Latest

LatestAI news

Google announces Gemini 4 Carbon, limited at first to cyber defenders

Google's blog, dated 28 October 2026, announces Gemini 4 Carbon as its next frontier model, which it calls its strongest model, with cybersecurity and long-horizon work as the focus. The post says Google is in the US government's voluntary process for pre-release model testing. Starting this week, Carbon is available to a subset of cyber defenders. Access then phases to further trusted partners through the Fairwind Program. Google says it is still gathering feedback and iterating on guardrails before Carbon is available to developers, businesses and consumers.

Source: Google (opens in a new tab), 28 Oct 2026

Google's blog, dated 28 October 2026, announces Gemini 4 Carbon as its next frontier model, which it calls its strongest model, with cybersecurity and long-horizon work as the focus. The post says Google is in the US government's voluntary process for pre-release model testing. Starting this week, Carbon is available to a subset of cyber defenders. Access then phases to further trusted partners through the Fairwind Program. Google says it is still gathering feedback and iterating on guardrails before Carbon is available to developers, businesses and consumers.

The post says Carbon will launch at the same introductory price as Gemini 4 Argon: $2 per million input tokens and $10 per million output tokens, with cached input tokens priced at 95% off the input token price. After the introductory period expires, the price is $4 per million input tokens and $20 per million output tokens. The post does not say when that introductory period ends, and it does not state a context window.

Google says Carbon ranks first on version 5.0 of the Artificial Analysis Intelligence Index at 59. It gives SEC-bench Pro at 88.2% for finding vulnerabilities, ExploitGym at 54.8% for exploitation, and CWE-bench at 71.0% for patching. It says those external benchmarks were evaluated without guardrails. On its own measures, Google says Carbon cut successful cheating on Terminal-Bench 4.0 to 0.0%, with cheat attempts at 0.4%, and that high-severity misalignment alerts in internal live coding fell by almost 80%. It also says Carbon's attack success rate on Gray Swan's Indirect Prompt Injection benchmark was 0.0%, against 0.7% for Argon.

What it means for business: Carbon is not generally available. The introductory prices are $2 and $10 per million input and output tokens, then $4 and $20, and the post does not date the change. A model Google is limiting to cyber defenders, and that it describes as strong at finding and exploiting vulnerabilities, is a reason to wait for the wider release notes before connecting it to business systems.

All items in AI news