Some interesting news today - Google is doubling the introductory API prices for Gemini 4 Argon on 1 December 2026, giving developers and businesses until the end of November to use the new AI model at its launch rates - the model launched on 13 October and is designed for complex tasks involving coding, research, finance and other multi-step workflows.
The increase affects API usage rather than automatically changing the price of a consumer Gemini subscription. Businesses building their own applications, automation systems and AI-powered software are the main audience affected by the announcement.
How Much Will Gemini 4 Argon Cost?
Gemini 4 Argon currently costs $2 per million input tokens and $10 per million output tokens. From 1 December, those rates will rise to $4 and $20 respectively.
| API usage | Until 30 November | From 1 December |
|---|---|---|
| 1 million input tokens | $2 | $4 |
| 1 million output tokens | $10 | $20 |
| 10 million input + 2 million output tokens | $40 | $80 |
These figures exclude any applicable taxes and assume the same standard token rates throughout. Google's pricing documentation also says cached input tokens receive a 95% discount, with their price increasing from $0.10 to $0.20 per million tokens.
For a developer testing a new application, the difference might initially be small. For a business processing millions of tokens every day, however, doubling the rates could make a noticeable difference to its monthly AI bill.
What Does the Price Increase Mean for Businesses?
Gemini 4 Argon is aimed at demanding work, including coding, complex reasoning and tasks that require several steps, and Google has also highlighted its use in software engineering and enterprise workflows - businesses using Argon should review their expected token usage before the new prices take effect, as they may be able to reduce costs by routing straightforward requests to cheaper models.
Developers should also benchmark alternatives rather than assuming the most capable model is necessary for every request. A less expensive model may be sufficient for routine classification, summaries or simple customer-support tasks, while Argon could be reserved for work where stronger reasoning makes a measurable difference.
Source: Google Gemini API documentation and Google's Gemini 4 Argon announcement.