What Is Gemini 4 Argon, and Can You Use It Yet? (October 2026)

TL;DR: Gemini 4 Argon is Google's most capable AI model, announced on September 30, 2026. You almost certainly can't use it yet: Google released it only to vetted cybersecurity defenders through its Fairwind Program, because it is good enough at finding and fixing software flaws that it could help attackers too. Google says paid API customers and Google AI Ultra subscribers come next, then everyone else, but it has not given a date. Until then, nothing changes in your Gemini app.

If you saw "Google's most powerful model yet" in the headlines and opened Gemini to try it, you didn't miss a setting. Argon isn't there. This post covers what Argon is, why Google is holding it back, who gets it first, and what that means for you if you use Gemini at work.

What is Gemini 4 Argon?

Argon is the first model in Google's Gemini 4 generation. Google's announcement, written by Koray Kavukcuoglu (senior vice president of Google DeepMind and Google's chief AI architect), describes it as built for "complex workflows across real-world software engineering, enterprise knowledge work like legal and finance, and cybersecurity defense."

The headline claims from Google:

• Long output. Argon can produce up to 1 million tokens in a single response, up from 64,000 in Google's earlier models. In practice, that means very long documents, reports, or code changes without stopping partway through.
• Coding and long tasks. Google reports 77.9% on DeepSWE v1.1, a test of real-world software engineering, and first place on AutomationBench.
• Professional work. Google says Argon leads the Vals Index, which covers finance, legal, tax, and coding tasks.
• Video and charts. 91.7% on LVBench, a long-video understanding test.
• Security. Tied for first on CWE-bench, which measures whether a model can fix software vulnerabilities, and leading results on a prompt-injection robustness test.

According to Engadget, Argon also takes the place of the Gemini 3.5 Pro model Google had been expected to ship.

Keep in mind that these are Google's own numbers on a model that almost no one outside Google and its partners can test yet.

Why can't you use it yet?

Google says Argon "can autonomously find, validate, and patch critical software vulnerabilities." That is useful for defenders, but the same skill could make cyberattacks easier. So Google is starting with a small group.

Argon is "rolling out to a set of trusted cyber defenders through our Fairwind Program," Google wrote, adding that "safely releasing frontier capabilities at this level requires a phased approach." Google says it is taking part in the U.S. government's voluntary process for pre-release model access and collecting feedback from early testers before a wider release. AFP reported that the launch came a day after Google and other AI companies signed a voluntary White House pledge on AI risks.

This isn't new in the industry. Anthropic has kept its most capable cyber-focused model, Claude Mythos, in a restricted program for vetted organizations while releasing a safeguarded version to the public. Google is now following a similar pattern.

Who gets Gemini 4 Argon next?

Google has named the order but not the timing:

1. Now: vetted cyber defenders in the Fairwind Program.
2. Next: paid Gemini API customers and Google AI Ultra subscribers.
3. Later: developers, enterprises, and consumers more broadly, which Google says will happen "as soon as possible."

As of October 1, 2026, Google's subscriptions page does not list Argon on any plan. Google AI Ultra, the plan named as first in line, starts at $99.99 per month in the US (with a $199.99 tier for higher limits). Google has not said when Google AI Pro ($19.99), Google AI Plus ($4.99), or free users will get Argon, and it has not published plan-level access yet. We'll update this post when it does.

How much will it cost?

For developers, Google lists an introductory API price of $2 per million input tokens and $10 per million output tokens, with cached input at a 95% discount. After the introductory period, the standard price is $4 and $20. Google has not said when the introductory period ends.

For most professionals, the API price matters less than the plan question above. In the Gemini app, you'll get Argon through your Google AI plan, not as a separate purchase, once Google opens it up.

Should you change anything now?

Not really. Here's a practical way to think about it:

• Don't upgrade to Google AI Ultra just for Argon. Ultra subscribers are in line, but there is no date. Paying $99.99 a month for a model you can't use yet doesn't make sense.
• Don't switch assistants based on launch benchmarks. Google's numbers look strong, but they come from Google. Independent tests, and your own trial on real work, are what count. If you're comparing tools, our ChatGPT vs Claude vs Gemini guide covers the models you can use today.
• If you work in IT or security, ask your vendors whether they are part of the Fairwind Program. That is the only route to Argon today.
• Check your Gemini model picker from time to time. When Argon reaches your plan, it will appear in the model dropdown in the prompt bar.

The bigger story is the pattern. Two of the three major AI labs now hold back their most capable models from the public at first because of security risk. Expect "available to vetted partners first" to become a normal part of big model launches.

Sources

• Google, Gemini 4 Argon: our next era of frontier intelligence (Koray Kavukcuoglu, September 30, 2026)
• Google, Google AI plans and pricing (checked October 1, 2026)
• TechCrunch, Google releases Gemini 4 Argon, called its most powerful model yet (September 30, 2026)
• Engadget, Google's first Gemini 4 model is 'Argon' (October 2026)
• The Guardian, Google rolls out new Gemini AI model but restricts access over safety concerns (October 1, 2026)
• AFP via Arab News, Google restricts access to new AI model over safety concerns (October 1, 2026)