Google shipped a native Gemini app for macOS. Hit Option + Space and a floating chat window appears, no browser required. It's free, works with your Google account, and syncs everything across devices.
The main draw is contextual awareness. Share a specific window with Gemini and it can see what's on your screen. Working through code, reviewing a document, or stuck on a spreadsheet? The AI has eyes on it. Google built this using Apple's ScreenCaptureKit, which gives granular window-level access rather than capturing your whole display. The floating chat panel is designed not to steal focus, so you can keep working while you ask questions.
There are restrictions. macOS Sequoia 15.0 is the minimum, and Intel Macs need not apply.
The app also bundles creative tools: image generation and video creation via Veo. On the web, these feel like novelties you'd visit once and forget. On desktop, they're more interesting. Generate an image and drag it into a Keynote deck. Create a video clip and drop it into a project folder. The point isn't the tools themselves, but removing the friction between making something and actually using it somewhere else.
The web version works fine, so why bother? Speed and convenience. A system-wide hotkey makes Gemini feel less like a website and more like a native utility, closer to Spotlight than ChatGPT. That's Google's play here. Get Gemini into muscle memory as something you summon, not something you visit. Whether that's enough to matter depends on what they build on top of this foundation.