Loading page…
The global rollout pairs personalized outfit previews with a shopping library. The image model’s realism claims do not settle how much confidence shoppers should place in a rendered look.
Listen to this story
The rollout extends ChatGPT’s shopping role from product discovery toward visual evaluation and saved consideration: users can preview items against their own photo and keep listings and generated previews in an in-app Library. OpenAI began the global rollout on October 1, using ChatGPT Images 2.5, released September 8; the model’s claimed up-to-50% latency reduction is versus Images 2.0, not a promised try-on speed. The tools focus on the pre-purchase stage after a reported retreat from weak-performing instant checkout, but whether shoppers trust generated previews remains unknown.
A Try On button appears in shopping results, and users can also upload an item image, including a web screenshot, to request a preview.
Favorites stores discovered products alongside generated try-on images in an app Library, preserving both the item and its visualization.
Users can request products based on a style description or upload a celebrity outfit photo to find pieces available to buy.
ChatGPT shoppers can now preview clothing and accessories on a photo of themselves, then save products for another look later. OpenAI announced the global rollout of Virtual Try On and Favorites on October 1, adding a visual decision step to the assistant’s shopping results.
The two features handle different parts of browsing. Virtual Try On generates an image showing how an item might look on the person shopping. Favorites keeps discovered products available for later reference. Together, they let users experiment with a look and retain a shortlist inside ChatGPT, rather than treating each shopping conversation as a one-off search.
The starting point is a selfie or full-body photo. Users upload that reference to visualize clothing or accessories on themselves. A new Try On button appears in ChatGPT’s shopping results, giving the preview a place alongside the products the assistant has found. TechCrunch describes this as a way to see how an article might look, rather than simply viewing its product image.
There is also a route for something found elsewhere. Users can upload an image of an item, including a screenshot from the web, and ask ChatGPT to try it on. That means the workflow does not have to begin with a product surfaced in ChatGPT’s own shopping results. The item image supplies the clothing reference; the user’s photo supplies the person for the visualization.
Favorites adds a Library in the app where users can save products they discover. OpenAI says those items will sit alongside their try-on images. The distinction is useful: saving a product preserves the shopping find, while keeping the generated image preserves the visual experiment. Users can return to both later, instead of having to rediscover an item or recreate its preview before reconsidering a purchase.
OpenAI also describes shopping paths that start with an idea rather than a particular item. A user can describe a style and ask ChatGPT to find the pieces needed to complete it. Alternatively, they can upload photos of celebrities’ outfits and ask for the items available to buy. Those examples explain the discovery side of the workflow: the assistant can help translate a visual reference or style description into products.
OpenAI says the shopping features use ChatGPT Images 2.5. The company introduced that model on September 8, so October’s news is the shopping rollout, not the model’s debut. In its earlier announcement, OpenAI said Images 2.5 better preserves subjects from reference photos, produces more natural lighting and richer textures, and follows editing instructions more reliably across repeated changes.
The company also claims up to 50% lower image-generation latency than Images 2.0—meaning less waiting for an image. That figure describes the model comparison, not a guaranteed try-on completion time. The quality improvements are company claims, too. More natural lighting is an image-quality claim, not a stated guarantee that a garment will fit. The announced shopping use is a visualization of appearance.
This is not OpenAI’s first attempt to make ChatGPT useful for online shopping. TechCrunch reports that the company previously pivoted away from an instant checkout feature after it performed poorly. The new tools address a different part of the process: looking at an item and keeping it under consideration. They add help before a purchase decision, rather than making checkout the central attraction.
OpenAI is also entering an established category. Digital Trends notes that Google shipped a similar virtual try-on feature in July 2025. ChatGPT’s addition therefore brings a familiar shopping function into a conversational assistant; it does not introduce virtual try-on itself. Whether shoppers choose ChatGPT for this task remains unresolved, as does how much weight they will give an AI-generated version of themselves when deciding what to buy.
Loading discussion...
Join the conversation
Explain what a preview could settle—and what you would still need to know.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.