Gemini 4 Argon: who gets access and what the announced prices mean

Google starts Gemini 4 Argon access with Fairwind partners. We explain announced pricing, the output limit and how CodeMender fits.

Glowing faceted core inside a protective ring

Google introduced Gemini 4 Argon on September 30. Initial access is restricted: Fairwind offers the model to selected cybersecurity partners. For people waiting for a new option in the everyday Gemini app, this marks the beginning of a staged release.

What Google announced

According to the announcement, Argon targets complex work, including coding, and raises the output limit to one million tokens. That is a generation limit, not the input context size. Broader access through the paid API and Google AI Ultra is planned later, without a specific date.

Announced introductory rates are $2 per million input tokens and $10 per million output tokens, rising to $4 and $20 after that period. No end date was specified.

Who qualifies for early access

The Fairwind conditions include participant vetting and authorized defensive uses. Organizational requirements include secure authentication and access controls; reselling access is prohibited. Joining the program should therefore be treated as a separate process rather than an automatic benefit of a consumer subscription.

Where CodeMender fits

Alongside the model, Google Cloud offers CodeMender, an agent for finding, validating and fixing vulnerabilities. It supports multiple models, and developers can review and approve proposed changes within their existing tools. Having CodeMender in a workflow does not, by itself, establish access to Argon.

A team evaluating such a service can separate three decisions: whether the code review workflow suits its needs, whether the intended model is available and what a completed assessment costs. This avoids choosing a product around the advertised rate of a model that the team cannot yet use.

What this means for everyday users

Our assessment: changing a subscription solely because of this announcement is premature. First confirm access in your own account and identify the task you want the new model to perform. For a current model comparison, see our Sonnet 5.5 explainer. Before submitting work files, review the AI data privacy checklist.

Discussion

Join the conversation

Stay on topic and respect other readers. Your first comment may appear after editorial review.

Leave a comment

Your email address will not be published. Required fields are marked with an asterisk.

By submitting a comment, you agree to moderation and to the storage of the information you provide under our privacy policy.