Jamal shops for work shirts online every few months. He uses a wheelchair, and the fitting rooms at his local mall sit up a narrow ramp he would rather skip. So he orders the same shirt in three sizes, keeps one, and mails two back. That is just how it has worked for him.
Now picture this instead: Jamal uploads one selfie and sees how each shirt actually falls on him, before anything ships. That is not a hypothetical anymore. OpenAI just built ChatGPT’s new virtual try-on feature to do exactly that, and it could change how a lot of us shop online.
Clothing is the most-returned category in online retail. About 26% of Americans sent back something they bought online last year, according to Capital One Shopping data. Apparel return rates run between 20% and 30%, and some categories spike as high as 50%. Each return costs a retailer $10 to $65 once you count shipping, repackaging, and restocking.
Guess who pays for that math? We do, through higher prices and stingier return windows. Getting it right the first time means real money and real time back in your pocket.
How ChatGPT’s Virtual Try-On Actually Works
OpenAI rolled this out globally on October 1, 2026, as a free feature inside ChatGPT. Spot a jacket, a dress, or a pair of sneakers while chatting with the bot, and a “Try On” button now shows up in the shopping results. Tap it, upload a selfie, and ChatGPT generates an image of you wearing the item. You can also drop in a screenshot from a random website and ask ChatGPT to show you wearing that instead.
The feature runs on ChatGPT Images 2.5, a newer model OpenAI says renders more natural lighting and texture. ChatGPT also saves every try-on image to a new Favorites library, so you can compare sweaters side by side later instead of digging through old browser tabs.
Shoppers who used Google’s earlier virtual try-on tool shared those images far more often than they shared ordinary product listings, according to Google’s own reporting on the feature.
Who This Actually Helps
This is not just a convenience for online shopping fans. It is a real lift for:
- Parents of growing kids, who cannot try a hoodie on a 12-year-old between school pickup and soccer practice.
- People with disabilities who find in-store fitting rooms inaccessible or exhausting.
- People in small towns without a mall or department store nearby.
- Anyone on a tight budget who cannot absorb the cost of fixing a wrong-size order.
- Caregivers and busy professionals squeezing errands into a lunch break.
A good fit should not require a guess and a trip to the post office.
Part of a Bigger Shift in How We Shop
ChatGPT is not alone here. PYMNTS Intelligence data shows roughly 23% of U.S. consumers, about 61 million people, already start retail searches with AI, and clothing makes up nearly 28% of those AI-assisted purchases. We have seen the same pattern with Safari’s price-watching Notify Me feature, and with free ChatGPT workshops teaching older adults to navigate the internet more safely. AI is quietly taking over the tedious, risky parts of shopping, and virtual try-on is a low-stakes way for anyone, at any age, to get comfortable using it.
None of this makes online shopping foolproof. Sizing charts still vary by brand, and an AI-generated image is a preview, not a guarantee. But cutting your odds of a bad guess by even half means fewer trips to the post office and fewer dollars lost to return fees.
Next time you are eyeing something online and feeling unsure, tap “Try On” before you check out. If it saves you a single return, it already paid for itself.
Related Reading
- Safari’s Notify Me Feature Watches Prices So You Don’t Have To
- Free ChatGPT Workshops Are Teaching Seniors to Outsmart Scammers
- Instacart’s Clementine AI Finally Answers “What’s for Dinner?”
- PYMNTS: OpenAI Adds Virtual Try-On Feature to ChatGPT
- Ringly: 42 Ecommerce Return Statistics You Need to Know in 2026