Gemini 4 Argon: what the Nov 30 intro price gets you

Gemini AI, Google’s family of AI models and its chat app, now has a new top model you can buy: Gemini 4 Argon. For developers paying per token through the API (the way software connects to a model), Google’s introductory price of $2 per 1M input tokens and $10 per 1M output tokens ends on Nov 30, 2026, and the price then doubles to $4 and $20. Google’s posts give that date only for per-token prices, not for Gemini app plans.

This is an analysis of Google’s own documents and one independent leaderboard; we did not run our own tests of Argon. Prices below were checked on 2026-11-03.

A token is a chunk of text a model reads or writes (see What is AI? for the basics). An introductory price is a temporary list price that Google has said will rise on a set date. Thinking tokens are the tokens a model spends reasoning before it answers, and Google bills them as output.

The short version

If you pay per token through the API: Google’s Nov 30 date is the one that matters. The same workload costs twice as much per token from Dec 1, so price it before you commit.
If you use the Gemini app: Google gives the Nov 30 date only for per-token API prices. Google lists Argon on Google AI Pro and Google AI Ultra, and the plans page we read shows monthly prices with no Nov 30 date.
If you want the strongest Google model: that is not Argon. Google says Gemini 4 Carbon is its strongest model, but access is limited for now.

What Gemini 4 Argon is

Google says Argon is its flagship reasoning model, built for complex software engineering, math and science problems, multi-step agents and long outputs of up to 1M tokens (Gemini API pricing page). It launched on Oct 13, 2026, per Google’s launch post. Google’s model page lists these specs:

Model ID: gemini-4-argon.
Inputs: text, image, video, audio and PDF. Output: text only. It does not generate audio or images.
Context window: 1,048,576 tokens (the most text the model can hold at once), shared by input, thinking and output. A normal response can reach 65,536 tokens, and Google’s long-decode mode (a setting for very long single responses) goes up to 1,048,576 tokens minus your input.
Thinking levels: low, medium and high settings for how long the model reasons before it answers. Google’s what’s-new page says medium is the default.
Tools: function calling (the model asks your software to run a task), structured outputs (answers in a fixed format), code execution, computer use (operating a screen), file search, and grounding (checking answers against Google Search or Google Maps).
Caching: implicit only, meaning Google reuses repeated prompt text of at least 4,096 tokens automatically and charges less for it. Google says explicit caching (where you create a cache yourself) isn’t supported, and that the temperature, top_p and top_k settings (controls for how random the output is) are ignored.

Where Argon fits in Gemini AI

Argon is one model in a larger Gemini lineup. Google’s Gemini plans page lists “Access to Gemini 4 Argon” on Google AI Pro, and “Highest access to Gemini 4 Argon” on Google AI Ultra. Google AI Plus lists its Flash model and no Argon line, and a banner on the page says to choose AI Pro or AI Ultra to get Argon.

Google’s launch post says Argon is also available in the Gemini app and AI Mode in Search for those subscribers. The plans page adds that Argon in AI Mode is available in select countries.

Above Argon sits Gemini 4 Carbon. In its Oct 28 post, Google calls Carbon “our strongest model” and says it is starting with a subset of cyber defenders through its Fairwind Program, with wider access “as soon as possible”. So as of 2026-11-03, Argon is the top Gemini model you can buy.

For more on Carbon, see our news post and this week’s roundup. To compare the Gemini app with other assistants, see ChatGPT vs Claude vs Gemini vs Grok: AI assistants compared.

Gemini 4 Argon pricing (checked Nov 3, 2026)

These are Google’s Standard paid-tier list prices per 1 million tokens, in US dollars, from the Gemini API pricing page (checked 2026-11-03):

	Through Nov 30, 2026	From Dec 1, 2026
Input	$2.00	$4.00
Output (including thinking tokens)	$10.00	$20.00
Cached input (implicit caching only)	$0.10	$0.20

What changes on Dec 1: input and output prices double, per Google. Google’s launch post says the end date applies to the Argon model intended for public access.
Who it affects: API users who pay per token. Gemini app plans are priced per month: Google’s plans page shows Google AI Pro at $19.99/month and Google AI Ultra from $99.99/month (US page, checked 2026-11-03). Google’s posts state the Nov 30 date for per-token prices only.
Thinking tokens cost money: Google says every token the model processes is billed, including thinking tokens, and output pricing includes them. A job that thinks at length can cost more than the headline per-token price suggests.
Other tiers: Google’s page also lists Batch, Flex and Priority prices. We don’t print them here, so see the pricing page for those numbers.
Google Cloud: on Google Cloud, a footnote on the pricing page says the through-Nov 30 price is “Promotional pricing provided through 50% credits back on net spend on select models within a given period.”
No free API tier: Google’s page says Argon isn’t available on the Gemini API’s Free tier.

How Argon ranks

Google says Argon ranks #1 on the Vals Index, which measures work in finance, coding, legal and tax (Google’s launch post). We haven’t checked that claim against the index.

On the independent Artificial Analysis leaderboard, checked 2026-11-03, Gemini 4 Argon (high) scores 50 on the Artificial Analysis Intelligence Index. The same table shows Gemini 4 Carbon (high) at 59, Claude Opus 5.5 (max with fallback) at 58 and GPT-6 Astra (max) at 57, and GPT-6.1 Sol (max) at 55 and Claude Sonnet 5.5 (max with fallback) at 54 also score above Argon. So on that index, Argon is not at the top.

Each score is for the reasoning setting shown in brackets, such as “high” or “max”, and the table changes often, so treat the numbers as a dated snapshot, not a verdict. A higher score from a model with limited access doesn’t help most buyers either.

Who should pay now, and who should wait

Developers on a deadline: if you need Argon’s long outputs or coding and agent features now, the intro price is lower per token than what Google lists from Dec 1. We haven’t tested Argon, so try a small batch first.
Teams with large volumes: price your workload at $4/$20 before you commit. The doubling hits hardest when thinking tokens and long outputs are a big share of the bill.
Casual users: the Gemini app plans are the simpler route, and Google gives the Nov 30 date only for per-token API prices. If you want a no-cost way to try Gemini, see our best free AI tools.
Anyone waiting for Carbon: Google gives no date beyond “as soon as possible”, and says Carbon launches at the same introductory price as Argon.

Strengths and limits

Strengths, as Google documents them

A 1M-token window shared by input and output, with long outputs up to the same limit.
Inputs across text, image, video, audio and PDF, plus function calling, code execution and computer use.
A cached-input price of $0.10 per 1M tokens through Nov 30.

Limits

No Gemini API free tier.
Text-only output.
Explicit caching isn’t supported, and sampling settings such as temperature are ignored.
Token prices double on Dec 1.
Not Google’s strongest model, and not at the top of the Artificial Analysis index.

How we checked this

On 2026-11-03 we read Google’s pricing page, model page, what’s-new page, launch post, Carbon post, Google Cloud pricing page and Gemini plans page. We read the Artificial Analysis leaderboard on the same day. We did not run Argon ourselves, so we make no claim about its speed or quality.

Every statement about what Argon does is Google’s, and prices come only from Google’s pages. Pricing pages change, so check the date next to each number.

FAQ

How much does Gemini 4 Argon cost?

Google lists $2 per 1M input tokens and $10 per 1M output tokens through Nov 30, 2026, and $4 and $20 from Dec 1 (checked 2026-11-03). Output includes thinking tokens.

When does the Gemini 4 Argon introductory price end?

Google says Nov 30, 2026. From Dec 1, the $4 input and $20 output prices apply.

Is Gemini 4 Argon on the free API tier?

No. Google’s pricing page says Argon isn’t available on the Free tier. For free plans, see our best free AI tools.

Is Gemini 4 Argon in the Gemini app?

Google says yes, for Google AI Pro and Google AI Ultra subscribers, and in AI Mode in Search in select countries.