OpenAI adds virtual try-ons to ChatGPT's shopping results

Why it matters

OpenAI is shifting ChatGPT shopping toward discovery and visualization after its checkout push struggled. The preview may help shoppers imagine a product, but it cannot verify fit, and OpenAI has shared no conversion or accuracy data.

OpenAI has added virtual try-on images and saved product favorites to ChatGPT's shopping experience, putting the AI assistant closer to the moment when a shopper decides whether an item is worth buying. TechCrunch reported the global rollout on October 1st. Sam Altman co-founded OpenAI in 2015; OpenAI did not identify an individual product lead for the launch.

The feature extends ChatGPT's consumer shopping experience to product visualization and saved items. Shoppers can ask for suggestions, preview an eligible item on themselves and keep products they may want to revisit in the same conversation.

A picture, not a fitting

In ChatGPT shopping results, eligible clothing and accessory listings have a "Try on" button. A user can upload a selfie or full-body image to generate a preview, or upload a picture of an item and ask ChatGPT to show how it might look on them. The OpenAI help page says users can save products to Favorites or sort finds into folders in their ChatGPT Library.

Those images are visualizations, not fit assessments. OpenAI cautions that a try-on may not accurately represent the product or the user's appearance and does not guarantee fit or size. Shoppers still need the merchant's measurements and product details to judge whether a garment is likely to fit. The system can make a product easier to imagine on a person; it does not establish how the fabric drapes, how a size will feel or whether the item matches its listing exactly.

The feature uses ChatGPT Images 2.5, which OpenAI introduced on September 8th. OpenAI says the model produces more natural lighting and richer textures, follows editing instructions more reliably and reduces image-generation latency by up to 50% compared with Images 2.0. Those are OpenAI's claims about the image model; the company has published no accuracy results for clothing previews. OpenAI has not provided try-on benchmarks for garment fidelity, body representation or sizing.

Users' reference photos are saved for future try-ons, according to the help page, and can be changed or deleted in Settings. That makes photo handling part of the product experience, rather than a one-time upload. The documented controls cover managing the saved reference image; they do not make a generated preview a reliable guide to fit.

Discovery gets another chance

The launch follows a change in OpenAI's approach to shopping. TechCrunch reported that OpenAI's earlier instant-checkout feature did not perform well and that the company pivoted from that effort. The new features emphasize browsing, visualizing and saving products, with ChatGPT helping shape what a user considers before the merchant handles the transaction.

Google introduced a clothing try-on feature in 2025, and TechRadar reported that Perplexity added try-on to its AI shopping experience in 2025. ASOS and AIUTA launched a try-on system tied to ASOS's catalog. ChatGPT's route is distribution through a general-purpose assistant: users can start with a style description, an item photo or an existing product listing instead of opening a dedicated fashion tool.

That conversational route could help users move from an idea to a shortlist. TechCrunch reported that ChatGPT can suggest outfits based on a style description or identify purchasable pieces in a celebrity outfit image. Whether that discovery brings shoppers to products they actually buy depends on catalog coverage and the relevance of the listings. OpenAI has not disclosed how many products support try-on, which retailers are eligible or whether the feature changes shopping conversion.

OpenAI's shopping results link users to merchants, and the help center says product results are selected independently by ChatGPT, not as ads or through partnerships. Users need to know whether suggestions reflect their request or a commercial placement as shopping becomes more prominent inside the assistant. Favorites gives users a way to keep track of what they found, but OpenAI has not reported usage or purchase data for the new tools.

For Altman and OpenAI, the product move expands what ChatGPT is asked to do: move from answering questions about products toward helping people picture, organize and compare them. Virtual try-on gives the assistant a visual step in that process. Its credibility will depend on keeping the promise narrow: a useful preview of appearance, with fit and product accuracy still left to measurements, listings and the retailer.