming to bolster its competitive position against rivals like OpenAI and Anthropic in the advanced AI and cybersecurity sectors. This launch follows the recent release of Gemini 4 Argon. Google states that Gemini 4 Carbon represents a significant advancement in cybersecurity applications, demonstrating superior performance in vulnerability detection, validation, and remediation. The model reportedly generates secure code at more than double the success rate of its predecessor, Gemini 4 Argon, and is designed to handle complex, long-running engineering and coding tasks. Initially, access to Gemini 4 Carbon is being provided to a select group of cybersecurity defenders through Google's Fairwind Program for further testing of its safety protocols. The pricing for Gemini 4 Carbon is set at $2 per million input tokens and $10 per million output tokens, mirroring the introductory rates of Gemini 4 Argon.