Google Announces Gemini 4 Carbon, Its Next Frontier Model, with Cyber Defenders Getting First Access
Google announced Gemini 4 Carbon, its next frontier model following Argon's launch two weeks ago, topping cybersecurity benchmarks and ranking first on the Artificial Analysis Intelligence Index v5.0; cyber defenders will be the first to receive access this week, with broader availability coming soon.
Google announced Gemini 4 Carbon, positioning it as its next frontier model and the direct successor to Gemini 4 Argon, which launched just two weeks earlier. The company is leading with cybersecurity as Carbon's flagship capability: the model scored 88.2 per cent on SEC-bench Pro, a test of AI's ability to identify software vulnerabilities; 54.8 per cent on ExploitGym, which measures exploitation ability; and 71 per cent on CWE-bench, which tests patch generation. Carbon also ranked first overall on the Artificial Analysis Intelligence Index v5.0 with a score of 59, and placed fourth on WebDev Arena with an 89-point Elo improvement. Beyond security, Google highlighted significant gains in coding and agentic tasks, as well as improvements in full-stack web development and multimodal understanding. On safety, Carbon cuts unsupported claims by approximately twice the rate of Argon, reduces overreach by 26 per cent, and decreases under-service by 29 per cent. Google attributed the fast release to its researchers and models working together to advance capabilities more quickly. Introductory pricing starts at $2 per million input tokens and $10 per million output tokens, rising to $4 and $20 per million at standard rates, with cached input tokens at 95 per cent off the input price. Cyber defenders will gain access first this week, with the model described as coming soon to developers, businesses, and consumers.



