What happened
Google introduced Gemini 4 Carbon on October 28, 2026, two weeks after Gemini 4 Argon, in a post by Koray Kavukcuoglu of Google DeepMind. Google calls it its strongest model and highlights long-horizon engineering and cybersecurity defense.
Google cites a score of 59 on version 5.0 of the Artificial Analysis Intelligence Index, first in that ranking, and state-of-the-art results on SEC-bench Pro (88.2%), ExploitGym (54.8%) and CWE-bench (71.0%). It also reports a WebDev Arena rating of 1768, up 89 points.
Who gets it and what it will cost
This week Carbon goes to a subset of cyber defenders, with more trusted partners added gradually through the Fairwind Program. Google says it is part of the U.S. government's voluntary pre-release testing and will open Carbon to developers, businesses and consumers once it has iterated on guardrails, without naming a date.
Pricing is announced but not live: the same introductory $2 per million input tokens and $10 per million output tokens as Argon, cached input at 95% off, and $4 and $20 once the introductory period ends. On October 30 the Gemini API documentation had no Carbon model ID.
What it means for you
Google's safety claims are specific: no successful cheating on Terminal-Bench 4.0, roughly half the unsupported claims of Argon, 26% less overreach and a 0.0% attack success rate on Gray Swan's indirect prompt injection test. All come from Google's internal or chosen evaluations.
If you build on Gemini, keep working with Argon, whose introductory API price runs through November 30, 2026, and save a few real tasks to rerun on Carbon when it opens. Security teams interested in early access should look at the Fairwind Program.
Go to the source
Sources & further reading
Based on the official documentation below. Features and access can change; check the provider before making a decision.
Was this page useful?
Help us make our coverage more useful. This is feedback on the page, not a rating of the tool.
Go a little further