Google sets Argon pricing at $2 per million input tokens and $10 per million output

Google has put a price on Argon, its new AI service. Developers will pay $2 for every million input tokens and $10 for every million output tokens.

The pricing was announced in late September 2026. Google has not said which division is behind the product.

What the numbers actually mean

A token is a chunk of text that an AI model reads or writes, often a word or part of one.

Input tokens are what you feed the model: your prompt, your documents, your instructions. Output tokens are what the model produces in response.

Under Argon’s pricing, output costs five times as much as input. A developer who sends Argon one million tokens and gets one million back would pay $12 in total: $2 for the input and $10 for the output.

An app that pushes 100 million input tokens through Argon would spend $200 on the reading side. If the model then generates 100 million tokens of output, that half of the bill comes to $1,000.

Why output costs more

Generating a response takes more computation than processing a prompt. The model has to produce its answer piece by piece, and each new token requires work. Providers pass that cost along, and Argon follows the same pattern.

Token-based billing is the standard model for AI APIs, so Argon is not asking developers to learn a new pricing system.

Where Argon fits in Google’s lineup

The service is expected to be positioned alongside models like Gemini, Google’s flagship AI family. Google has not tied Argon to any named partnerships, and it has not drawn direct comparisons to its other products.

No public details are available on Argon’s capabilities, context window, or release date, and major technology news outlets have not yet reported on the service in detail. This suggests Argon may be in an early stage of development or testing.

What this means for developers and investors

The five-to-one gap between output and input rewards products designed around heavy reading and concise answers. Retrieval tools, classification systems, and summarization features tend to consume far more text than they produce and would sit on the cheaper side of Argon’s meter. Products that generate long-form text, code, or extended conversations will feel the $10 output rate more sharply.

For investors, the key question is whether Argon expands Google’s reach or simply splits attention between Google’s own offerings.

Watch for three things next: any disclosure of Argon’s capabilities and context window; a release schedule; and how Google positions Argon against Gemini.