What Is Gemini 4 Argon?

Gemini 4 Argon is Google’s latest flagship AI model, presented as a system for demanding professional and developer work. Google’s description emphasizes “deep work”: complex tasks that may require multiple steps, sustained reasoning, and work across specialist domains. The company names coding, finance, legal work, cybersecurity, and science among the areas where it expects Argon to be useful.

The launch happened in stages. Google announced the model on September 30, 2026, but initially restricted access while it gathered feedback and evaluated safeguards. The first cohort included trusted cyber defenders and early testers. Reporting describes the cybersecurity access initiative as the Fairwind Program. Google then announced broad availability on October 13, opening access to developers and enterprise customers worldwide through its relevant products and platforms.

This distinction matters for anyone looking up the release date. September 30 marks the announcement and limited initial rollout; October 13 is the date associated with broad availability. The staged approach reflects Google’s stated intention to test and strengthen safeguards before expanding access to a more powerful model. It should not be mistaken for evidence that every product, account type, or region necessarily has identical access conditions.

Gemini 4 Argon Pricing

Google launched Argon with introductory usage-based rates of $2 per million input tokens and $10 per million output tokens. Cached input tokens are priced at 95% off the input-token rate. Google’s official announcement says the promotional period expires on November 30, 2026. After that, the stated rates for the public-access Argon model are $4 per million input tokens and $20 per million output tokens.

These are token rates, not a fixed subscription price. The amount a project spends depends on the tokens it sends to and receives from the model and the applicable pricing period. Cached inputs can substantially reduce the cost of eligible repeated input, but the exact impact depends on how a particular application uses cached tokens. Do not equate the introductory input rate with the complete cost of a workflow: output tokens have their own rate, and the rates change after November 30.

For a simple illustration, one million input tokens billed at the introductory input rate would cost $2 before considering output tokens or any other applicable conditions. One million output tokens at the introductory rate would cost $10. The arithmetic provides a starting point for estimating consumption; it is not a full quote for a production deployment. Teams should check the applicable Google product, account, and billing terms before committing to a budget.

Where Developers and Businesses Can Access Argon

Google’s official announcement describes Gemini 4 Argon as globally available to developers and enterprises through the API in AI Studio and the Gemini platform. Google Cloud is also identified as an access route in coverage of the broad launch. For consumer-facing access, the announcement lists Google AI Ultra and Pro subscribers in the Gemini app and AI Mode in Search. Google Antigravity access is listed for Google AI Ultra...