Google has released a native Gemini application for Mac. Gemini is the last of three major AI services to receive a separate app for Mac: OpenAI and Anthropic did so much earlier. However, the delay is compensated by functionality.
What the application can do
The application requires macOS Sequoia (15.0) or newer and runs exclusively on Apple Silicon. It can be called from anywhere using Option + Space for a mini-chat or Option + Shift + Space for fullscreen mode. Access is also available through the Dock and menu bar.
The key difference from the web version is desktop context. A user can share the active window, and Gemini provides explanations or hints about what is visible on screen, including local files. For example, if you open a complex chart and ask "what are the three main conclusions?" — the model provides an instant summary.
Any window on Mac can be passed to Gemini for contextual assistance. For the model to read full pages in the browser, you need to manually enable accessibility for Gemini in system settings — Privacy and Security → Accessibility.
Free — but with limits
The application is free to download, but free access is limited. Subscriptions: Google AI Plus — $7.99/month, Google AI Pro — $19.99/month, Google AI Ultra — $249.99/month.
There is also a new feature for paid users: the Temporary Chat feature allows you to have conversations that are not stored in history and are not used for model training.
Privacy: what you should know before granting screen access
Screen reading is a convenient feature, but it requires an informed decision. Gemini activity can be managed through Google Account → Data and Privacy → "Gemini Apps Activity", where you can disable logging or shorten the auto-deletion window.
"Gemini is suitable for everyday use, but I would not enter sensitive personal data into it without prior setup"
Tom's Guide, comparison of ChatGPT, Gemini, Claude, and Perplexity privacy
Google says the Mac app is just the first step toward a "personal, proactive, and powerful desktop assistant", and announces more news in the coming months.
If Google truly integrates Gemini with Gmail, Google Photos, and Calendar in desktop mode — as it has already done on the web through the Personal Intelligence feature — will the choice between convenience and privacy be equally obvious for users who have already granted screen access?
