Google unveils Gemini 4 Carbon AI model with record-breaking cybersecurity capabilities

Google unveils Gemini 4 Carbon AI model with record-breaking cybersecurity capabilities Google has officially announced Gemini 4 Carbon, its latest frontier AI model following the release of Gemini 4 Argon. Marking a major leap forward for the Gemini ecosystem, Carbon lands at the #1 spot on the updated v5.0 Artificial Analysis Intelligence Index with a score of 59, showcasing massive gains in complex coding, multi-step agentic workflows, and an unprecedented focus on cybersecurity defense. Carbon is engineered to span the entire cybersecurity defense lifecycle—from vulnerability detection and risk validation to root-cause remediation. It notched record-breaking benchmark scores. Vulnerability discovery andrisk or exploitation validation: 88.2% on SEC-bench Pro, autonomously finding complex flaws in real-world software. 54.8% on ExploitGym, assessing whether vulnerabilities are actively exploitable to help teams prioritize patches. Root-cause patching: 71.0% on CWE-bench, translating validated findings directly into secure code patches. Google's Chrome team noted that Carbon generated high-quality security patches more than 10 times better than competing frontier models. Advanced agentic & coding proficiency: Designed to manage complex, long-running engineering chores over multi-day spans, Carbon significantly minimizes hallucinations, fabrications, and task overreach/under-service. Enhanced safety and guardrails: Google tightened safety mechanisms alongside the model's increased capability. Internal testing demonstrated that Carbon brought successful cheating attempts on Terminal-Bench down to effectively 0%, while maintaining stricter adherence to scope and project constraints. Access and gated rollout: Unlike broader consumer rollouts, initial access to Carbon is strictly gated. It is rolling out to a curated subset of cyber defenders and trusted partners through Google's Fairwind Program, with wider availability planned later. Introductory pricing: When accessible via APIs, Google has slated introductory pricing of $2 per million input tokens and $10 per million output tokens (with cached input tokens receiving a 95% discount), matching the pricing tier of its Argon counterpart.