OpenAI launched virtual try-on and a saved-products feature for ChatGPT globally on 1 October, extending the assistant’s shopping results into a place to preview clothing and keep track of items. A photograph can be used to visualise clothing or accessories, while a new Favorites library holds products for later reference, TechCrunch reported.
Key points
- A “try on” button appears in ChatGPT shopping results for clothing and accessories.
- Users can also supply a picture of an item, including a screenshot, to request a try-on.
- Favorites saves products in a Library alongside try-on images.
- OpenAI says the features use ChatGPT Images 2.5.
ChatGPT Images 2.5 powers the try-on
The try-on starts with a selfie or full-body photograph. From a shopping result, the new button lets a user ask ChatGPT to visualise how a selected item might look on them. A picture of an item can serve as the starting point instead: someone could upload a screenshot from the web and ask to see that item tried on.
That makes the image the centre of the feature. The supplied photograph provides a view of the person, and the selected clothing or accessory provides what ChatGPT is asked to place in that view. It is closer to holding a garment up against a photograph than putting it on in a fitting room: the result is a picture intended to help with a choice.
Choosing an outfit online could therefore begin with a photograph and an item found in a shopping result. The try-on would make it possible to view the item against that photograph, while the saved product could be returned to later.
OpenAI says the features use its newly launched ChatGPT Images 2.5 model. The company claims the model produces lighting that looks more natural and textures with more detail, follows image-editing instructions more reliably and reduces the wait for generated images. Each of those qualities matters to an image being used to consider an item of clothing, though the try-on described here is a visualisation of a possible appearance.
Favorites puts products in a ChatGPT Library
Favorites gives shopping results a place to go after a search ends. Products saved through the feature sit in a Library within ChatGPT, alongside the user’s try-on images.
OpenAI also says ChatGPT can respond to a description of a desired style by looking for pieces to make up the look. Alternatively, a photograph of a celebrity’s outfit can be used to ask for the purchasable items in it. Both routes start with an idea of what to wear, rather than a product already identified in a shopping result.
Google and Pinterest already serve fashion discovery
The addition of image-led shopping puts ChatGPT into an activity for which Google and Pinterest have long been used: finding inspiration and then looking for something to buy. TechCrunch reported that OpenAI had already changed course on an earlier ChatGPT shopping experiment after its instant-checkout feature performed poorly. The new features centre on finding, viewing and saving products instead.
Another approach is to put suggestions in front of people before they ask for them. The AI startup Instinct has begun pushing product recommendations, an approach that some people considered too much like advertising, according to TechCrunch. ChatGPT’s new try-on begins with an item selected from a result or supplied in an image.
OpenAI enters this part of shopping after Google: the company launched its own virtual try-on feature in 2025.
