Google Introduces Argon AI Service with Competitive Pricing for Developers

Google has officially launched pricing details for its new AI service, Argon, aimed at developers. As of late September 2026, the service will charge $2 for every million input tokens and $10 for every million output tokens. This pricing structure reflects a common model in AI APIs, where output costs significantly more than input, with Argon’s output priced at five times the input rate.

A token is defined as a segment of text that AI models process, with input tokens representing the data fed into the model and output tokens being the responses generated by the model. For instance, if a developer sends one million tokens to Argon and receives one million back, the total cost would amount to $12.

The pricing strategy indicates that applications requiring extensive reading and concise outputs, such as retrieval tools and summarization systems, will benefit from the lower input costs. In contrast, applications that generate lengthy content or complex interactions may incur higher expenses due to the output pricing.

Argon is expected to complement Google's existing AI models, including the Gemini series. However, Google has not disclosed specific details regarding Argon’s capabilities or its release timeline, suggesting that the service may still be in development or testing phases.

In related news, Google’s Gemini 4 Argon has achieved a score of 53 on the Artificial Analysis Intelligence Index, matching OpenAI’s GPT-6 Astra. This score positions Google among the top three AI labs, with Argon demonstrating fewer hallucinations and improved agentic skills compared to its predecessors.

As the rollout progresses, Google plans to provide access to trusted cyber defenders first, with a broader release expected for paid API customers. The introductory pricing is set to double after the promotional period, highlighting the urgency for developers to explore Argon’s capabilities.