Shoppers browsing apparel catalogs inside conversational prompts can now inspect outfits directly on their own likenesses without leaving the application window or visiting third-party tailoring visualization websites. OpenAI rolled out ChatGPT virtual try-on globally across desktop and mobile clients this week, enabling buyers to render clothing items onto uploaded portraits [1]. The update integrates dedicated visual simulation algorithms. Instead of guessing how a tailored jacket or summer dress might drape in reality, buyers can evaluate generated previews before making retail purchasing decisions [2].
How ChatGPT Virtual Try-On Functions
The virtual dressing system activates whenever a product listing for apparel appears during a routine conversation with the chatbot. Through ChatGPT virtual try-on, users tap on a displayed piece of clothing, select the dedicated Try on button, and upload a selfie with a full-body photograph to generate their customized visual preview [1]. The software does not restrict users solely to items surfaced by the conversational engine. Shoppers can also upload external screenshots of garments discovered elsewhere across the web, including celebrity outfits or formal wedding attire, to prompt identical simulated fittings within the chat window [2].
Underlying these visual transformations is ChatGPT Images 2.5, an updated generation architecture that OpenAI deployed in September. Company representatives stated that the refreshed model generates more natural lighting and realistic fabric textures than older systems, preventing garments from appearing like flat digital stickers pasted over three-dimensional human bodies [2]. While OpenAI expanded conversational voice chat capabilities to process spoken dialogue in real time, this release directs generative resources toward fine-grained spatial modeling and retail product visualization. The model reconstructs fabric folds convincingly.
Platform consistency spans native mobile applications and desktop web browsers, ensuring that visual generations remain synchronized across devices. Shoppers can test outfits or accessories like scarves, coats, and shoes through single prompts [3]. The processing takes only moments. Because rendering runs on remote infrastructure, mobile hardware demands remain modest during complex image generation tasks [1].
What Can ChatGPT Do Now for Shoppers?
Practical evaluations demonstrate that the feature operates best when users pair specific budget parameters with clear aesthetic criteria. Tech journalist Elton Jones tested the capability for Tom’s Guide, issuing prompts requesting tailored jackets under a set price limit, casual workplace outfits, and versatile sneakers designed to complement existing personal wardrobes. The assistant proposed candidate items from web catalogs before rendering each piece directly onto Jones’s test photo. Having integrated ChatGPT virtual try-on into everyday product queries, such tailored workflows allow consumers to experiment with unfamiliar aesthetic styles, bold color palettes, or seasonal outerwear layers without committing retail funds upfront before inspecting appearance [2].
Beyond singular visual tests, OpenAI introduced dedicated curation tools that let buyers bookmark items directly from chat interfaces. Users can click a bookmark icon to save garments to Favorites or organize discoveries into custom folders for upcoming events and seasonal planning [2]. Engadget reporter Ian Carlos Campbell observed that these saved items populate an in-app Library, allowing consumers to revisit product specifications later [1]. Jones concluded that previewing fits, archiving images, and visiting physical retailers to inspect fabrics in real time represents a compelling shopping sequence [2].
Can a synthetic photograph replace the physical dressing room entirely? Synthetic imagery illustrates drape, cut, and color coordination, but it cannot verify exact physical sizing or seam comfort [3]. Jones noted that artificial lighting simulations still require eventual physical confirmation at retail storefronts. Sizing tolerances vary across clothing manufacturers [2].
Managing Virtual Try-On in ChatGPT Settings
Reference photographs persist across active sessions. OpenAI stores submitted user portraits to streamline recurring ChatGPT virtual try-on requests without requiring fresh uploads for every query [2].
Managing these stored likenesses requires navigating through the account preference panel. To inspect or modify saved portraits, users must head to Settings > Personalization > Reference photos, where the interface displays every active reference portrait registered to the account. From this dedicated menu, consumers retain the authority to add supplementary photos, swap existing full-body pictures with updated poses, or delete reference material permanently with a single click. Maintaining accurate reference angles improves subsequent simulations significantly because ChatGPT Images 2.5 calculates proportions directly against the original photograph’s contours. Removing outdated pictures ensures that subsequent wardrobe simulations reflect current hairstyles and bodily dimensions accurately without retaining unwanted personal photographic records across company server infrastructure [2].
The system accommodates diverse portrait types depending on the intended garment category. Close-up selfies provide sufficient structural detail for hats, eyeglasses, and neckwear, whereas coats and trousers necessitate uncropped full-body captures with clear peripheral spacing [2]. Providing clean backgrounds minimizes visual artifacts around limb boundaries. Testing multiple poses allows shoppers to verify whether hemlines sit appropriately when standing upright. When users fail to supply clear full-body angles, the image generator must extrapolate leg proportions, occasionally yielding slight geometric distortions around footwear [1].
Photo Reference Privacy and Data Training
Uploading personal portraits to cloud-hosted neural networks introduces standard data governance questions that consumers should weigh before activating the tool. Campbell emphasized in Engadget that, by default, images uploaded from personal consumer accounts can be utilized by OpenAI for model training purposes unless individuals explicitly opt out. Similar governance debates surfaced when Meta Muse shopping tools and personal privacy safeguards entered public testing. Users concerned about facial privacy must actively verify their account settings to disable training permissions before submitting unedited portraits [1].
OpenAI maintains opt-out controls within standard account preferences, allowing individuals to exclude conversation histories and uploaded attachments from corporate training corpora [1]. Opting out prevents engineers from ingesting private dressing photos into subsequent multimodal model iterations. Data protection policies place the burden of opt-out selection onto individual consumers rather than establishing automatic default shielding [3]. Privacy advocates urge caution. Deleting reference images from the personalization tab removes them from active virtual fitting workflows immediately [2].
Because ChatGPT virtual try-on operates across interconnected mobile and web platforms, uploaded reference photographs remain accessible across all logged-in devices until intentionally purged. Account holders who utilize shared tablets or public workstations should remain mindful of saved previews stored within browser sessions [2]. The system stores rendered results in recent chat transcripts. Users can clear specific conversations to scrub generated dressing images from immediate chat history [1].
Comparing OpenAI Shopping Tools to Competitors
The concept of using artificial intelligence to preview garments on personal photos did not originate in OpenAI’s research labs. Google added an identical capability to Google Search in 2025, enabling users to upload portraits and visualize apparel across various retail listings [1]. In testing artificial intelligence models such as Google Gemini, Apple Siri, and Anthropic Claude, technology reviewers have tracked how conversational bots gradually evolve into comprehensive shopping assistants [2]. OpenAI integrates the tool natively.
OpenAI’s broader commercial strategy demonstrates sustained interest in digital commerce, even though the company does not operate a dedicated retail marketplace on the scale of established competitors like Amazon. The organization previously experimented with transaction completion inside chats through its Instant Checkout feature, which has since been discontinued. Engadget clarified following initial reporting that OpenAI does not receive a financial cut of retail transactions executed through its platform. Instead, the firm generates commerce revenue by serving merchant advertisements directly inside conversational threads [1].
The consistent rollout of shopping tools over the past year suggests that OpenAI views retail integration as a viable commercial frontier. While generative artificial intelligence infrastructure requires massive capital expenditure to operate, visual try-on features bridge conversational recommendations directly with tangible commercial spending at retail partner outlets [1]. Consumers interested in testing the capability can launch ChatGPT on desktop or mobile, search for an apparel item, and upload a test portrait to evaluate their prospective wardrobe in simulated reality [2].
- ONLINE NEWS Campbell, I. C. (2026, October 1). ChatGPT can now help you virtually try on clothes. Engadget. [Article Link]
- ONLINE NEWS Jones, E. (2026, October 2). You can now virtually try on clothes with ChatGPT before you buy them — here’s how it works. Tom’s Guide. [Article Link]
- ONLINE NEWS TechRepublic. (2026). OpenAI launches virtual try-on for clothes and accessories in ChatGPT. TechRepublic. [Article Link]
1 comment