Latest / Frontier models and safety

Model releaseFrontierMediaUSConfirmed

Google announces Gemini 4 Carbon, with limited access for cyber defenders

See how this story connects

Google announced Gemini 4 Carbon on 28 October 2026. Google says the frontier model ranks first on Artificial Analysis Intelligence Index v5.0 with a score of 59. Google reports 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for exploitation and 71.0% on CWE-bench for patching, evaluated without guardrails. From this week Google is making Carbon available to a subset of cyber defenders, and will phase access to additional trusted partners through its Fairwind Program. Developers, businesses and consumers get access later, once Google has iterated on guardrails.

Why it matters

Google says Carbon more than doubles the success rate of secure code generation over Gemini 4 Argon on internal benchmarks, and its Chrome team reported patches of more than 10 times better quality than another lab’s frontier model. The introductory price is $2 per million input tokens and $10 per million output tokens, rising to $4 and $20 after the introductory period.

When
28 Oct 2026
ID
WTA-N-2026-0477
Who
Google, Google DeepMind, Wiz, Chrome; Koray Kavukcuoglu

Summary written with AI by wheresthe.ai, from Google. Checked against the source on 28 Oct 2026.

Source copy: saved on 28 Oct 2026 to check this story and not republished, because the text belongs to its publisher. Also held by the Internet Archive.

Next stop on this Storyline: Elsewhere: InternationalWiz links a month of AI developer platform breaches to theft of AI keys

Cite this story

WTA-N-2026-0477: “Google announces Gemini 4 Carbon, with limited access for cyber defenders.” wheresthe.ai, 28 Oct 2026. https://wheresthe.ai/d/us-google-announces-gemini-4-carbon-with-limited-access-for-2026-10/ (source: Google).

Spotted an error? Email [email protected].