SDSignal Desk

ChatGPT adds virtual try-on and saved Favorites in OpenAI's latest push into shopping

· Updated · Signal Desk newsroom

Analysis of reporting by TechCrunch. How we work

Image: TechCrunch

OpenAI launched two ChatGPT shopping features globally: a Try On button that renders clothing and accessories on a user's photo, and Favorites that save products and try-on images to a Library. Both rely on the new ChatGPT Images 2.5 model, and they follow an Instant Checkout feature that TechCrunch says underperformed.

Key takeaways

  • Try On works from a selfie or full-body photo, or an uploaded screenshot of an item.
  • Favorites save products and try-on images to the ChatGPT Library.
  • The features use ChatGPT Images 2.5, which OpenAI says improves lighting, textures and speed.
  • OpenAI also pitches style prompts and celebrity-outfit lookups linked to purchases.
  • Google launched virtual try-on last year, and Instant Checkout reportedly underperformed.

What launched

TechCrunch reports that OpenAI has launched two shopping features in ChatGPT worldwide. The first is virtual try-on: a Try On button lets users see clothing and accessories on themselves using a selfie or full-body photo, or by uploading a screenshot of an item they found elsewhere. The second is Favorites, which saves products, along with any try-on images, to a Library inside ChatGPT.

Both features depend on ChatGPT Images 2.5, a new image model that OpenAI says improves lighting, textures, adherence to edit instructions and latency. OpenAI is also promoting style prompts and lookups of celebrity outfits that convert into purchase links.

Competitive context

This is crowded territory. Google launched its own virtual try-on last year, and Pinterest has long served fashion discovery. TechCrunch notes that OpenAI's earlier Instant Checkout feature underperformed, which makes this launch look like a search for a consumer commerce experience that sticks. The company has also been expanding ChatGPT's plug-ins with app-like interfaces, which could eventually connect discovery to retailer experiences.

Retailers are experimenting from the other side as well; Albertsons recently described reimagining retail with AI. For brands, ChatGPT becomes another referral surface to integrate with or compete against, alongside Google's try-on and shopping graph.

Questions about accuracy and likeness

Virtual try-on is most useful when it helps shoppers judge fit and style and reduces returns. It is least useful when a generated image flatters the fit or misrepresents fabric. OpenAI's claims about Images 2.5 are company statements; independent testing of how faithfully it renders size and drape will matter. Shoppers should still check sizing charts and return policies on retailer sites.

Celebrity outfit lookups raise separate questions about likeness and endorsement, and saved Favorites could become a source of preference data. TechCrunch's report does not indicate that Favorites is tied to advertising, but OpenAI has begun testing ads in ChatGPT, so how shopping data is used is worth watching. More in our OpenAI coverage and generative AI coverage.

Who feels it

Shoppers
Faster visualization and saved shortlists, with sizing still to verify at the retailer.
Retailers
Another discovery surface beside Google's try-on and Pinterest.
OpenAI
Needs commerce features that convert after the Instant Checkout miss.

What to watch

  1. Accuracy complaints and any return-rate evidence.
  2. Whether Favorites becomes an advertising surface.
  3. Pushback on celebrity outfit features.

FAQ

What does Try On do?

It renders clothing or accessories onto a user's selfie or full-body photo.

What are Favorites?

Saved products and try-on images stored in the ChatGPT Library.

Which model powers it?

ChatGPT Images 2.5.

Is this new for the market?

No. Google launched virtual try-on last year.

Read the original

Continue at the source.

TechCrunch

Companies: OpenAI

More on OpenAI

All OpenAI coverage →