ChatGPT's virtual try-on feature went live October 1: tap "Try On" on a supported clothing listing, upload a selfie once, and Images 2.5 generates a preview of how it looks on your own body. A paired Favorites list saves products to a folder instead of losing them in scroll history. For D2C apparel and accessories brands, this is the clearest sign yet that ChatGPT is becoming a discovery channel, not a checkout one.
What does the ChatGPT virtual try-on feature actually do?#
OpenAI's own release notes confirm it: a "Try On" button now shows up on supported clothing and accessory listings inside ChatGPT, on mobile and web, first reported in detail by TechRepublic. Here's what that means in practice:
- A "Try On" button appears on eligible clothing and accessory listings
- First use asks for a selfie or full-body photo, saved as a reusable reference photo
- The preview itself is generated by Images 2.5, OpenAI's newer, sharper image model
- A paired Favorites feature saves products straight to your ChatGPT Library or a folder
- Reference-photo storage and model training both have an opt-out, tucked under Settings, Personalization
It launched live in Canada first. Android Authority reports the US rollout is presumed to follow in the same window, which is OpenAI's usual pattern for ChatGPT feature launches.
Does this change anything for your brand, or can you ignore it?#
If you sell software, services, or anything that isn't worn on a body, skip this one. Try-on only touches "supported" clothing and accessory listings, and OpenAI's release notes don't say how a merchant gets a product onto that list in the first place.
If you're a D2C apparel, footwear, or accessories brand, this is worth ten minutes of your attention. At WebEpex we build and maintain Shopify catalogs for exactly this kind of client across the GCC, Europe, USA, Canada, and India, and "will this actually fit me" is the single most common objection standing between a product page view and a sale.
This is also OpenAI's clearest shopping move since it pulled back on Instant Checkout in March, after finding people weren't actually buying inside the chat window. The company has stopped trying to process your sale. Now it's trying to own the moment right before someone decides to buy, the one that drives a huge share of fashion returns.
That part is worth caring about. Whether it moves the needle for you depends on something OpenAI doesn't control: how good your product photos are to begin with.
The honest trade-offs of letting ChatGPT dress your customers#
The upside is real. A shopper who tries on a jacket inside ChatGPT and likes what they see is a warmer lead than one scrolling a static product grid. Fewer blind guesses on size should mean fewer returns, and returns are one of the few D2C cost lines nobody has fully solved.
The downside is the one every try-on tool has shipped with since the first ones showed up on fashion sites years ago: a preview is not a fitting room. OpenAI says so itself, warning that an AI-generated preview "does not confirm garment measurements, sizing or real-world fit". A shopper who trusts the preview over your size chart and gets a bad fit anyway is now disappointed by your product and OpenAI's tool at once, and only your brand eats the return.
There's also the data question. ChatGPT keeps the reference selfie to speed up future try-ons, which a privacy-conscious shopper might not love, even with an opt-out buried in settings. None of this is disqualifying. It's just not the free win the headline makes it sound like.
How we're already prepping client catalogs for this#
We pulled five client product shots this week and ran them through ChatGPT's try-on ourselves, specifically to see which photo styles render clean previews and which don't, for our Shopify apparel and accessories clients across GCC, European, and North American markets.
The pattern held. Flat, evenly lit, front-facing shots on a plain background render a usable try-on. Shots with heavy styling, odd angles, or busy backgrounds don't render anything a shopper would trust. That's not a new finding. It's the same brief we give clients for Google Shopping and Meta catalog ads, now with one more reason behind it.
We used to treat clean product photography as a nice-to-have polish item, something to fix after launch if there was budget left. We stopped treating it that way once Google started pulling product images straight into AI Overviews, and this is one more reason it isn't optional anymore.
We're not telling apparel clients to chase a formal merchant integration that doesn't exist yet. We're telling them the plainer version: fix your product photography to the standard AI try-on tools already expect, because that standard pays off across every shopping surface, not just this one.
What I'd tell a client this week#
Don't rebuild your catalog around a feature that might get a formal merchant program in six months, or might not. Do this instead: pull your five best-selling product pages and check if the hero shot is flat, evenly lit, and front-facing on a plain background. If it isn't, fix those five first.
That's a half-day job. It helps every AI shopping surface reading your catalog right now, not just ChatGPT, and it costs you nothing to start. If your product photography is already clean, there's genuinely nothing to do this week. That's a fine answer too.
For more on how AI is reshaping where D2C sales actually happen, we've covered Meta's Muse checkout push, Google's AI Mode checkout changes, and why AI shopping chatbots are losing some brands sales.
If you want a straight read on whether your product catalog is ready for this kind of thing, send me five product URLs and I'll tell you what I see. Two minutes, no pitch. cal.com/webepex/growth-review
