Google has made Gemini 4 Carbon — its strongest model — generally available to developers, enterprises, scientists and consumers, ending the defenders-only rollout that began two weeks ago. Why it matters: the model built for cybersecurity and long, multi-step agent work is now something any paying developer can call, and it arrives with a preview feature that turns one prompt into a coordinated team of agents.
November 11, 2026
What happened
When Google introduced Carbon on October 28, access was limited to vetted cybersecurity defenders through its Fairwind Program. In a November 10 announcement, Google opened the gates: Carbon is now available through the Gemini API in Google AI Studio and through Gemini for business for all paying developers and enterprises. It is also rolling out to Google AI Ultra subscribers in Google Antigravity, the Gemini app and AI Mode in Search.
The price has not changed with the wider release. Carbon carries the same introductory API pricing as Gemini 4 Argon: $2 per million input tokens and $10 per million output tokens.
The new piece: /swarm in Antigravity
Alongside the wider rollout, Google is previewing a feature called /swarm inside Google Antigravity, its agent-focused development tool. A single prompt becomes a coordinated team of agents that you can track in real time. Developers can assign Carbon to the roles that demand the deepest reasoning — architectural design, coordination and auditing — while faster, lighter models work in parallel isolated workspaces, with adversarial verifiers testing every change.
Google is positioning Carbon for jobs that test a model’s limits: complex multi-step coding and debugging, open-ended scientific research, and end-to-end cybersecurity work. It is explicit that Carbon uses more tokens than Argon to explore and verify, and that Argon remains its recommended choice for the majority of software engineering and knowledge work.
What Google claims on performance
Google says Carbon ranks first on CursorBench 4.0, Perplexity’s WANDR and Zapier’s AutomationBench, at more than twice the cost efficiency per task of the next best external model. On a new Multimodal Knowledge Agent benchmark built with Artificial Analysis — designed to test agents working over dense files with charts, images and tables — Carbon ranks second while costing more than six times less per task than the top model. These are Google’s reported results, not independent hands-on testing by us.
Google also says it has expanded its free scientist access program: participants who received Argon in Antigravity can now use Carbon as well, free for six months, with applications still open on a rolling basis.
Why it matters
The two-week gap between “defenders only” and “everyone paying” tells you where Google thinks the risk sits: Carbon’s cyber capabilities are the headline and the hazard at the same time, and Google says its safeguards blocked 99% of attack trajectories in simulated testing. For everyone else, the practical news is simpler. The strongest Gemini model is now a normal line item at Argon’s introductory price — and /swarm is the clearest sign yet that the frontier labs expect single agents to give way to supervised teams of them.
Frequently asked questions
How can I try Gemini 4 Carbon?
Through the paid Gemini API in Google AI Studio or Android Studio, through Gemini for business, or — if you subscribe to Google AI Ultra — in the Gemini app, AI Mode in Search and Google Antigravity.
What does Gemini 4 Carbon cost?
Google lists introductory API pricing of $2 per million input tokens and $10 per million output tokens, the same as Gemini 4 Argon. Google calls the pricing introductory, so expect it to change.
Should I switch from Argon to Carbon?
Google itself says no for most work: Argon remains its recommended model for the majority of coding and knowledge tasks. Carbon is aimed at the hardest, longest jobs — and it uses more tokens getting there.
Sources: Google (blog.google), Android Headlines, The Neuron.