Gemini 4 Argon price and access: is it free and who gets it

Google has rolled out Gemini 4 Argon, its strongest model, to everyone who pays: Gemini app and AI Mode subscribers, plus developers through the API. There is no free tier for it at all. Here is what the model does, what it costs, and whether a short-video creator needs it — or should spend that $20 somewhere else.

What Gemini 4 Argon actually is

Argon is the top model in the Gemini 4 line. Its strength is not images or video but long reasoning work: reading code, digging through documents and spreadsheets, and acting as an agent that completes a chain of steps on its own.

Two things set it apart from a regular chatbot:

• it runs on a unified limit of up to 1 million tokens for input and output combined — the previous output ceiling was 64K tokens, so it can both read a lot and produce a very long result;
• it is built for agent-style work: not one reply, but a task broken into steps and carried to the end.

Argon does not generate video. That is a job for other Google models, so if you need footage or images, this is the wrong tool.

Where you can use it, and on which plan

Google also opened Argon to a set of security defenders through its Fairwind Program — a closed track for people protecting infrastructure.

What it costs

For a regular user, the price is the price of the subscription. Google's consumer plans as of October 2026:

For developers the price is per token, and there is a deadline attached:

So the current rate is introductory and doubles on December 1. If you run anything that calls the API — your own tool, a client project, an automation — budget for that now.

The free Gemini tier got smaller a week earlier

A week before the Argon launch, Google cut what free accounts get. Since October 9, 2026, free users are left with Flash-Lite only — the lightweight model for everyday summarizing and brainstorming. Flash moved behind AI Plus at $4.99, and Pro behind AI Pro at $19.99.

For most creators that is the bigger story: free Gemini is noticeably weaker than it was at the start of the month.

The benchmark numbers, and how to read them

Google compares Argon with rival flagships. These are the results the company published itself:

Two caveats. First, the vendor picks the benchmarks, and Argon loses some of them — command-line work, for one. Second, independent scoring is more modest: in the Artificial Analysis Intelligence Index, Argon sits at 53 against 58 for Claude Opus 5.5 and 53 for GPT-6 Astra, while in the Vals Index of professional finance, legal and coding work Argon leads at 68.9% — less than a point ahead of Opus 5.5. Figures reported by VentureBeat.

In plain terms: a strong model in the same class as its rivals, not a leap above them.

On the practical side, Google says Argon helped its own engineers compress a Chromium optimization effort from months into weeks. Trade press adds internal numbers — 35% more code generated and 33% fewer rollbacks, plus issues found that cut YouTube server load by 45%. None of that can be verified from the outside; it comes from the company.

What this means if you make short videos

Honest answer: not much directly. Argon is tuned for code, analysis and agent tasks, not for editing or shooting. Two uses are real, though.

Going through long footage

That 91.7% on long-video understanding means the model can find the right moment in an hour-long recording and explain what happens there. For a clipper that is the daily grind: watch a three-hour stream, pull the segments worth cutting. It is a model capability rather than a button in a product today, but the direction is clear.

Reading your own analytics

Exports from YouTube Studio or TikTok, dozens of videos compared, patterns pulled out of the numbers — "read a lot of boring data and draw conclusions" is exactly where these models are strong.

What to do about it:

1. Do not switch tools because of one launch. If your current setup works, Argon will not speed up your edit.
2. If you pay for AI, pay for one subscription, not three. The flagships are within a few points of each other.
3. Test on your own task. Take one real case — a stream breakdown, a script, an analytics export — and run it through whatever you already have.

Which AI for which job

Related reads: Is GPT-6 free in ChatGPT, Claude Opus 5.5 for clipping and UGC and the best AI video editors.

What is still unclear

• Whether Google will extend or replace the introductory API rate after November 30 — the increase is announced, nothing else is.
• Whether Argon's video-reading ability will show up as a product feature or stay an API capability.
• How request limits inside AI Pro and AI Ultra change — Google does not publish the exact numbers.

FAQ

Is Gemini 4 Argon free?

No. Free accounts have had Flash-Lite only since October 9, 2026. Argon starts at the AI Pro plan, $19.99 a month.

How much does Gemini 4 Argon cost for developers?

Through November 30, 2026: $2 per million input tokens and $10 per million output tokens, with cached input 95% cheaper. From December 1: $4 and $20.

Is Gemini 4 Argon better than ChatGPT or Claude?

On Google's own benchmarks it leads in coding, legal and agent tasks. Independent indexes show a much tighter race, and Claude Opus 5.5 is ahead on the Artificial Analysis index. For scripts and everyday writing you will barely notice a difference.

Can Gemini 4 Argon make videos?

No. It is a reasoning, coding and analysis model. Video generation belongs to other models.

Do I need AI Ultra?

Only if you want Argon inside Antigravity, Google's agent development environment. For the Gemini app and AI Mode in Search, AI Pro is enough.

While the model race goes on, the steadiest income is still work with a price on it. On Prime Oracles brands post tasks through "Create campaign"; a creator opens "Content rewards", hits "Take to work" and gets paid for the views their own video earns.