Google Unveils Gemini 4 Carbon, a New Frontier AI Model Focused on Cybersecurity

Google has announced Gemini 4 Carbon, the successor to Gemini 4 Argon, as a new frontier AI model that it will not release to the public yet. According to Google, Gemini 4 Carbon achieves a new threshold in coding, agentic tasks, and scientific analysis. In cybersecurity, the model reportedly shows a “significant leap” in vulnerability detection, validation and remediation, and is twice as likely to generate code free of vulnerabilities compared with Argon. The Chrome team claims Carbon is ten times better at producing high‑quality patches than non‑Google models, and it has risen 89 points on the WebDev Arena leaderboard to rank fourth. In benchmark tests, Gemini 4 Carbon outperforms Argon and beats GPT‑6 Astra, Claude Opus 5.5, and Claude Fable 5.1 by a few points.