Automate Basics
New model · Google

Google unveils Gemini 4 Argon for complex work tasks

By Automate Basics, written with AI from Google's original

· 2 min read

Hand comparing printed charts and documents on a sunlit meeting table
Image: AI-generated illustration.

Google has announced Gemini 4 Argon, a model built for extended coding, research and document tasks. Access is limited to trusted cyber defenders and testers for now, with paid API customers and Google AI Ultra subscribers due to receive it later.

What changed with Gemini 4 Argon

Google says Gemini 4 Argon is designed to carry out complex tasks that require sustained reasoning across several steps. Its stated uses include software development, financial research, legal drafting and cybersecurity defense. Google has expanded the model’s output limit to 1 million tokens, giving it room to produce much longer responses during a task.

Google also says Argon can work with charts, lengthy videos and collections of documents. For someone preparing a report or reviewing a complicated project, the relevant change is the model’s capacity to keep working through a larger task rather than stopping after a short response. Those capabilities are Google’s claims, not a guarantee that an output will be accurate or ready to use without review.

Who can use Gemini 4 Argon?

Gemini 4 Argon is currently going to trusted cyber defenders through Google’s Fairwind Program. Google says it is gathering feedback and strengthening safeguards before broader access. The company plans to start that wider release with paid API customers and Google AI Ultra subscribers, followed by other developers, enterprises and consumers. It has not given a date for those groups to receive the model.

Google lists an introductory usage price of $2 for a million input tokens and $10 for a million output tokens. Cached input tokens receive a 95% discount from the input price. The announcement does not say how long introductory pricing will last or specify a separate Argon price for Google AI Ultra subscribers.

How should I try Gemini 4 Argon at work?

Gemini 4 Argon is not yet broadly available, so most readers cannot try it immediately. When access reaches your workplace, a useful first task would be a contained piece of research: ask the model to examine a set of documents and a related chart, then produce a draft summary that a colleague can check. That fits the document and visual work Google describes without handing over a final decision.

Check whether the summary reflects the supplied material, whether figures match the chart and whether important qualifications have been left out. For legal, financial or security work, keep the appropriate human reviewer involved. Google says safeguards are still being developed ahead of wider release, and its announcement does not provide a general availability date.

Frequently asked questions

Is Gemini 4 Argon available to everyone?

No. Google says Gemini 4 Argon is rolling out first to trusted cyber defenders through its Fairwind Program. Broader access is planned to begin with paid API customers and Google AI Ultra subscribers, but Google has not announced when they will receive it.

What kinds of work is Gemini 4 Argon meant to handle?

Google describes Gemini 4 Argon as a model for extended tasks involving coding, financial research, legal drafting and cybersecurity defense. It also says the model can analyze charts, long videos and groups of documents. Work involving consequential decisions still needs a person to check the result.

Tools in this piece

Written with AI from Google's original and published after automatic checks: every figure here appears in the original, and no sentence is copied from it. The picture is AI-generated. The original is the authority.

Source: Gemini 4 Argon: our next era of frontier intelligence, Google, 30 Sept 2026.