Gemini 4 Argon Price Doubles December 1, 2026: What Your API Bill Becomes

Gemini 4 Argon's API price goes from $2 / $10 to $4 / $20 per million tokens on December 1, 2026. What common tasks will cost, and how to keep your bill flat.

By the PromptSaver team · Updated November 16, 2026 · 4 min read

The short answer: Google's introductory price for Gemini 4 Argon ends on November 30, 2026. From December 1, the API costs $4 per million input tokens (up from $2) and $20 per million output tokens, including thinking tokens (up from $10). Cached input goes from $0.10 to $0.20 per million (Google).

Every Argon request will cost twice what it costs today. If you call Argon through the API, you have two weeks to decide whether to pay that, cut your usage, or move some of the work to another model.

This is an API price. Subscriptions to the Gemini app are priced separately.

What changes on December 1

Gemini 4 Argon, per million tokensUntil Nov 30, 2026From Dec 1, 2026
Input$2.00$4.00
Cached input (implicit caching)$0.10$0.20
Output, including thinking tokens$10.00$20.00

Google's model page also notes that Argon isn't available on the free tier, and that every token the model processes is billed, including thinking tokens. Prices were checked on Google's own page on November 16, 2026.

What common tasks will cost

Here's what 1,000 runs of six everyday jobs cost on Argon before and after the change, at list prices (input / output tokens in brackets; divide by 1,000 for one run).

Task (tokens in / out)Until Nov 30From Dec 1
Product description (300 / 250)$3.10$6.20
Support reply with context (1,500 / 300)$6.00$12.00
Email or blog draft (800 / 1,500)$16.60$33.20
Code fix (4,000 / 800)$16.00$32.00
Long document summary (20,000 / 800)$48.00$96.00
Reasoning task (2,000 / 6,000)$64.00$128.00

At 100,000 support replies a month, that's $600 a month today and $1,200 from December 1. To run your own numbers, the free AI cost calculator already switches Argon to the new price on December 1, so you can see both.

The reasoning row is the one to watch. Thinking tokens are billed as output, so a model that thinks for a few thousand tokens before answering costs far more than its sticker price suggests, and doubling the output price doubles that too.

How Argon compares after the increase

From December 1, Argon's list price matches Claude Opus 5.5 ($4 / $20, per Anthropic, checked November 16) and is twice the price of Claude Sonnet 5.5 ($2 / $10). On OpenAI's pricing page as of October 30, GPT-6.1 Sol was also $2 / $10.

Same list price doesn't mean same bill. Models use different numbers of tokens for the same text (current Claude models count roughly 35% more than Gemini for English), and some write much longer answers than others. Test your own prompts before switching on price alone. Our LLM API pricing comparison covers every model side by side.

Five ways to keep your bill flat

  1. Move simple jobs off Argon. Tagging, short rewrites and extraction rarely need a top model. Google's Gemini 3.1 Flash-Lite costs $0.25 / $1.50 per million tokens, a fraction of Argon's new price.
  2. Cap answer length. Output is the expensive side, five times the input price. Asking for "under 150 words" or setting a maximum output length cuts the part of the bill that's doubling the most.
  3. Limit thinking. If a task doesn't need deep reasoning, use a lower thinking setting where the API offers one. Hidden thinking tokens are billed at the full output price.
  4. Reuse the same opening text. Implicit caching discounts repeated input by 95%. Put the instructions and reference text that stay the same at the start of every request, and only the changing part at the end.
  5. Send less context. Chat history and pasted documents are re-sent with every request. Summarize long threads instead of resending them in full. Our guide to reducing AI costs covers each of these in more detail.

What to do before December 1

  • Find out how much of your bill is Argon. Check your Google Cloud or AI Studio billing for the last month and double it. That's your December bill at the same usage.
  • Test a cheaper model on your real prompts this month, while Argon is still at the old price, so you can compare quality and cost side by side.
  • Update your budget. If you charge customers per request or per seat, check your margins against the new price.

Frequently asked questions

When does Gemini 4 Argon's price go up?

On December 1, 2026. Google's introductory price of $2 per million input tokens and $10 per million output tokens is available through November 30, 2026.

How much will Gemini 4 Argon cost from December 1?

$4 per million input tokens and $20 per million output tokens, including thinking tokens. Cached input goes from $0.10 to $0.20 per million tokens.

Does the price increase affect the Gemini app?

The announced change is to the Gemini API. Gemini app subscriptions are priced separately, so check Google's plan page for those.

Is Gemini 4 Argon available for free?

Not through the API. Google's model page says Argon isn't available on the free tier.

What's the cheapest alternative to Gemini 4 Argon?

It depends on the job. For simple tasks, Gemini 3.1 Flash-Lite costs $0.25 / $1.50 per million tokens. For harder work, Claude Sonnet 5.5 lists at $2 / $10, the same as Argon's current intro price. Test quality on your own prompts before switching.

Sources

Gemini 4 Argon prices and dates: Google AI for Developers, checked November 16, 2026. Claude prices: Anthropic, checked November 16, 2026. Gemini 3.1 Flash-Lite: Gemini API pricing, checked November 16, 2026. Task costs are our own math from those list prices.

Gemini 4 Argon Price Doubles December 1, 2026: What Your API Bill Becomes | PromptSaver