[DEVELOPING] Google unveils Gemini 4 Carbon, cyber-focused frontier model, first only to vetted defenders
Google DeepMind announced Gemini 4 Carbon, which ranks first on the Artificial Analysis Intelligence Index v5.0 with a score of 59; VentureBeat reports Claude Opus 5.5 at 58 and GPT-6 Astra at 57. Google reports 88.2% on SEC-bench Pro and 54.8% on ExploitGym. Access is limited to trusted cyber defenders in the Fairwind program, with no general-availability date. Introductory pricing is $2/$10 per million input/output tokens, rising later to $4/$20. A 'defense-first' release of a model with exploitation capability is relevant to the AI-driven cyber-spend theme. To watch: the timing of wider API access and whether a technical report is published.