The short version
● What Gemini's Mac app can now operate your Mac - native apps and websites alike.
● Control A live preview in the chat, and you approve each action before it runs.
● Who gets it Google AI Plus, Pro or Ultra subscribers.
● Where United States, English only, Mac only, for now.
Google has given the Gemini desktop app on macOS the ability to complete tasks on your behalf — navigating native Mac applications and browsing websites in the background while you watch it happen in a live preview inside the chat window. You can pause it, stop it, and approve what it is about to do before it does it.
The examples Google gives are deliberately unglamorous, which is the right signal: filling out forms, repeating the same task across a batch of files, sorting files into folders by whatever criteria you describe, and building a slideshow out of files or an online dashboard. These are the jobs people actually lose afternoons to.
What it needs
| Requirement | Detail |
|---|
| Device | Mac. The Gemini app for macOS specifically |
| Subscription | Google AI Plus, Pro or Ultra |
| Region | United States |
| Language | English |
| Approval | You approve actions before they execute |
| Other platforms | Not announced |
Google did not say when this reaches other countries or other operating systems, and it has not given the feature a product name of its own — it is simply computer use inside the Gemini Mac app.
Why the approval step is the interesting part
An agent that can click things on your machine is only useful if you trust it near your files, and trust is not something a launch post can grant. Putting a live preview and a per-action approval in front of every step is the honest answer to that: it makes the agent slower and it makes it auditable, and for a first release aimed at file sorting and form filling, slower and auditable is the correct trade.
It also sets the bar for everyone shipping desktop agents. The question stops being whether a model can drive a GUI — several can — and becomes whether you can watch it and stop it.
Where it sits against the rest of Gemini
This is a client feature rather than a model release, so it does not change anything about the Gemini API or the models underneath. If you are weighing Gemini against the alternatives on price rather than features, our 19 November roundup covers the current model line-up and what each tier costs.
FAQ
Can Gemini control my Mac?
Yes, in the Gemini app for macOS. It can navigate native Mac apps and browse websites in the background, with a live preview and per-action approval.
Do I need a paid plan for Gemini computer use?
Yes. It requires a Google AI Plus, Pro or Ultra subscription.
Is Gemini computer use available on Windows?
Not yet. Google announced it for the Mac app and has not said when it reaches other platforms.
Is it available outside the US?
Not at launch. It starts in the United States, in English.
Does Gemini ask before it does something?
Yes. You can pause, stop and approve actions before they run, and watch the agent work in a live preview.