Google Announced Pricing for New Argon AI Model
The company set developer costs for its latest model, establishing a wide price gap between input and output tokens.
Updated on Sept. 30, 2026 in Artificial Intelligence
Live Poll
Would you pay for AI services that charge based on the amount of text generated?
Google has announced pricing for its new AI service, Argon, set to join the company's existing Gemini model family. Developers will be charged $2 per million input tokens and $10 per million output tokens for the service.
Why it matters
The pricing structure reflects the technical reality that generating model responses requires significantly more computational resources than processing incoming prompts. By segmenting costs, Google aims to align development fees with the specific hardware demands of generative tasks.
Argon carries an output-to-input cost ratio of 5:1, with developers paying $10 per million output tokens compared to $2 per million input tokens. This benchmark highlights the higher computational load required for token generation during inference.
The players
A global technology company focused on search, cloud computing, and the development of large-scale artificial intelligence models like Gemini and Argon.
The details
Argon pricing is based on tokens—the fundamental units of text or code that models process as input and produce as output. Because the model must perform iterative, step-by-step computation to generate an answer, the output process demands more server-side resources than reading an input prompt. This asymmetric computational cost explains the five-fold difference in price between input and output tokens.
Timeline
September 2026: Google announced the official pricing structure for Argon.
The Tech Race
Argon enters the market as a new offering intended to operate alongside Google's established Gemini model family. This move follows a broader industry trend of differentiating AI services by task-specific computational costs rather than uniform usage fees.
Developers building on Argon will need to account for a $12 total cost for every one million tokens processed with an even split of input and output. Scaling to 100 million tokens will result in a $1,000 charge for output generation, compared to $200 for the input phase.
The takeaway
The pricing disparity underscores the high compute requirements of generative inference over simple data ingestion. Developers should monitor the release of public API benchmarks for Argon to determine if its performance justifies the 5:1 output-to-input cost premium.
Further reading
For broader context on how providers structure costs for developers, see the latest updates in Artificial Intelligence.
Source note: This article includes information reported by Crypto Briefing.
Live Poll
Would you pay for AI services that charge based on the amount of text generated?
