Gemini for macOS - your native AI desktop app
1 min read
Originally from gemini.google
View source
My notes
Summary
Google shipped a native Gemini macOS app that puts an LLM one keystroke away via Option+Space, with optional screen-window sharing so the model can answer based on what’s actually on screen. It’s free, Apple Silicon only, requires macOS Sequoia 15.0+, and syncs chat history across devices via the Google account.
Key Insight
- Native desktop entry point, not a wrapper: Option+Space opens mini chat, Option+Shift+Space opens full chat. Both are remappable in Settings.
- Contextual grounding via window sharing: instead of pasting code/docs, you share a specific window. Enabling Accessibility in System Settings lets Gemini read full browser pages, not just visible viewport.
- Bundles Google’s creative stack inside the app: Nano Banana (image generation) and Veo (video generation) are one click away, no separate site.
- Hard constraints: Apple Silicon only (no Intel Macs), Sequoia 15.0 or later. This mirrors Apple Intelligence’s hardware floor and quietly sidelines older machines.
- Free in all countries where Gemini is available, in all supported languages. Notable competitive move against ChatGPT desktop (which gates advanced features behind Plus).
- Menu bar + Dock + shortcut = three parallel entry points. Reduces the “where is the app” friction that killed adoption of the old Gemini web-only flow.
- Sync story: same Google account = chat history + memory across devices. This is the lock-in, once memory builds, switching apps has a cost.