Google on October 28 announced Gemini 4 Carbon, describing it as its strongest model with state-of-the-art cybersecurity and long-horizon capabilities1,2. The model is an improvement on Gemini 4 Argon, which shipped just two weeks earlier, but like Argon it will follow a staged release that Google says could take a while.
Google said Gemini 4 Carbon reaches a new threshold in coding, agentic tasks, and scientific analysis. The cybersecurity claims are the most specific: Google said the model shows a "significant leap" in vulnerability detection, validation, and remediation, and is twice as likely to generate vulnerability-free code compared to Gemini 4 Argon. The Chrome team claimed Gemini 4 Carbon is 10 times better at generating high-quality patches compared to non-Google models.
On benchmarks, Google said Gemini 4 Carbon increased its WebDev Arena score by 89 points, now ranking fourth on that leaderboard. Separately, the model ranked first on Artificial Analysis' new Intelligence Index.
ANALYSIS The back-to-back Argon and Carbon releases mark a rapid cadence for Google's frontier line, though the staged-access pattern limits near-term external validation of the claimed gains.