Google has opened up its strongest AI model yet. Gemini 4 Carbon is now available to paying developers and to Google AI Ultra subscribers, after two weeks in the hands of a small group of cyber defenders.
If you have been waiting to try it, here is the short version: it costs $2 per million input tokens and $10 per million output tokens until 30 November 2026, then the price doubles. And most people still cannot use it in the Gemini app, because it is limited to Google’s most expensive plan.
What is Gemini 4 Carbon?
Gemini 4 Carbon is Google’s top model in the Gemini 4 family. Google announced it on 28 October 2026, as described in Google’s announcement, and made it generally available on 10 November 2026, in a post by Koray Kavukcuoglu, Google DeepMind’s senior vice president and chief AI architect.
Google calls Carbon its strongest model, built for work that tests the limits of a model’s intelligence: long, multi-step coding and debugging, open-ended scientific research, and cybersecurity. It uses more tokens than Gemini 4 Argon, the model below it, because it explores a problem more thoroughly, reasons more deeply, and runs longer checks on its own work.
Google says Carbon ranks first on three benchmarks it names: CursorBench 4.0, Perplexity’s WANDR, and Zapier’s AutomationBench, at more than twice lower cost per task than the next best outside model. Those are Google’s own claims, so treat them as the company’s word rather than an independent test.
Who can use Gemini 4 Carbon?
Access depends on who you are and what you pay for:
- Developers can use it through the paid tier of the Gemini API in Google AI Studio and Android Studio. There is no free tier. According to the pricing details, Gemini 4 Carbon is not available on the free tier.
- Businesses get it in Gemini for business, Gemini Enterprise, Google Antigravity, and CodeMender.
- Consumers need Google AI Ultra, Google’s most expensive plan. Ultra subscribers get Carbon in the Gemini app, in AI Mode in Search, and in Antigravity.
If you are on the free plan, Google AI Plus, or Google AI Pro, you cannot use Carbon in the Gemini app yet. Google has not said when, or if, that will change.
How much does Gemini 4 Carbon cost?
The pricing is simple, but the deadline matters. Until 30 November 2026, Carbon costs $2 per million input tokens and $10 per million output tokens, the same introductory price as Gemini 4 Argon. From 1 December 2026, the price doubles to $4 per million input tokens and $20 per million output tokens.
Thinking tokens, the tokens Carbon uses while it reasons through a problem, are billed as output tokens. Cached input rises from $0.10 to $0.20 per million tokens on 1 December as well.
That means November is the cheap month to test it. If you want to try Carbon on a real project, doing it before the price rise is the sensible move.
What Gemini 4 Carbon is best at
Google pitches Carbon for the hardest jobs, not everyday ones. Here is where it is meant to shine:
Long coding jobs. Carbon is built for sustained work across large codebases, such as tracking down a bug that crosses several parts of a system, or moving code between packages. Google says its own teams have seen Carbon keep working for days to get through dead ends on complex engineering problems.
Cybersecurity. This is the area Google talks about most. In simulated cyberattacks, Carbon blocked 99 percent of attack paths from an automated red-team agent, and its per-attempt breach rate fell to 0.8 percent, down from 3.1 percent on Argon. Google also says Carbon helped show that two bugs in popular open-source software could be exploited, and that security firm Wiz used it to find zero-day bugs at two large software companies.
Research and science. Carbon shows stronger reasoning across physics, maths, and life sciences, according to Google. It can pull together experimental data, test ideas, and work through multi-step problems in a subject area.
Knowledge work. For legal, finance, and business teams, Carbon combines visual understanding and reasoning with deep document search, to pull grounded answers out of large sets of dense documents and data.
Even so, Google itself still recommends Argon for most coding and office work. Carbon is for the longest, hardest jobs, and it burns more thinking tokens doing them.
What is the /swarm feature?
Alongside Carbon, Google is previewing a feature called /swarm in Antigravity, its agent-based coding tool. A single prompt turns into a coordinated team of agents that you can watch in real time.
Carbon can take the roles that need the deepest reasoning, such as planning the architecture, coordinating the work, and checking it, while lighter models handle routine tasks in separate workspaces. It is a preview for now, not a finished product.
Should you use Gemini 4 Carbon?
It depends on what you need it for. If you are a developer or a business with a hard, long-running problem, such as a large migration, a deep security review, or open-ended research, Carbon is now the strongest model Google sells, and November is the cheapest month to try it.
If you are an everyday Gemini user on the free plan or Google AI Plus or Pro, you cannot use it yet, and there is no date for when that changes. For most daily coding and writing work, Google’s own advice is to stick with Argon.
The bottom line is simple: Carbon is Google’s most capable model, it is finally open to paying users, and its low introductory price ends on 30 November 2026.
Frequently asked questions
When did Gemini 4 Carbon become available?
Google made it generally available on 10 November 2026, two weeks after it first went to a small group of cyber defenders through its Fairwind Program.
How much does Gemini 4 Carbon cost?
It costs $2 per million input tokens and $10 per million output tokens until 30 November 2026. From 1 December 2026, the price rises to $4 and $20 per million tokens.
Can I use Gemini 4 Carbon for free?
No. It is not available on the Gemini API free tier, and in the Gemini app it is limited to Google AI Ultra subscribers.
What is the difference between Gemini 4 Carbon and Gemini 4 Argon?
Carbon is built on Argon but uses more tokens to explore, reason, and check its work. Google says Argon is the better pick for most everyday coding and knowledge work, while Carbon is for the hardest, longest jobs.
