On July 14, 2026, Google announced a new home screen for desktop Google Images that arranges images according to users' interests. Images are gathered from across the web and update in real time. Collections of saved images appear as tabs at the top of the gallery, letting users resume browsing from where they left off. Separately, Google is enabling text-to-image generation within AI Overviews in regular Search. Google Images will handle browsing and saving, while AI Overviews handles generation—Google is aligning the entry points for both finding and creating images within Search.
Image Search That Begins Before You Type a Query
Until now, the basic flow of Google Images was straightforward: a user enters a search term, and Google returns relevant images. The new home screen changes that order. Even without a clear phrase in mind, users can browse images matched to their interests, save the ones that catch their eye, and move on to related images from there. It's a design that brings scenarios you'd recognize on sight but struggle to put into words—like scouting travel destinations or redecorating a room—into the search experience itself.
Saving isn't the endpoint of browsing. When a user adds an image to a collection, that collection appears at the top of the gallery, allowing them to return to the same theme later. By cycling through browsing, saving, and revisiting within a single screen, search shifts from a one-off query into an ongoing activity. When Google redesigned desktop Images in 2019, it introduced a side panel that let users compare selected images while keeping them visible. This time, a home screen that arranges images according to interest has been added ahead of that.
This feature requires signing into a Google Account. While Google hasn't listed specific recommendation signals unique to the new gallery, it has explained that across its search services in general, profile information, search history, saved items, and activity on Google's sites and apps may be used for personalized recommendations, depending on settings. Users can separately turn off "Search history" and "Personalized recommendations." The convenience is tied to choices about which activities get saved and fed into recommendations.
From Mobile in 2025 to Desktop Now
This announcement doesn't mark the first appearance of an image discovery feed. On November 12, 2025, Google had already announced an "Images" tab in the Google app for US users, showing interest-matched images daily on both Android and iOS. Users could browse images, save them to collections, and use discovered images as a starting point for further searches. The 2026 change brings this same flow to the desktop home of Google Images.
In the 2026 version, the gallery updates in real time, and saved collections become tabs on the main screen. On a larger screen, users can compare multiple options side by side, making it easier to sustain longer-term research into things like travel, fashion, or interior design. Google has long allowed users to organize images, links, and places into collections through its existing "Saved" feature. What's new here isn't the saving feature itself, but the fact that the path back from saved content to further browsing has been placed front and center in Google Images.
That said, the rollout is limited in scope. The new Google Images home will roll out over the coming weeks to English-language users in the US on desktop. Sign-in is required. At least as of the announcement, Japan and the Japanese language are not part of the initial rollout.
The Pinterest Resemblance Lies in the Save-and-Return Browsing Flow
The underlying concept resembles Pinterest. Pinterest recommends Pins on its home feed based on saves, searches, and browsing history, and lets users fine-tune recommendations by board or area of interest. Google Images similarly surfaces visual candidates before a user types anything into the search box, letting them save images to collections and return to them later. However, Google hasn't stated that saved images directly influence recommendations in the new gallery. What's similar is the screen-level flow of browsing recommended images, saving them, and revisiting them.
On the other hand, the underlying data foundations differ. Pinterest builds its own interest data from Pins, boards, and user behavior. What Google arranges in the new gallery are "images from across the web," backed by its search index and Google Account activity. Google Images has long included pathways from discovered images to original web pages or product information. The bigger story isn't the Pinterest-like appearance, but the fact that a search index is being re-presented as a personalized feed.
This makes it critical for publishing sites to know how many people actually click through from images to the original page. When Google redesigned Images in 2019, it explained that showing brand and price information on product images—along with stock availability and reviews—would make it easier for users to visit publisher and retailer sites. This latest announcement doesn't explain how site names or links will be displayed within the gallery, or how clicks will be measured. Whether a screen designed for continuous browsing increases referral traffic, or whether browsing simply stays contained within Google, remains to be seen once the rollout's display and data become available.
Bringing Nano Banana 2 Lite to AI Overviews
The other update is a feature that generates images from text within AI Overviews. When the desired image doesn't exist on the web, users can describe what they want in words and generate a new image without leaving the search results page. Google is starting with English and will expand over the coming weeks to regions where AI Mode's image generation is already available.
Here too, what's new isn't image generation itself, but the entry point. On June 30, 2026, Google had announced the introduction of Nano Banana 2 Lite into AI Mode and other features. Now, that same generation capability is coming to AI Overviews, which appear in regular search results. Users will be able to switch between Google Images, which finds existing images from the web, and AI Overviews, which generates images that don't yet exist, within the same search experience.
The model being used is Nano Banana 2 Lite, formally named Gemini 3.1 Flash Lite Image. Google describes it as the fastest and lowest-cost model in the Nano Banana lineup. According to developer specifications, it generates a 1K-resolution image in four seconds at a cost of $0.034 per image. These figures don't guarantee the response time or cost of AI Overviews specifically. Still, the choice of Lite—built for speed and high-volume processing—over Nano Banana Pro, which prioritizes quality above all, aligns with a design built to handle large numbers of generation requests on the search results page.
Generated images carry a SynthID watermark, and Google states that AI-generated content can be verified through Search, the Gemini app, and other tools. However, this announcement didn't go into detail about how labels will be displayed on the generation screen, how users will distinguish generated images from original search results, or whether there are limits on how many images can be generated. As image-finding and image-creating features sit side by side, clear labeling that lets users instantly tell the two apart becomes essential.
A $60.399 Billion Search Business and Undisclosed Monetization
According to Alphabet's Q1 2026 earnings, revenue from "Google Search & other" reached $60.399 billion, up 19% year-over-year. Google Images is part of this search business, and the value of turning visual discovery into sustained engagement is significant—rather than waiting for users who already have a specific search term in mind, Google gains touchpoints from the moment someone starts casually browsing outfits or travel destinations.
That said, it's not clear that the new gallery will immediately boost ad revenue. Google Images has long carried product information and shopping ads, but Google hasn't announced any ad formats, ad inventory, or advertiser-facing measurement metrics specific to the new gallery. Whether this new screen will become ad inventory at all remains undetermined. The two outcomes worth watching first are whether improved recommendation accuracy increases browsing and saving, and whether click-throughs to publisher sites are maintained.
The timeline for expansion beyond the US and into other languages hasn't been disclosed either. Before this reaches users in Japan, Google will need to work through regional and language rollout, along with clarity around recommendation settings, pathways back to original sites, and how AI-generated images are displayed. Whether Google Images takes hold as an entry point for browsing images without a search term will ultimately depend less on how skillfully images are arranged, and more on how effectively users are carried from what they find to where they go next.