Gemini 4 Carbon: Who Can Use It, and Argon's Dec 1 Price

Only a small group of cyber defenders picked by Google can use Gemini 4 Carbon right now. As of November 2026, Google has not given a date for anyone else. The Google model you can actually use today is Gemini 4 Argon. It's in the Gemini app for Google AI Pro ($19.99 a month) and Google AI Ultra subscribers, and in the paid Gemini API for developers.

If you build on the API, put one date in your calendar. Argon's introductory price ends November 30, 2026. From December 1, the price doubles to $4 per million input tokens and $20 per million output tokens.

The short version: Argon is something you can buy, and Carbon is something you have to be chosen for.

Who can use Gemini 4 Carbon, and is it in the Gemini app?

As of November 2026, Gemini 4 Carbon is available only to "a subset of cyber defenders," according to Google's Carbon launch post from October 28, 2026. It is not in the Gemini app. Google says it will "gradually phase access to additional trusted partners" through its Fairwind Program, a vetted access scheme for security teams.

So you can't switch Carbon on today in any of these places:

• The Gemini app or AI Mode in Search. Google hasn't named a consumer plan that includes Carbon. TechRepublic reports that "a paid Gemini subscription does not guarantee Carbon access."
• The Gemini API. Google says it is still gathering feedback and iterating on guardrails before releasing Carbon to developers. Guardrails are the safety limits on what a model will help with.
• Business products. These have the same status. Google says Carbon will come "as soon as possible," with no date.

Think of Carbon as a new aircraft still in test flights with a few hand-picked crews. It exists, and Google says it flies well, but you can't book a seat yet.

Which model can you actually use, by role?

As of November 2026, most readers can get Gemini 4 Argon today through a paid plan or the paid API. Only selected cyber defenders can get Gemini 4 Carbon, and even Fairwind membership doesn't promise it. This table combines Google's Argon post, Carbon post, Fairwind page and plan page. Find your row and stop guessing.

If your row says "No" under Carbon, that's not a temporary glitch on your account. It's the current policy.

What is the difference between Gemini 4 Argon and Carbon?

Gemini 4 Argon is the model you can use today. Gemini 4 Carbon is Google's newer model, built on Argon and announced two weeks after Argon launched, but access is still restricted. Google calls Carbon "our strongest model with state-of-the-art cybersecurity and long horizon capabilities." "Long horizon" means working through long, multi-step tasks without losing the thread.

Treat Carbon's scores as Google's own reporting. SEC-bench Pro is a test of finding software vulnerabilities, and Google says Carbon's score was measured without guardrails. TechRepublic also notes that benchmark results "do not guarantee comparable performance in production environments."

Argon's "1M unified input-output token limit" means a single API request's input and output share one budget of about a million tokens. Google calls this "industry-leading." In everyday terms, a very long contract or report plus a long answer can fit in one request.

Which Google AI plan includes Gemini 4 Argon?

As of November 2026, Google AI Pro ($19.99 a month) and Google AI Ultra ($99.99 or $199.99 a month) include Gemini 4 Argon in the Gemini app and AI Mode in Search. That's according to Google's Argon launch post and Google's subscriptions page. The Free plan lists the lighter Flash-Lite model instead.

Is Gemini 4 Argon free? Not on the Free plan, going by Google's plan page. For developers, Google's Argon materials point to the paid tier of the Gemini API.

Does the December 1 rise change my subscription? The rise is a per-token API price. The plans above are flat monthly fees, and Google's Argon post describes the change only in token terms.

How much will Gemini 4 Argon cost after November 30?

From December 1, 2026, the Gemini 4 Argon API price rises from $2 to $4 per million input tokens and from $10 to $20 per million output tokens. The introductory rates run through November 30, 2026, per Google's Argon post and Gemini 4 Argon developer guide. Google says the new prices apply to the publicly available Argon model.

A token is a small chunk of text, often part of a word. "Input" is everything you send the model, and "output" is what it writes back.

Here's a worked example. Picture a small support-reply tool that reads 20 million tokens of customer emails and help docs a month and writes 4 million tokens of replies.

Same work, double the bill. Output tokens cost five times as much as input, so if your app writes a lot (long reports, drafts, code), the output line is the one to watch.

One caveat applies to caching. Caching is when Google stores text you send repeatedly, like the same long instructions, and charges less for it. Google lists cached input at 95% off alongside the introductory price, so check the developer guide for the cached rate after November 30 before you put it in a forecast.

When is Gemini 4 Carbon coming, and what hasn't Google announced?

As of November 2026, Google has not announced a date for Gemini 4 Carbon in the Gemini app, the API, or business products. Its own heading says only "Coming soon to developers, businesses, and consumers." TechRepublic sums it up: the announcement "confirms broader availability is planned but provides no timetable."

What Google has announced about Carbon pricing:

• It will launch at the same introductory price as Argon: $2 per million input tokens and $10 per million output tokens, with cached input 95% off.
• After its introductory period, the price will be $4 per million input tokens and $20 per million output tokens.

What Google has not announced, so don't plan around any of it:

• A release date for Carbon in the Gemini app, AI Mode, or the Gemini API.
• Which consumer plans (Plus, Pro, or Ultra) will include Carbon.
• When Carbon's introductory price period ends. Don't assume it's November 30 like Argon's.
• Whether Fairwind members will all get Carbon. TechRepublic notes that admission "does not guarantee Carbon access."

TechRepublic's practical read is that most organisations "will have to rely on existing models rather than plan deployments around Carbon." For most readers, that means Argon. If you have a budget line labelled "Carbon, someday," you can quietly delete it.

How do I get into the Fairwind Program?

To request Fairwind access, you contact your Google Cloud account team, according to Google DeepMind's Fairwind page. The page says the program gives approved defenders "exclusive access to Gemini 4 Argon with cybersecurity defense capabilities." Google's Carbon post says Carbon access will be phased in gradually to additional trusted partners through Fairwind.

This isn't a sign-up form. Per the Fairwind page, member organisations agree to:

• User-level authentication, so every person using the model is individually identified.
• Phishing-resistant MFA (multi-factor sign-in that a fake login page can't trick) and access controls.
• Limit access to internal cybersecurity, incident response, or penetration testing teams, and track which employees use it and how.

Google also runs background checks on applying organisations "to verify security history and analyze their record of ethical operations." Academic labs that focus on defensive benchmarking are welcome to apply.

If you aren't eligible, you still have options. Google says you can protect your code with CodeMender using publicly available models, along with other Google AI Threat Defense products.

Should I switch to Argon before the price changes?

Switch to Gemini 4 Argon because it fits your work, not to beat the deadline. Google hasn't announced any way to keep the introductory price after November 30, 2026, so the new rate applies to the public Argon model from December 1 no matter when you started. What November does give you is a cheaper month to test.

If you're moving code to Argon, Google's developer guide flags a few things to check before you flip the switch:

• Model ID: gemini-4-argon.
• Thinking levels: low, medium, or high, with medium as the default. This controls how much the model reasons before answering.
• Sampling settings are ignored. Custom values for temperature, top_p and top_k (dials for how random the answers are) are ignored and return a deprecation warning. If your app relies on them, test the output.
• Longer outputs. Earlier Gemini models capped output (thinking plus response) at 64K tokens. Argon can generate past that.

App subscribers can relax, because you pay a flat monthly fee, not per token. Your only question is whether Pro's limits are enough or Ultra's extra headroom is worth it.

What should you do before December 1?

Before December 1, 2026, the one decision that can't wait is your Gemini 4 Argon API budget. Everything about Carbon can wait, because Google hasn't set a date. Here's your to-do by role:

• Everyday users: If you want Argon, Google AI Pro ($19.99 a month) is a plan Google lists as including it. Don't hold out for Carbon, because Google hasn't named a plan for it.
• Small business owners: Give staff who use Gemini daily a Pro seat now. If you pay for a tool built on the Gemini API, ask the vendor whether December 1 changes your bill.
• Developers: Log one week of real token use, multiply it out to a month, and price it at $4 / $20. Then test the gemini-4-argon model ID and your thinking level while November rates still apply.
• Security teams: If you qualify, contact your Cloud account team about Fairwind now, since Google is phasing access in gradually. If you don't, set up CodeMender with publicly available models instead of waiting.

Prices and plan limits here were checked on Google's own pages in November 2026; they change often.