Google began a worldwide rollout of Gemini 4 Argon on Oct. 13, offering the model through its paid Gemini API to developers and enterprises, and to Google AI Pro or Ultra subscribers in the Gemini app and Search AI Mode. Google Antigravity access is limited to Ultra subscribers. API prices start at $2 per million input tokens and $10 per million output tokens through Nov. 30, then rise to $4 and $20, respectively.
The launch changes who can try Argon and where: developers and businesses can call it through the API, while consumer access depends on a paid Google AI plan. The announcement does not describe an open-weight release or a free API tier. Google says its public model has safeguards; cybersecurity defenders can apply through the Fairwind program for a version without cyber guardrails.
Argon’s token budget is shared across the request and answer
Google’s API documentation lists a unified limit of 1,048,576 tokens. The prompt and everything the model generates, including internal thinking and the visible answer, count toward that same total. A standard synchronous response can produce up to 65,536 output tokens. For a longer answer, Google’s long-decode option runs the request in the background and continues generation in successive chunks; it does not add a second token allowance.
The introductory API rate lasts through Nov. 30. From Dec. 1, Google’s published price doubles for both input and output tokens. The published rates are per million tokens; actual usage cost depends on the request’s input and generated output.
Public access keeps cyber safeguards in place
Google says Argon was built for multi-step work in areas such as software engineering, finance, law, science and cybersecurity. Its launch post also distinguishes the public model from Fairwind access: cybersecurity defenders can apply through Fairwind for a version without cyber guardrails, while the public release retains protections. The company’s reported capability claims and examples are not independent evaluations, and access remains subject to plan and API terms.
Sources
- Google’s Oct. 13 availability announcement
- Gemini API token limits and long-decode documentation
- Illustrative image by SimonWaldherr, CC BY-SA 4.0, via Wikimedia Commons
- CC BY-SA 4.0 license
