Google Ships a Native Gemini App for Mac Built in 100 Days
Google launched a free native Gemini app for Mac with screen sharing, window context, image and video generation, and a global Option+Space shortcut - built in pure Swift with 100+ features in under 100 days.
4 min read
TL;DR
- Google launched a free native Gemini app for Mac - pure Swift, Apple Silicon, macOS Sequoia or later
- Option+Space opens a mini chat anywhere; Option+Shift+Space opens the full interface
- Screen and window sharing let Gemini see what you're looking at and respond in context - code, documents, spreadsheets, anything visible
- Includes Gemini Live (voice), Deep Research, image generation (Nano Banana), video generation (Veo), music generation, Canvas, and NotebookLM integration
- A small team built 100+ features in under 100 days using Google's AI coding tool Antigravity - Sundar Pichai said the prototype came together in days
Google has been conspicuously absent from the Mac desktop while OpenAI's ChatGPT app and Anthropic's Claude desktop have been available for months. That gap closed today.
The Gemini for Mac app launched April 15 as a free native download. It's not an Electron wrapper or a PWA - the team built it in 100% native Swift, which explains why early reviews are calling it fast.
What it does
The core interaction model: hit Option+Space from anywhere on your Mac and a mini chat opens. Ask a question, get an answer, dismiss it. Option+Shift+Space opens the full experience. Both shortcuts are customizable.
The standout feature is window and screen sharing. Select "Share Window" after activating Gemini and it can see whatever you're looking at - a code editor, a spreadsheet, a PDF, a design file. Ask questions about visible content without copy-pasting anything.
The full feature set includes:
- Gemini Live - real-time voice conversation
- Deep Research - intensive multi-step information gathering
- Canvas - collaborative workspace
- Nano Banana - high-speed image generation
- Veo - AI video generation
- Music generation
- NotebookLM integration - connect to your existing research notebooks
- Cross-device sync - chat history follows your Google account across Mac, web, and mobile
It's available for users 13+ in all countries where Gemini is supported. Free to download, with paid subscription tiers for advanced features.
100 features in 100 days
The development story is the more interesting angle. A small Google team used Antigravity - Google's internal AI coding tool - to go from idea to prototype in days and ship 100+ features in under 100 days. CEO Sundar Pichai highlighted this as a demonstration of what AI-assisted development looks like at Google's scale.
The all-Swift codebase means the app runs natively on Apple Silicon with no translation layer. Early benchmarks from tech reviewers note snappy response times and low memory usage compared to Electron-based competitors.
How it compares
| Feature | Gemini for Mac | ChatGPT Mac | Claude Desktop |
|---|---|---|---|
| Screen/window sharing | Yes (native) | Yes (screenshot) | No |
| Voice conversation | Yes (Gemini Live) | Yes (Advanced Voice) | No |
| Image generation | Yes (Nano Banana + Imagen) | Yes (DALL-E) | No |
| Video generation | Yes (Veo) | No | No |
| Global shortcut | Option+Space | Option+Space | No default |
| Native Swift | Yes | No (Electron) | No (Electron) |
| Free tier | Yes | Yes (limited) | Yes (limited) |
| Code context | Window sharing | Screenshot | Claude Code (CLI) |
Gemini's advantage is breadth: it's the only desktop AI app shipping voice, image gen, video gen, screen context, and research tools in one native package. Claude's strength remains coding (via Claude Code's terminal integration). ChatGPT sits in the middle.
The window-sharing approach is more fluid than screenshot-based context because it updates in real time as you work. Whether Gemini's answers are as good as Claude's or GPT's for specific tasks is a separate question that our April rankings cover in detail.
The rough edges
Early user reports note a few issues: occasional lag on first launch, some UI quirks in the mini chat overlay, and the absence of a dedicated coding mode (Gemini sees your code through window sharing but doesn't integrate with editors the way Claude Code or GitHub Copilot do). Voice support via Gemini Live isn't available in all regions yet.
Google says updates including expanded voice support are coming.
Sources:
