Google's Gemini 4 Carbon tops AI rankings with cybersecurity focus

Google announced Gemini 4 Carbon on October 28, calling it its strongest AI model so far, with a focus on cybersecurity and long-running software engineering where Carbon can find software flaws, check whether attackers could exploit them, and repair their root causes. Carbon arrives 15 days after Gemini 4 Argon became broadly available, and it improves on Argon on every benchmark Google published.

Carbon ranks #1 on the recently updated v5.0 of Artificial Analysis Intelligence Index at 59, with leading results in SEC-bench Pro (88.2%) for finding vulnerabilities, ExploitGym (54.8%) for exploitation, and CWE-bench (71.0%) for patching. This score of 59 places it one point above Claude Opus 5.5, ending months in which Google's best model trailed its rivals. According to the company, Carbon makes roughly half as many unsupported claims as Argon, writes caveats directly into its work instead of guessing and flags conflicting information before acting.

Google is making Carbon available to a subset of cyber defenders during the week of the announcement, then gradually adding trusted partners through its Fairwind Program, while gathering feedback and refining guardrails before offering Carbon to developers, businesses and consumers through its products. The Fairwind Program already has more than 650 partners, gives priority to governments, national cyber agencies, critical infrastructure operators and major technology platforms, and runs background checks on applicants. Google is also taking part in the U.S. government's voluntary testing process for models before release.