Google announced Gemini 4 Carbon on October 28. It described the AI model as focused on cybersecurity and long-horizon engineering tasks.

Starting this week, Google plans to give some cyber defenders access, then gradually extend it to other partners through its Fairwind Program.

Google said Carbon ranked first in version 5.0 of the Artificial Analysis Intelligence Index.

The company also said Carbon outperformed Gemini 4 Argon in secure code generation in internal benchmarks.

Google is gathering feedback and refining guardrails before a wider release to developers, businesses, and consumers.

Introductory pricing will be US$2 per million input tokens and US$10 per million output tokens.

The announcement did not name any early-access partners or specify eligibility criteria for cyber defenders.

Separately, Google DeepMind introduced a Frontier Safety Framework in May 2024 and later updated it.

🔗 Source: Google

Recent Google developments