Google DeepMind Launches Gemini 4 Carbon With Staged Cybersecurity Rollout

Google DeepMind released Gemini 4 Carbon on October 28, 2026, ranking first on the Artificial Analysis Intelligence Index v5.0 with a score of 59. The model posts state-of-the-art cybersecurity benchmark results: 88.2% on SEC-bench Pro, 54.8% on ExploitGym, and 71.0% on CWE-bench. On the GraphWalks long-context benchmark (256K to 1M tokens), Carbon scores 96.2%, compared to Argon's 84.2%, GPT-6 Astra's 71.8%, and Claude Opus 5.5's 66.8%. Because of its exploit capabilities, access is being staged: first to a subset of cyber defenders, then to trusted partners via the Fairwind Program, and later to developers, businesses, and consumers. Introductory pricing is $2 per million input tokens and $10 per million output tokens, with prices set to double to $4 and $20 later. The model reduces successful cheating on Terminal-Bench 4.0 to 0.0% and cuts high-severity misalignment alerts by nearly 80% in live coding environments. Carbon achieves a 0.0% attack success rate on the Gray Swan Indirect Prompt Injection Benchmark.