Google Unveils Gemini 4 Carbon, Its Next Frontier Model

Google DeepMind announced Gemini 4 Carbon, its next frontier model, which ranks first on version 5.0 of the Artificial Analysis Intelligence Index with a score of 59, Google said in a blog post. Google calls it its strongest model for cybersecurity and long horizon engineering tasks. Carbon scored 88.2% on SEC-bench Pro for vulnerability detection, 54.8% on ExploitGym for exploitation, and 71.0% on CWE-bench for patching when evaluated without guardrails.

Google says Carbon can work on coding tasks for days, including finding and patching bugs across large codebases and low level systems. In internal tests, it more than doubled the success rate for generating secure code compared with Gemini 4 Argon and reduced unsupported claims by about half.

Access begins with a subset of cyber defenders, followed by additional trusted partners through the Fairwind Program. Google is also participating in a voluntary United States government process for model testing before release and plans broader availability for developers, businesses, and consumers.

Introductory pricing will be $2 per million input tokens and $10 per million output tokens, with cached inputs discounted by 95%. After the introductory period, pricing will rise to $4 per million input tokens and $20 per million output tokens.