Gemini 4 Argon is built for longer work tasks: Google’s announced features, API prices and access limits

Google announced Gemini 4 Argon on September 30, starting with selected cybersecurity partners. Its Fairwind programme vets organisations and restricts access to approved teams. This is not a general release to everyone using the Gemini app.

Google describes Argon as designed for sustained software, legal and financial tasks, including analysis of documents, charts and video. These capabilities are company claims.

Google says the output-token limit rises from 64,000 to one million. This measures generated output, rather than the amount of material supplied to the model.

Introductory API pricing is $2 per million input tokens and $10 per million output tokens, rising subsequently to $4 and $20. Developers pay according to usage.

Google says the later rollout will start with paid API customers and Google AI Ultra subscribers. It has not announced a public-release date. Fairwind partners are separately prohibited from reselling or sharing their restricted model access.