Google doubles the prices for Gemini 4 Argon in the Gemini API on December 1, 2026. Input tokens and output tokens will then each cost twice as much as before. According to the Gemini API pricing page, the current rates apply until November 30, 2026.
Affected are Argon calls made via an API key, for example in your own scripts, automations, or add-on programs. The price list only affects the API, not the Gemini app.
What will Argon cost from December?
The pricing page lists four tiers that will double from December: Standard, Batch, Flex, and Priority. Batch and Flex cost half the Standard tier, Priority almost double.
Tier
Until November 30 (input / output, US dollars per million tokens)
From December 1 (input / output, US dollars per million tokens)
Standard
2 / 10
4 / 20
Batch and Flex
1 / 5
2 / 10
Priority
3.60 / 18
7.20 / 36
Cached inputs in the Standard tier will also cost twice as much from December. The tokens the model generates internally for thinking also count as output and thus fall under the higher rate. A run with one million input tokens and 100,000 output tokens, including internal thinking, costs 3 in the Standard tier today by our own calculation and 6 from December. All amounts are in US dollars; a euro price for Germany is missing.
What can Argon do and what is missing?
Argon accepts text, images, video, audio, and PDF. The model delivers responses only as text. The context window holds 1,048,576 tokens, and a response may by default be 65,536 tokens long.
Argon at a glance
Gemini 4 Argon in the Gemini API, as of November 3, 2026
Input of text, images, video, audio, and PDF
Batch and Flex at half the standard price
Free tier
Explicit caching
Image and audio output
Live API
Source: Google AI Studio
Image and audio output as well as the Live API are missing. Caching runs automatically; Argon does not offer explicit caching. The model is not available in the free tier, so every token is billed.
The Gemini app is not part of the API price list. According to our report on Gemini 4 Carbon, Argon is available there in the Google AI Pro and Ultra plans; the entry-level AI Plus plan does not include Argon. For Carbon, Google adopts Argon's introductory price at launch.
Is the price jump a trend?
No uniform trend is apparent: At the end of October, OpenAI cut API prices by at least 50 percent with GPT-6.1 Terra. Prices for AI models are thus moving in both directions. For a budget, the model name in the calculation therefore matters, not just today's price.
What this means for you
Check in November which tools and automations call Argon with your Gemini key. The costs of these calls will double in December unless you change something.
In future, send non-urgent tasks via the Batch tier. From December, it costs as much as the Standard tier does today. Also test a lower thinking level (low or medium instead of high) and check quality as well as token consumption. Thinking is billed as output.
The Gemini tutorial on TutKit shows how to get started with prompts and practice using Gemini.
Google announces the price change on the Gemini API page, which we accessed on November 3, 2026.
