Google Gemini finally has a Mac app, and it's gunning for ChatGPT's desktop lead

Google’s Gemini now runs natively on macOS. The company shipped a dedicated Mac app on April 15, built entirely in Swift by a small team that Google says went from idea to prototype in days.

That makes Gemini the last of the big three AI assistants to land on the Mac desktop. OpenAI’s ChatGPT has been there for months. Anthropic’s Claude beat it too. The question isn’t whether Google is late (it is), but whether what it shipped is worth switching for.

What you actually get

The app installs from gemini.google/mac and drops two things onto your system: a dock icon and a menu bar spark. The real entry point is a keyboard shortcut.

Option + Space opens what Google calls the “mini chat,” a Spotlight-style floating window that sits on top of whatever you’re working in. It stays visible for 10 minutes by default (configurable up to 30), so you can fire off quick questions without losing your place. Option + Shift + Space opens the full chat window when you need more room.

The feature list reads like a greatest-hits of everything Google has built for Gemini’s web and mobile apps:

• Deep Research for multi-step investigation tasks
• Image generation via Nano Banana
• Video generation via Veo
• Music creation
• Canvas for collaborative editing
• Voice interactions with multiple voice options
• File sharing from local files, Google Drive, Google Photos, and NotebookLM

The standout feature for desktop use is Share Window. You can link any open application window (Pixelmator Pro, your IDE, a browser tab, a spreadsheet) and Gemini will analyze what’s on screen. It’s read-only, so it won’t click buttons or type for you. Think of it as screenshotting your work into the conversation, except it updates as you keep talking. You’ll need to grant Accessibility permissions in System Settings for it to read full browser pages.

One notable gap at launch: Gems (Google’s custom AI personas) aren’t available yet. You’ll still need the web or mobile app for those. Josh Woodward, VP of the Gemini App, described this as a first release. “We recruited a small team. They built 100+ features in less than 100 days. 100% native Swift,” he said.

The app includes a model switcher with three tiers: Fast, Thinking, and Pro. Chat history and conversation memory sync across every device when you’re logged into the same Google account, so a research thread you started on your phone picks up exactly where you left it on the Mac. That cross-device continuity is something neither ChatGPT nor Claude handles as cleanly.

How it compares to ChatGPT and Claude on the desktop

Here’s the honest breakdown as of April 2026.

ChatGPT’s desktop app has the deepest macOS hooks right now. It had a head start, and OpenAI used that time to build tighter OS-level integrations. Its voice mode is still considered the strongest of the three. If you want a general-purpose assistant that handles the widest range of tasks with the least friction, ChatGPT is the safer pick today.

Claude’s desktop app plays a different game. Anthropic’s bet is on deep context (up to 1 million tokens with Opus 4.6) and precise instruction following. Claude also has Computer Use features that let it interact with your machine more directly than Gemini’s read-only window sharing. For developers and writers who throw large codebases or long documents at their AI, Claude remains hard to beat.

Gemini’s edge is the Google ecosystem. If your life runs on Gmail, Docs, Sheets, and Meet, no other AI assistant can pull context from those services as natively. Deep Research goes further than the competition’s equivalents, pulling from more sources with longer reports. And on raw cost, Gemini is aggressive: API pricing for Gemini 3.1 Pro sits at $1.25/$5 per million input/output tokens, compared to $2.50/$15 for GPT-5.4 and $15/$75 for Claude Opus 4.6.

On benchmarks, GPT-5.4 and Gemini 3.1 Pro are tied at 94 on aggregate scores, with Claude Opus 4.6 close behind at 92. But benchmarks rarely capture what matters most on a desktop: how fast the thing launches, how well it stays out of your way, and whether it actually understands what’s on your screen when you need it to.

The Windows version is a different story

Google also released a “Google” app for Windows on April 14, one day before the Mac launch. Don’t confuse the two. The Windows app is a broader Google Search portal that includes Gemini access, not a dedicated Gemini-first experience. PCWorld’s hands-on called it “just another Gemini portal.” It’s English-only at launch and uses Alt + Space instead of the Mac’s Option + Space.

The gap between the Mac and Windows apps tells you something about Google’s priorities. The Mac app was built by their Antigravity team as a native Swift showcase. CEO Sundar Pichai noted on X that the team “went from an idea to a native Swift app prototype in a few days.” The Windows app feels like it was built to check a box.

The pricing question

The app download is free. What you get depends on your subscription:

• Free tier: limited usage, basic features
• AI Plus ($7.99/month): expanded access
• AI Pro ($19.99/month): full Gemini 3.1 Pro
• AI Ultra ($249.99/month): everything, including priority access
• Workspace customers: full access through existing business plans

Workspace customers get the app through their existing business plans. Google confirmed it’s available immediately to both Rapid Release and Scheduled Release domains, turned on by default for all organizations with Gemini enabled. Admins can control access through the Generative AI settings in the Workspace Admin console.

For the 750 million people already using Gemini across web and mobile, the desktop app is mostly about convenience. The system requirements are modest (macOS Sequoia 15.0 or later), and the DMG download weighs in at a reasonable size from gemini.google/mac.

What this means for you

If you’re already paying for Google Workspace or a Gemini subscription, download it. The mini chat window alone is worth the install, it’s the fastest way to get an AI answer without leaving your current app. If you’re evaluating all three desktop assistants, your decision probably comes down to ecosystem: Google services users should pick Gemini, heavy coders should try Claude, and everyone else will find ChatGPT the most polished all-rounder for now.

Google described this launch as “the foundation for a truly personal, proactive and powerful desktop assistant.” That’s a promise, not a product. But with Gemini set to power upgraded Siri features in iOS 27 later this year, the Mac app might just be a warmup for something bigger.