OpenAI has added ChatGPT new capabilities—this time to help youtry on clothes online. On October 1, OpenAI announced the global rollout of two shopping features: “virtual try-on,” which lets you upload a selfie or full-body photo to see how a garment or accessory looks on you; and “Favorites,” which saves desired items to an in-app library for later review.
How Virtual Try-On Works
The process is highly intuitive. When you search for products in ChatGPT , a Try On button appears among the results. Click it and upload a selfie or full-body photo, and it will “dress” you in the item so you can see the effect instantly. Beyond selfies, you can also upload a product image directly (e.g., a webpage screenshot) forChatGPT to try on for you. ChatGPT Upload a photo to try on
- : Selfie or full-body photo—to preview how clothes or accessories look on you;Upload a product image
- : Webpage screenshot or product image—to have it tried on instantly;Webpage screenshots or product images—just let it try them on for you
- FavoritesSave desired products to your Library, archiving them alongside try-on images
This feature is powered by the newly released ChatGPT Images 2.5 modelOpenAI claims this version generates more natural lighting, richer textures, and follows editing instructions more reliably, while also reducing image generation latency. For highly detail-sensitive scenarios like “trying on clothes,” texture and lighting realism directly determine whether results are “presentable” versus “clipping through clothing.” This continues the rapid iteration trend in multimodal models, with models like Gemini 3.8 Live already achieving real-time conversational-level visual understanding
Not just virtual try-on:ChatGPT Your personal shopping advisor
OpenAI clearly isn’t satisfied with being just a “virtual fitting room.” Its demonstrated shopping capabilities also include:
- Describe a style for outfit recommendations:Tell ChatGPT Specify a style you’d like to try, and it will help you source all required items for a complete outfit
- Upload celebrity outfits to find matchesUpload celebrity street-style photos, and it will identify purchasable equivalents shown in the image
The latter directly advances the “fashion inspiration discovery” category long dominated by Pinterest and Google—both platforms have long served as primary entry points for users seeking outfit inspiration before converting to e-commerce purchases. AI shopping assistants are evolving from “helping you place orders” to “helping you discover and curate outfits” ChatGPT Competitive landscape: Google is already ahead
Notably, virtual try-on is not OpenAI’s innovation. Google launched its virtual try-on feature last year, and Google has recently intensified its shopping-related initiatives—including prior reports on its evolution from “helping you search” to “helping you order”
This latest release is largely about “catching up + filling gaps,” completing both visual try-on and Favorites functionality Gemini tested e-commerce shopping in India Reporting on the evolution from “help you search” to “help you order.”ChatGPT This round of features is largely about “catching up + filling gaps,” adding visual try-on and saving items.
OpenAI has also taken detours in the shopping space—it previously attempted an “instant checkout” feature, but abandoned it due to poor performance. The path toward AI shopping assistants remains uncharted for all players: going too far (like agentic AI startup Instinct, which proactively recommends products) risks alienating users, while playing it too safe fails to retain them.ChatGPT This “try-on + save” combination represents a relatively cautious entry point.
What This Means for Ordinary Users
The most immediate benefit isReduces the “landmine rate” of online shopping.The biggest fear when buying clothes is “looks great on the model, disastrous on me.” While virtual try-on cannot fully replicate real-world wear, it at least helps filter out clearly unsuitable options in terms of cut, color coordination, and proportion. For frequent online shoppers, this saves not only return shipping costs but also time spent repeatedly returning and exchanging items.
That said, a reality check is warranted: current AI try-on results remain “reference-grade”—fabric drape and skin-tone fidelity under real lighting conditions still fall short, and they cannot serve as sizing references. They are better suited for judging “does this style suit me?” rather than “what size should I buy?”
Frequently Asked Questions (FAQ)
ChatGPT Is virtual try-on free?
This feature relies on ChatGPT image generation capabilities, typically counted against ChatGPT account image generation quotas. Free and paid users may have different available quotas; specifics are subject to OpenAI’s official policy.
ChatGPT How do you use virtual try-on?
exist ChatGPT Click the Try On button in
ChatGPT product search results, upload a selfie or full-body photo to preview how the item looks on you; alternatively, upload a product image for AI-generated try-on.
Both rely on image generation for visual try-on, delivering similar core experiences. The key difference lies in ecosystem integration: Google’s try-on is deeply embedded within e-commerce and search workflows, whereas ChatGPT excels in conversational interaction—enabling natural-language style descriptions or uploading celebrity outfits to find matches.
Can virtual try-on results be used as sizing references?
No. AI try-on is intended primarily to assess whether style, color coordination, and cut suit you—it delivers “reference-grade” results and cannot replace official size charts or physical try-ons.
Summarize
ChatGPT Global rollout of virtual try-on and save functionality advances the AI shopping assistant from “help you search” further toward “help you try, help you style.” Powered by the visual capabilities of the Images 2.5 model and enhanced by natural-language interaction, it launches a new challenge to Pinterest and Google in the fashion discovery space. For users, it’s a practical tool to reduce online shopping misfires—but its output remains “reference-grade,” not a substitute for real-world try-ons.
Want to Discover More Useful AI Tools? Explore Our AI Model Library and Tool Comparison Engineor continue reading:GPT-6.1 Sol benchmark · GPT-6 Sol/Luna evaluation · ChatGPT Voice Agent · All Reviews.