Google released the Gemini desktop app for Windows on 10 September 2026, and it opens with Alt + Space. That is the same global hotkey OpenAI's ChatGPT Windows app has used for its companion window since launch. If you already run both, the second one you install is going to fight the first one for a keystroke you use all day.
The announcement on Google's blog is short and leads with convenience: "Press Alt + Space on your PC at any time to open Gemini over your active work." Underneath that sentence is the part that actually matters to anyone who makes things for a living. The overlay carries Nano Banana for images and Gemini Omni for video, which means text, image and video generation are now one keystroke away from Photoshop, Premiere, PowerPoint or whatever else happens to be in front of you. The download page puts the requirement at Windows 10 or later on both x64 and ARM64.
Almost every outlet covered this as "Google brings Gemini to Windows." That framing misses both the collision and the catch. The collision is the hotkey. The catch is a one-line footnote at the bottom of Google's own post: "Google AI subscription required. Availability varies. 18+." The app is free and global. The three features worth installing it for are not.
What Google Shipped
The Windows client is a native overlay rather than a browser tab. Google describes it as "lightweight and quiet," and the pitch is that it sits on top of your active window instead of pulling you out of it. Google's verbatim framing is "available today globally for Windows 10 and 11," while the download page is more precise and says Windows 10 or later, x64 and ARM64.
Three capabilities ride along inside the overlay, and each maps to something a creator already does:
Gemini Spark, described by Google as "your 24/7 personal AI agent," takes multi-step handoffs. We covered what Spark actually does when it landed inside Google Photos, and the desktop version is the same agent with a wider surface.
Nano Banana generates images. Google's example is making a custom image for a presentation, which undersells it: this is the same image model that has been driving generation inside Google Earth.
Gemini Omni handles video. Google's phrasing is "direct a high-quality video with Gemini Omni," and Omni 1.1 Flash is the current generation of that model.
Context comes from your Google account. Google's example is asking Gemini to "draft a project summary by pulling information directly from your Google apps like Gmail and Google Drive." A side panel exposes Notebooks and recent conversations. Google says more native desktop capabilities are rolling out over time, which is the usual signal that the first release is a floor rather than a ceiling.
Google Took the Hotkey ChatGPT Already Had
This is the part no coverage led with, and it is the thing that will bite people in the first ten minutes.
OpenAI's ChatGPT app for Windows binds its companion window to Alt + Space. A current shortcut reference for the ChatGPT desktop apps lists it plainly: "Open ChatGPT companion window: Alt Space" on Windows, customizable under Settings, App. Google has now shipped a second global overlay on the identical binding, with no acknowledgement anywhere in the launch post that the key was already spoken for.
It is not a Windows-only accident either. On macOS, Google's download page says "press Option+Space to open the mini chat, or Option+Shift+Space for the full chat experience." OpenAI's macOS chat bar has been on Option + Space since companion mode arrived in August 2024. Google picked the incumbent's hotkey on both platforms. The Mac client shipped in April 2026, so Windows users are getting the collision five months after Mac users did.
Read charitably, Alt + Space and Option + Space are simply where a global assistant belongs, and two companies reached the same conclusion. Read practically, it does not matter why. If you use ChatGPT for drafting and Gemini for image work, which is an extremely common split right now, you are picking one to demote on day one.
Three Assistants, One Windows Desktop
Windows now has three AI assistants competing for the same reflex. Here is what each one actually gives you at the keystroke level.
| Assistant | Global hotkey on Windows | Image generation | Video generation | Where context comes from |
|---|---|---|---|---|
| Gemini app (new) | Alt + Space | Yes, Nano Banana | Yes, Gemini Omni | Gmail, Google Drive, Notebooks |
| ChatGPT app | Alt + Space (rebindable in Settings, App) | Yes | Not in the companion window | Chat history, connected apps |
| Microsoft Copilot | Win + C, plus a dedicated Copilot key on newer hardware | Yes | No | Windows and Microsoft 365 |
The column that decides this for creative work is video. Copilot has no video generation at the overlay. ChatGPT's companion window is a text and image surface. Gemini is the only one of the three that will produce a video clip from a floating panel on top of your editing timeline. Whether the clip is good enough to use is a separate argument, and we ran that comparison when we put Omni against Veo, Sora, Runway and Kling.
Copilot's position is worth noting because Microsoft has been quietly retreating from it. The Copilot key documentation now describes remapping that dedicated key to Right Ctrl or the Context Menu key, which is an unusual thing to document for a button you shipped on millions of keyboards. The assistant that owns the operating system is loosening its grip on the hardware key at the same moment Google is claiming a software one.
What Is Free and What Is Not
Google's footnote does a lot of work: "Google AI subscription required. Availability varies. 18+." The download page repeats it as "certain features are available to users 18 and older with a Google AI subscription."
Neither sentence enumerates which features. Reading it against the body of the announcement, the three capabilities Google leads with are Spark, Nano Banana and Gemini Omni, and those are the same three that sit behind Google's paid consumer tiers elsewhere. The honest summary is that the app is free, global and available in every language the Gemini app supports, while the reasons a creator would install it are gated, and the gating varies by country.
We are deliberately not printing tier prices here. The subscription lineup has moved more than once this year, third-party trackers disagree with each other on what sits in which tier, and Google's post does not state a price at all. Google's own subscriptions page is the only source worth trusting on that, and it is the one to check before assuming Spark or Omni is included in what you already pay for.
How to Set This Up Without Breaking Your Existing Hotkey
Thirty minutes, and you will know whether this earns a place in your workflow.
1. Decide which assistant owns Alt + Space before you install. Whichever one you reach for without thinking should keep the default. Do this first, because discovering the conflict after installation means diagnosing it while a hotkey misbehaves.
2. If ChatGPT loses, rebind it rather than Gemini. ChatGPT exposes the companion window hotkey as a configurable setting under Settings, App. Google's launch post documents no equivalent control, so the app with the adjustable binding is the one to move.
3. Install from Google's page, not a search result. The client is at gemini.google/desktop. Confirm your machine matches: Windows 10 or later, x64 or ARM64.
4. Test the overlay against a real application, not a blank desktop. Open the tool you actually work in, trigger the hotkey, and watch whether the panel steals focus or holds it. An overlay that yanks focus out of a timeline is worse than a browser tab, and you will only find that out against real software.
5. Run one image and one video generation before you judge it. The Nano Banana and Gemini Omni paths are the only things the Windows app does that a browser tab does not do more comfortably. If both are gated on your account or in your country, the app is a hotkey attached to a chat window, and you can stop there.
Frequently asked questions
Does the Gemini Windows app conflict with the ChatGPT desktop app?
Yes, on the default binding. Both use Alt + Space as their global overlay shortcut on Windows. Only one process can own a global hotkey at a time, so whichever app claims it will leave the other unresponsive to that keystroke. ChatGPT's hotkey is configurable in its app settings, so that is the practical one to move.
Do I need to pay to use the Gemini app on Windows?
The app itself is free and available globally in every language the Gemini app supports. Google's own footnote states that a Google AI subscription is required for some features and that availability varies by region, with an 18+ age requirement attached. Google does not list which features in the launch post, so check your subscription before assuming Spark, Nano Banana or Gemini Omni is included.
Can it actually generate video from the desktop overlay?
Google's post says you can "direct a high-quality video with Gemini Omni" from the app. That makes it the only one of the three major Windows assistants with a video path at the overlay, since Copilot has none and ChatGPT's companion window is text and images. Note that some coverage misreported the video model as Gemini Nano; Google's announcement names Gemini Omni.
Is there a Mac version, and is it the same?
Yes. The macOS client shipped in April 2026, five months before Windows. It requires an Apple Silicon MacBook running macOS Sequoia 15.0 or later, and it uses Option + Space for a mini chat and Option + Shift + Space for the full experience. That mini chat and full chat split does not exist on the Windows client.
What are the exact system requirements?
Google's blog post says Windows 10 and 11. The download page is more specific and says Windows 10 or later on x64 and ARM64, which means ARM-based Windows machines are supported rather than left out. Where the two disagree, the download page is the one to trust.
Is this different from Gemini in Chrome or the Gemini website?
The meaningful difference is the global hotkey and the fact that the panel floats above other applications rather than living in a browser window. The models are the same ones you would reach on the web. If you work primarily inside a browser, the desktop client adds a keystroke and little else. If you work inside desktop applications like Premiere, Photoshop or PowerPoint, the overlay is the entire point.
