Gemini 4 Argon Pricing Doubles on December 1: What the Intro Rates Mean for Your API Bill

Gemini 4 Argon costs $2 per million input tokens and $10 per million output tokens until November 30, 2026, and exactly double that, $4 and $20, from December 1. Google states both schedules on its own pricing page, so any API bill built on the introductory rate will double with no change in usage. It affects every team that moved a production workload to the model after its general availability on October 13.

Key facts

Gemini 4 Argon (gemini-4-argon) became generally available on October 13, 2026.
Standard rates: $2.00 input and $10.00 output per million tokens through November 30, 2026; $4.00 and $20.00 from December 1, 2026.
Cached input goes from $0.10 to $0.20 per million tokens. Only implicit caching is supported.
Batch and Flex run at $1 input and $5 output until November 30, then $2 and $10.
Priority runs at $3.60 and $18 until November 30, then $7.20 and $36.
The model is not available on the free tier, and thinking tokens are billed as output.

What the pricing page actually says

Every rate on Google's Gemini API pricing page for Argon carries two dates: a "through November 30, 2026" figure and a "starting December 1, 2026" figure. The change applies to all tiers, not only the standard one. Google's changelog records the October 13 general availability. A separate third-party breakdown published in late October reports the same schedule.

Older articles about the model describe a preview with no public API and no published batch pricing. Those pages predate general availability and should not be used for planning.

What it does to a real bill

The arithmetic is simple because both rates double. Take a workload of 100 million input tokens and 20 million output tokens a month, which is a modest support or coding assistant deployment.

Standard, through November 30: 100 x $2 + 20 x $10 = $400.
Standard, from December 1: 100 x $4 + 20 x $20 = $800.
Batch, through November 30: 100 x $1 + 20 x $5 = $200.
Batch, from December 1: 100 x $2 + 20 x $10 = $400.

These figures use the published list rates and ignore caching and grounding. They are our own calculation, not a Google example.

Output is the larger risk. Google bills thinking tokens at the output rate, and the model is built for long reasoning and long decodes up to 1M tokens. A job that produces long reasoning traces costs more per request than the visible answer suggests, and after December 1 each of those tokens costs $20 per million instead of $10.

Where the extra cost does not hit

Argon has no long-context surcharge on the pricing page, so a 900K-token prompt is billed at the same per-token rate as a short one. Search grounding gives 5,000 free requests a month shared across Gemini 3 and newer models, then $14 per 1,000 requests. Neither of those changes on December 1, so the doubling applies to token charges only.

What to do before December 1

Re-run your cost model at $4 and $20. If a feature only works financially at the introductory rate, you want to know that in November, not on the December invoice.
Move non-urgent work to Batch or Flex. Both tiers stay at half the standard rate, which at the new prices brings Argon back to $1 and $5 per million tokens for anything that can wait.
Measure thinking tokens per request. Log billed output tokens against the length of the returned answer for two weeks. That ratio is the number that decides your real unit cost.
Test a cheaper model on the easy traffic. Price moves of this size are the moment to route simple requests elsewhere. Our breakdown of OpenAI's Luna price cut shows how far a small tier has fallen, and the state-of-the-models update lists the alternatives at the time.
Do not rely on the free tier. Argon is not offered there, so prototypes need a paid key from the first call.

For wider comparisons, browse the tool directory or the full article archive.

Questions people ask

When does Gemini 4 Argon pricing go up?

On December 1, 2026. The introductory rates apply through November 30, 2026, and the pricing page lists the new rates as starting December 1.

Is there a free tier for Gemini 4 Argon?

No. Google's pricing page states that the model is not available on the free tier.

Are thinking tokens billed?

Yes. Google says every token the model processes is counted and billed, including thinking tokens, and the output price includes thinking.

Does batch pricing also double?

Yes. Batch and Flex move from $1 input and $5 output to $2 and $10 per million tokens, so they stay at half the standard rate.

Sources