Google Launches Gemini 4 Carbon, Its Most Powerful AI Model Yet

Google has unveiled Gemini 4 Carbon, its most powerful artificial intelligence model to date. Designed for cybersecurity defense and complex, long-running engineering tasks, Carbon claimed the top spot on the Artificial Analysis Intelligence Index v5.0 with a score of 59—the first Google model to do so.

Carbon is built to support the full cycle of cyber defense: identifying vulnerabilities, testing whether they can be exploited, and developing patches. In independent benchmark testing, it achieved industry-leading results:

• 88.2% on SEC-bench Pro for vulnerability discovery
• 54.8% on ExploitGym for exploitation
• 71.0% on CWE-bench for patching

Google’s Chrome team also found Carbon more than ten times better than rival large models at producing high-quality code patches.

Despite its impressive results, access is tightly controlled for now. Google is initially making Carbon available exclusively to vetted cybersecurity professionals through its Fairwind Program, with no firm date announced for broader public availability.

Introductory API pricing is $2 per million input tokens and $10 per million output tokens, matching its predecessor, Gemini 4 Argon, which launched two weeks earlier.

Source: Google Blog — Gemini 4 Carbon: Our Next Frontier Model