Google DeepMind announces Gemini 4 Carbon with cybersecurity and long-horizon capabilities

What happened

Google DeepMind announced Gemini 4 Carbon, reporting a score of 59 on Artificial Analysis Intelligence Index v5.0 and cybersecurity benchmark results of 88.2% on SEC-bench Pro, 54.8% on ExploitGym, and 71.0% on CWE-bench, evaluated without guardrails. Access begins this week for a subset of cyber defenders and will expand through the Fairwind Program before broader availability. Introductory pricing will match Argon at $2 per million input tokens and $10 per million output tokens, with cached input tokens discounted 95%.

Read next

• More changes in today’s briefingGoogle DeepMind announces Genie 4 for real-time world simulation with native audio

Reported by 6 sources

The story so far

Google DeepMind announced Gemini 4 Carbon with initial access limited to selected cyber defenders, while Artificial Analysis ranked it first on its Intelligence Index v5; the model is not yet publicly available. Google said access would begin this week and gradually expand to trusted partners through its Fairwind Program as it gathers feedback and refines guardrails before broader availability to developers, businesses, and consumers. Artificial Analysis gave Carbon a score of 59, including the highest score in four of nine evaluations. In its coding-agent comparison using Antigravity CLI, Carbon scored 59, behind Claude Opus 5.5 at 62. Google reported cybersecurity scores of 88.2% on SEC-bench Pro, 54.8% on ExploitGym, and 71.0% on CWE-bench, all evaluated without guardrails. Sundar Pichai said Carbon significantly improves on Argon in long-horizon tasks and has strengthened frontier safeguards. The restricted rollout limits who can try the model despite its published benchmark results. Nathan Lambert criticized Google for announcing scores for a model he could not publicly access.