Gemini 4 Carbon is the best AI you can't use
Google's top-ranked model sits behind a velvet rope with a price tag on it.
Gemini 4 Carbon is "our strongest model" by Google's own description, and it took #1 on an independent ranking. It scored 88.2% at finding real vulnerabilities. Almost nobody can use it.
Access starts for "a subset of cyber defenders" and then "trusted partners". Google cites pre-release testing with the US government and a "deliberate approach to rollout". Fair enough. Everyone else gets it "as soon as possible"; the floor plaque says the same.
It's a showroom model: spotlight, velvet rope, price tag taped to the pedestal. Input runs $2 per million tokens, output $10. Google promises no timeline for a wide release. The price is public. The release date is not.
Sources: Google (Oct 28, 2026); Ars Technica (Oct 28, 2026); Reuters (Oct 28, 2026).
- Google, "Introducing Gemini 4 Carbon" (Oct 28, 2026): the announcement itself; the strongest-model claim, the #1 rank, the 88.2% vulnerability score, the rollout tiers and the $2 / $10 price. blog.google
- Ars Technica, "Google announces Gemini 4 Carbon, its latest frontier AI model that you can't use" (Oct 28, 2026): the headline, and Google making no promises on the timeline for a wide release. arstechnica.com
- Reuters, "Google launches Gemini 4 Carbon AI model with cybersecurity focus" (Oct 28, 2026): independent confirmation of the launch. reuters.com
Current as of Oct 29, 2026, one day after the announcement. The image is an AI-generated illustration. Not affiliated with anyone.