Gemini 4 Argon Doubles in Price on 1 December

Gemini 4 Argon's introductory API pricing ends on 30 November 2026. From 1 December the rate doubles: input from $2.00 to $4.00 per million tokens, output from $10.00 to $20.00, and cached input from $0.10 to $0.20. Argon has no free tier on the Gemini API at any price, so every request is on a paid key. The same doubling applies to the Batch and Flex tiers.¹

The headline is a 100 percent increase with a date on it, which is the sort of thing that gets forwarded around an engineering team. The more useful reading is that most of it is avoidable, and not by acting before the deadline.

The tiers, measured

Prices measured on 27 October 2026. Vendor plan terms, allowances and prices change without notice, so treat every figure on this page as a dated snapshot rather than a permanent fact. Where a number here differs from the vendor's own pricing page, the vendor's page is correct and the figure above is simply older than it.

Read the Standard and Batch rows together and the deadline changes shape. Batch and Flex cost exactly half of Standard, permanently, because they trade latency for price. That means a batch workload paying $4.00 / $20.00 after 1 December is paying what it would have paid on Standard before it, and a team that moves work from Standard to Batch absorbs the entire increase.

Cache pricing moves the same way, $0.10 to $0.20 per million cached tokens, which is still an order of magnitude below the input rate. Caching is implicit only on Argon, so there is no explicit cache to manage, and no way to pin a prefix deliberately.

The deadline only bites interactive work

Anything that can sit in a queue until Google has spare capacity should already be on Batch. Anything that cannot wait is on Standard, and that is the traffic the doubling actually reaches: interactive agents, autocomplete-style features, and any pipeline where a user is watching a spinner.

So the useful question before 1 December is not how to lock in the price. It is which of your Argon calls genuinely need to return in seconds. In most systems that is a smaller fraction than the bill suggests, and the difference between Standard and Batch is a configuration change rather than an architectural one.

Priority sits at $3.60 / $18.00, which is 1.8 times Standard, for guaranteed faster access under load. That is a real product and a real premium: it costs more than the post-December Standard rate does today.

The Contributor tier is half price for your data

One row in that table is not like the others. Contributor is priced at $1.00 / $5.00 per million tokens, the same as Batch, and the pricing page marks it plainly: content is used to improve Google's products.

That is the trade stated honestly rather than buried in a policy page, and it deserves to be read as a business decision rather than a discount. Every other Argon tier carries a note that content is not used to improve the products. Contributor is the exception, and the exception costs half.

For a side project or an open dataset, that may be a reasonable exchange. For anything touching client material, internal code or regulated data, it is not a saving at all.

What Argon is, and who can reach it

Argon is Google's flagship reasoning model, positioned for complex software engineering, mathematics and science, autonomous multi-step agents, and long-form generation up to a million output tokens. It reached general availability on 13 October 2026, after two weeks limited to vetted cyber defenders in the Fairwind Program.

It shares the same one-million-token context across input and output that the rest of the Gemini 3-and-later line does, and thinking tokens are billed as output. A long reasoning run is therefore charged at the output rate for everything it thinks as well as everything it says.

Reachability splits cleanly. On the API it is paid only, with no free tier. In the consumer products it sits behind Google AI Pro at $19.99 and Ultra, and it is absent from both the free plan and AI Plus at $4.99. That is the same access split that reshaped the free Gemini app in October, arriving in the API as well.

Grounding has its own free allowance

Grounding with Google Search carries 5,000 free search requests per month, shared across all Gemini 3 and newer models, then $14 per 1,000 requests. Grounding with Maps has the same shape at 5,000 requests per month.

Those allowances are shared across the model family rather than granted per model, which matters if you run Argon alongside Flash: the two draw from one pool. A customer request can trigger more than one search query, and Google charges for each query rather than each request.

That is the one place where a free allowance exists in a product that otherwise has none, and it is worth knowing precisely because the table above it says Not available five times.

Gemini 3.8 Flash has its own, later deadline

Flash is the tier most people actually run at volume, and its introduction price runs longer: $0.75 input and $3.75 output through 31 December 2026, then $1.50 and $7.50 from 1 January 2027. It is also genuinely free of charge on the free tier, unlike Argon.

So there are two separate price steps a month apart. Argon doubles on 1 December and Flash doubles on 1 January. A team using both should not treat them as one event, and should not assume a single budget review in December covers the year.

What to actually do before 1 December

Move queue-tolerant traffic to Batch. The saving is 50 percent and it is permanent, so it survives the deadline and every deadline after it.

Check what is running on Priority. At 1.8 times Standard, a feature that used Priority defensively is already the most expensive thing on the bill and will double with everything else.

Decide about Contributor deliberately. It is the only way to pay less than Batch and the price is your training data, on the record, in the pricing table.

And if you are choosing between this and the consumer side of the same decision, the arithmetic is different: the free Gemini app lost Flash and Pro in October and Argon was never part of it, while ChatGPT's Pro 200 allowance drops on 29 October at an unchanged price.

Frequently asked questions

When does Gemini 4 Argon's launch price end?
30 November 2026. From 1 December the standard rate doubles to $4.00 input and $20.00 output per million tokens.

Is there a free tier for Gemini 4 Argon?
No. The Gemini API pricing page marks the free tier as not available for every Argon tier, including Batch and Flex.

Does batching avoid the price increase?
It absorbs it. Batch and Flex cost half of Standard, so the post-December Batch rate equals the current Standard rate.

What is the Contributor tier?
Argon at half the Standard rate, on the condition that content is used to improve Google's products. Every other tier states that content is not used.

Do thinking tokens cost extra?
They are billed at the output rate. Argon's output price explicitly includes thinking tokens.

Does the same deadline apply to Gemini Flash?
No. Flash's introductory price holds through 31 December 2026 and rises from 1 January 2027.

Argon tier prices, the 30 November 2026 end date, Batch and Priority rates, Contributor data terms, grounding allowances and Flash pricing verified against Google's Gemini API pricing page on 27 October 2026. No free-tier allowance is estimated because Google publishes none for Argon.

Sources

1. Gemini API pricing: ai.google.dev/gemini-api/docs/pricing. Used for: Argon tier rates, the 30 November 2026 deadline, Batch/Flex/Priority and Contributor terms.

All sources are the vendor's own plan or documentation pages, checked on 27 October 2026. Plan terms change without notice, so confirm current limits before you rely on them.