The Maze: OpenAI added Virtual Try On and Favorites to ChatGPT on October 2. Shoppers can preview clothes or accessories using a photo of themselves, then save products and try-on images for later. That puts two familiar parts of fashion shopping—imagining an outfit and keeping a shortlist—inside the conversation. The described journey still ends on merchants’ sites. For retailers, the question is whether the product earns a place in that shortlist before the shopper ever reaches a store page. No sales uplift has been established.
The fitting-room idea starts with a photo. A shopper uploads a selfie or full-body image, then taps the try-on control in shopping results. They can also upload a picture of an item and ask ChatGPT to show it on them. The launch example pairs a shopper photo with a screenshot of a wedding-guest dress. This gives someone another way to judge a style before visiting a merchant, rather than asking them to imagine the item from a standalone catalogue image. It is visual evaluation, not evidence that a size will fit. The retailer still supplies the product that eventually arrives.
Favorites makes the shortlist persist. Products discovered in ChatGPT can be saved to Library and revisited alongside try-on images. A shopper can return to a dress they considered earlier rather than start the search again. Combining an item with its appearance preview also keeps the reason for considering it close to the saved choice. The commercial inference is straightforward: some repeat consideration can happen inside ChatGPT. A saved product is still only a candidate purchase. The launch supplies no evidence that these lists increase conversion or reduce the number of retailer visits.
The image engine explains the preview, not the fit. The new shopping controls build on ChatGPT Images 2.5, introduced in September. Its reference-photo editing gives OpenAI a way to combine a person and an outfit. But the try-on documentation warns that the generated result may misrepresent the product or appearance and guarantees neither fit nor size. That distinction matters for apparel sellers: an appealing preview cannot substitute for measurements, material details or a return policy. A confidence boost is plausible; fewer returns remain an untested claim.
The merchant still has to close the sale. After reviewing saved choices, the launch journey sends shoppers to merchants’ sites to buy. The two new features extend product discovery and evaluation; they are not presented as a fresh checkout announcement. For an operator, it makes sense to check whether product images and descriptions let a shopper understand what they are saving, then whether the store page resolves the remaining buying questions. That is an implication of this flow, not a new OpenAI merchant requirement. The announcement gives no feature-specific fee, adoption figure or regional rollout matrix.
Why it matters: ChatGPT is adding reasons to stay during the consideration stage and reasons to come back to a saved item. That could make the visit to a merchant more focused, while moving some early persuasion outside the store’s own pages. Retailers should watch the quality and intent of those visits rather than assume the new controls automatically create sales. The useful test is what happens after a preview: does the shopper find clear size, stock and return information, and then buy? A convincing image alone cannot answer that.
