OpenAI Adds Virtual Try-On and Favorites to ChatGPT Shopping

OpenAI said Thursday it is rolling out two new shopping tools inside ChatGPT worldwide: a virtual try-on feature for clothing and accessories, and a Favorites library for saving products found during a chat session, according to TechCrunch. Both features run on ChatGPT Images 2.5, a newly launched image model that OpenAI says improves lighting realism, texture detail, and instruction-following while cutting generation latency, TechCrunch reported.

What are ChatGPT's new shopping features?

The launch adds a Try On button to ChatGPT's shopping results, letting a user upload a selfie or full-body photo to see how a specific item might look on them, per TechCrunch's reporting on OpenAI's announcement. A second option, Favorites, saves products a user finds during a conversation into a Library inside the app, where they sit alongside any try-on images the user has generated, OpenAI told TechCrunch.

How does the virtual try-on tool work?

A shopper can upload their own photo, or an image of an item such as a screenshot from a retailer's website, and ask ChatGPT to render how the garment would look on them, according to the TechCrunch report. OpenAI is also positioning the assistant as a stylist of sorts: a user can describe a look they want and ask ChatGPT to shop for the individual pieces, or upload a photo of a celebrity's outfit and ask the model to find purchasable versions of those items, TechCrunch wrote.

What is Favorites and how does it work?

Favorites functions as a running list inside ChatGPT where saved products and try-on renders are stored together for later reference, per OpenAI's description to TechCrunch. The company did not disclose a rollout timeline for additional markets or a user count tied to the shopping features in the material reviewed.

Why is OpenAI entering AI shopping now?

The push follows a bumpier run for OpenAI's commerce ambitions. TechCrunch noted the company previously tried an instant-checkout feature inside ChatGPT that "ended up not performing well." The retreat from checkout toward try-on and discovery suggests OpenAI is betting that visual, selfie-based product exploration draws more engagement than one-click purchasing did. The timing also lands as rivals test — and sometimes misjudge — proactive shopping nudges: TechCrunch reported that agentic AI startup Instinct began pushing product recommendations to users, and that "some felt that the proactive recommendations were an overreach, more akin to ads than helpful suggestions."

How does this compare to Google and Pinterest?

TechCrunch framed the move as OpenAI stepping into territory "that Pinterest and Google have dominated in recent years as sources for fashion inspiration and discovery that can convert to sales for online retailers." Google is not a new entrant to the try-on category: the outlet noted Google launched its own virtual try-on tool last year. Whether ChatGPT becomes shoppers' default stop for outfit inspiration, TechCrunch wrote, "remains to be seen."

Selfie-driven image generation is also expanding outside shopping. Apps such as AlterEgo let users turn personal photos into stylized AI renders for non-commerce use cases, a reminder that the underlying selfie-to-image pipeline OpenAI is now applying to retail has broader consumer traction already.

What should shoppers and retailers watch next?

• Whether OpenAI discloses usage or conversion data tied to Try On and Favorites in a future update.
• Whether retailers integrate directly with ChatGPT's shopping results the way they have with Google and Pinterest catalogs.
• Whether OpenAI revisits checkout inside ChatGPT now that try-on and discovery are live.
• Whether Google or Pinterest respond with their own feature changes to the try-on category.

Glossary

Virtual try-on: A feature that uses a user's uploaded photo to generate an image showing how a specific clothing item or accessory would look on them.

ChatGPT Images 2.5: OpenAI's newly launched image-generation model, which the company says improves lighting, texture, and instruction-following while reducing generation delay, according to TechCrunch.

Agentic AI: AI systems designed to take actions on a user's behalf, such as recommending or purchasing products, rather than only answering questions.

The originating report is available from TechCrunch.