Google today made Gemini available as an installable program for PCs running Windows 10 and Windows 11, worldwide, according to a post on the company's news site credited to Erin Pettigrew, Director of Product Management for the Gemini App. The Windows client appears over whichever window is active when a key combination is pressed, draws on Gmail and Google Drive, hands multi-step jobs to the Gemini Spark agent and generates images and video, according to Google.
In Short
Google now has a Gemini program you can install on a Windows computer, and it pops up on top of whatever you are doing when you press two keys together. Lots of office and agency work happens on Windows PCs, and Microsoft already places its own assistant, Copilot, on those same machines. Some of the features Google lists, such as the Spark agent, have come with age, subscription and country limits on other platforms, and the post does not say whether those limits follow them to Windows.
A short post in three parts
Google's page gives the release a one-minute reading time and an audio version lasting two minutes and one second. Its substance fits under three numbered headings, followed by a closing paragraph on performance and availability. The download sits at gemini.google/desktop, according to the post, an address distinct from the gemini.google/mac domain Google uses for its macOS client.
The key combination
Pressing Alt and the space bar at any time brings Gemini up over active work, according to Google. The post offers two examples of what that is for: a quick fact-check on a document, and a few title ideas for a presentation, after which the user goes back to the task at hand.
The choice of keys carries some history on Windows. Long before any assistant claimed it, that pairing brought up the window menu, the small list holding the minimise, maximise and close commands. Windows Latest reported in January 2025 that Microsoft's Copilot app on Windows 11 had begun promoting the same combination as its default trigger, and a May 2026 guide by technology writer Dave Taylor described Copilot and OpenAI's ChatGPT client contending for it on a single machine. Gemini now joins that queue. Google's post does not say how the program behaves when another assistant already holds the shortcut, or whether a different combination can be assigned.
The Mac version offers more ways in. Google's German-language page for Gemini on macOS promoted an Option and space bar shortcut alongside a Command-Command gesture that sends on-screen content straight into a chat, and on July 29, 2026 Google added long-press Fn voice dictation and screen-aware editing for Mac users, in English only. Neither voice input nor a screen-capture gesture is mentioned in the Windows material.
Spark, Gmail and Drive in one window
Inside the application, users can reach everything they already use Gemini for, according to Google. Two functions are singled out. Multi-step tasks can be passed to Gemini Spark, which the post calls a "24/7 personal AI agent", and Gemini can draft a project summary by pulling information directly from Gmail and Google Drive.
Spark is Google's most visible consumer product in agentic AI, and its record is short. The agent debuted at Google I/O on May 19, 2026, built on Gemini 3.5 Flash with Model Context Protocol connections to Canva, OpenTable and Instacart, and limited to the United States. On June 30, 2026, Google extended Spark to macOS in beta, restricting it to Google AI Ultra subscribers aged 18 and over in the United States. Ultra costs $99.99 a month. Separately, thirteen further connected apps for Gemini were unveiled at Made by Google on August 12, 2026, spanning productivity, local, entertainment, music, home, health and lifestyle categories.
Images and video
The third heading covers creative work. Users can generate custom images with Nano Banana for a presentation or direct "high-quality video" with Gemini Omni, according to Google, which stresses that both run from the same place.
Neither model is new. Nano Banana first appeared in the Gemini app in August 2025 and had produced more than 5 billion images by October 13, 2025, when it spread to Search and NotebookLM. A successor, Nano Banana Pro, built on Gemini 3 Pro, reached advertisers inside Google Ads on November 20, 2025. Gemini Omni was unveiled at Google I/O 2026, where Google described it as combining Gemini reasoning with generative media, and the following day Google said it would bring the model into Asset Studio, the creative hub inside Google Ads.
Omni's consumer deployments have not been uniform. In Google Vids, Omni editing arrived on July 16, 2026 for paying subscribers only, and the Shorts remix feature built on Omni excluded the European Union and the United Kingdom when it was shown in May. The same image and video systems that advertisers meet in campaign tools are now reachable from a Windows desktop without a browser tab. Who can actually use them from that desktop is a separate matter.
What the post leaves unsaid
The closing paragraph describes the program as "Lightweight and quiet" and says it runs without slowing down a PC. No figures accompany the claim. There is no installation size, no memory or processor requirement, no distinction between x64 and Arm-based Windows machines and no statement on whether any processing happens locally. That last question has precedent. A privacy researcher documented in May 2026 that Chrome writes a 4 GB Gemini Nano model file to eligible Windows, macOS and Linux devices without asking, a practice he argued breaches European privacy law.
Eligibility is the larger gap. The post places a superscript marker beside both Gemini Spark and Gemini Omni, pointing to a "More Information" panel that is collapsed on the published page, and its contents are not visible in the captured version. What is on record comes from earlier documentation. Spark support pages reviewed in July 2026 required users to be 18 or older, to sign in with a personal Google Account rather than a work or school account, to hold a Google AI Pro or Ultra subscription and to have Keep Activity enabled. Pro is priced at $19.99 a month. The same pages stated that Spark was unavailable in the EEA, Nigeria, Switzerland and the United Kingdom, and that it worked only through the Gemini mobile app, the Mac app and the web app at gemini.google.com.
That documentation predates the Windows client, so it could not list it. Whether a Windows user in Frankfurt, Zurich or London will find Spark inside the new program is not addressed. The global availability in the post applies to the application. It does not claim that every feature inside the application is global.
The work-account condition also sits awkwardly beside the post's own examples. Checking documents, drafting project summaries and producing presentation imagery are office tasks, yet Spark, according to its July documentation, did not accept work or school accounts. A similar split applies elsewhere in Gemini: Business Profile management remained tied to personal Gmail logins in August 2026, leaving agencies that work through Workspace identities outside it.
One smaller inconsistency surfaced in the source material. Press material circulated alongside the post phrased the Windows availability in the future tense, while the post itself states the program is available today. The published post is the more authoritative of the two.
Mac first, Windows second
Gemini has been on Windows desktops for some time through the browser. Google put Gemini into Chrome for Mac and Windows desktop users on September 18, 2025, starting in the United States with English language settings, and on July 14, 2026 it added Chrome auto browse for MacOS, Windows and Chromebook Plus users, letting an agent complete tasks such as booking travel and filling in forms.
The standalone Gemini client followed a different order. The Mac received Spark in beta on June 30 and voice dictation on July 29. Windows receives its program 72 days after the Spark beta reached macOS, and without any of the Mac-specific input methods described in July. Google gives no schedule for closing the gap. "This is just the beginning for the Gemini app on Windows, with more native desktop capabilities rolling out over time," Pettigrew wrote.
An assistant on Microsoft's operating system
Google is placing its assistant on an operating system whose owner bundles a rival. Distribution alone has not settled that contest. Similarweb data covering twelve months to May 2026 showed Microsoft Copilot holding between 1.8% and 2.0% of generative AI website traffic throughout the period, despite its integration into Windows and Office. Gemini's share over the same window rose from 8.9% to 27.3%, while ChatGPT's fell from 76.4% to 52.7%.
The operating system's defaults remain contested ground. A Mozilla-published study found that Microsoft continues to use interface patterns across Windows 10 and Windows 11 that influence which browser users end up on, with new findings covering how Copilot handles links inside its app. Gemini enters that environment as a third-party program, subject to the same defaults as everything else.
How much desktop activity is at stake? Less than the attention suggests. Clickstream data from Datos placed AI tools below 2% of total desktop web visits in both the United States and Europe in the first quarter of 2026. The US share rose from 1.31% in the first quarter of 2025 to 1.65%, and the EU and UK figure climbed from 0.98% to 1.72% over the same comparison.
A measurement blind spot
Those figures describe web visits. An installed Windows program is not a browser visit to gemini.google.com, so desktop sessions that move from a browser tab into the client fall outside what web traffic panels are built to count. Similarweb's traffic share data, as described in January 2026 coverage of the chatbot market, excluded API usage and mobile app activity except where it showed up in referral traffic. If a meaningful share of Gemini's desktop use migrates into the Windows client, panel-based share figures would capture less of it, which is not the same as the use declining.
The referral side has its own history. Native apps have produced attribution gaps before: the Gemini iOS app was found in October 2025 to identify itself with an undocumented "GeminiiOS" user agent string. Google Analytics added a dedicated AI Assistant channel on May 13, 2026, classifying sessions whose referrer matches a recognised assistant such as Gemini. That classification depends on a referrer arriving at all. Google's own products have lost it before: in AI Mode, noreferrer elements in link code stripped referrer values until Google confirmed the problem on May 23, 2025, and the affected clicks registered as direct traffic. AI Mode is a search product rather than the Gemini app, but the episode shows how a single link attribute can move assistant traffic into the direct bucket.
The Windows post does not describe how links inside Gemini responses open, whether in the default browser or an embedded view, or what referrer information travels with them.
Scale without advertising
The installed base the Windows client draws on is large. Alphabet reported 750 million monthly active users for the Gemini app on February 4, 2026, and the figure crossed 900 million in May 2026. By the second quarter it had reached 950 million, with daily actives tripling year over year, according to Alphabet's second-quarter 2026 results, which also put Gemini model throughput at approximately 22 billion tokens per minute across first-party APIs.
None of that audience sees advertising inside the Gemini app. Dan Taylor, Google's vice president of global ads, publicly denied on December 8, 2025 that ads were coming to the Gemini app, and repeated in January 2026 that there were no plans for them, drawing a line between the assistant and AI Mode, where paid formats run. AI Mode passed one billion monthly active users and now carries sponsored placements that Alphabet executives cited as revenue drivers in July.
OpenAI has taken the other route on its own assistant. ChatGPT's advertising business reached a $1 billion annualised revenue run rate in under 200 days, according to a Digiday report on August 31, 2026, and Sensor Tower estimated that roughly 23% of ChatGPT ads placed on desktop in the United States between August 15 and August 30, 2026 were carousels. Desktop assistant surfaces already carry inventory at OpenAI.
Google's Windows client extends a surface the company has said carries no ads. The post, focused on keyboard shortcuts and creative tools, does not address monetisation, and the question of whether an assistant with 950 million monthly users and a new desktop footprint stays commercial-free remains Google's to answer.
Timeline
- August 2025: Nano Banana image model debuts in the Gemini app
- September 18, 2025: Gemini arrives in Chrome for Mac and Windows desktop users in the United States
- October 13, 2025: Nano Banana expands to Search and NotebookLM after more than 5 billion images
- October 2025: Gemini iOS app found using an undocumented "GeminiiOS" user agent string
- November 20, 2025: Nano Banana Pro, built on Gemini 3 Pro, reaches advertisers in Google Ads
- December 8, 2025: Dan Taylor denies plans for ads in the Gemini app
- February 4, 2026: Alphabet reports 750 million monthly active users for the Gemini app
- April 27, 2026: Datos places AI tools below 2% of desktop web visits in the US and Europe
- May 13, 2026: Google Analytics adds an AI Assistant default channel
- May 19, 2026: Gemini Spark and Gemini Omni unveiled at Google I/O 2026
- May 20, 2026: Google confirms Gemini Omni for Asset Studio at Google Marketing Live
- May 2026: Gemini app crosses 900 million monthly active users
- June 11, 2026: Similarweb data shows Copilot at 2.0% of generative AI web traffic, Gemini at 27.3%
- June 30, 2026: Gemini Spark reaches macOS in beta for US Ultra subscribers
- July 14, 2026: Chrome auto browse arrives for MacOS, Windows and Chromebook Plus
- July 14, 2026: Mozilla-published study documents browser-choice interface patterns in Windows 10 and 11
- July 16, 2026: Gemini Omni editing reaches Google Vids for paying subscribers only
- July 22, 2026: Alphabet reports 950 million monthly active users for the Gemini app
- July 29, 2026: Mac users get long-press Fn voice dictation in Gemini, English only
- July 30, 2026: Spark documentation lists the EEA, Nigeria, Switzerland and the UK as unavailable markets
- August 12, 2026: Google adds thirteen connected apps to Gemini at Made by Google
- August 31, 2026: ChatGPT advertising reaches a $1 billion annualised run rate
- September 10, 2026: Google makes the Gemini app available for Windows 10 and 11 worldwide
Related PPC Land coverage
- Google gives Gemini for Mac voice control over any open app: The July 29, 2026 Mac update adding Fn-key dictation, screen-aware editing and image generation in English.
- Google restricts new Mac AI agent to $99.99 Ultra subscribers: How Gemini Spark reached macOS in beta on June 30, 2026 behind age, country and subscription gates.
- Gemini Spark blocks EU and UK users as Google adds 160 countries: The support documentation listing Spark's account, subscription and geographic requirements.
- Google adds Granola to Gemini 12 days after it was sued over secret recording: The August 2026 connected-app expansion and the sequence of Spark releases since May.
- Google Vids gains Gemini Omni editing and avatars for paid users only: Gemini Omni's arrival in Workspace video tools with subscription and regional limits.
- Chrome auto browse lets AI agents book travel and fill forms alone: The July 2026 Chrome update bringing agentic browsing to Windows, MacOS and Chromebook Plus.
- ChatGPT drops to 52.7% as Claude triples its AI traffic share: Twelve months of Similarweb share data, including Copilot's flat position despite Windows distribution.
- Microsoft blocks browser choice for 1.4 billion Windows users, study finds: Research on interface patterns in Windows 10 and 11, including Copilot's in-app link handling.
- Gemini iOS app traffic revealed through undocumented user agent: An earlier case of a native Gemini client complicating referral attribution.
- Google Analytics adds AI assistant channel for ChatGPT, Gemini, Claude: The May 2026 default channel that classifies assistant referrals.
- Why Google won't put ads in Gemini - VP explains the strategy gap: Google's stated separation between the ad-free Gemini app and monetised AI Mode.
- ChatGPT ads pass $1bn run rate as financial services spend triples: OpenAI's advertising growth and desktop carousel adoption through August 2026.
- Researcher documents Chrome's silent 4 GB Gemini Nano download: The on-device model file disclosure and the European consent questions it raised.
Summary
Who: Google, through Erin Pettigrew, Director of Product Management for the Gemini App. The release affects Windows 10 and Windows 11 users worldwide, Microsoft as the owner of the operating system and of the competing Copilot assistant, and publishers and advertisers who measure traffic and attention from AI assistants.
What: A standalone Gemini program for Windows that opens over the active window with the Alt and space bar keys, provides access to Gemini Spark and to information from Gmail and Google Drive, and generates images with Nano Banana and video with Gemini Omni. The post gives no system requirements, pricing, language details or feature-level eligibility, and attaches collapsed footnotes to Spark and Omni.
When: Today, September 10, 2026, with further native desktop capabilities described as rolling out over time and no dates attached.
Where: Globally, for PCs running Windows 10 and Windows 11, downloadable from gemini.google/desktop. Earlier Spark documentation listed the EEA, Nigeria, Switzerland and the United Kingdom as unavailable markets, and the post does not say whether that changes on Windows.
Why: The release moves an assistant with 950 million monthly active users onto Windows PCs alongside Microsoft's own Copilot, while shifting some desktop activity into a client that web traffic panels are not built to count and whose referral behaviour Google has not documented.
Discussion