OpenAI rolls out virtual clothing try-on and saved item libraries in ChatGPT

Virtual try-on rollout

On 1 October 2026, OpenAI rolled out two e-commerce features across the web and mobile versions of ChatGPT: a Virtual Try On system and a saved-item tool named Favorites. The virtual try-on mechanism enables users to view how apparel and accessories look on their own bodies directly inside the chat interface. When a clothing item appears in a product card during a search, users can tap a dedicated "Try on" button and upload either a selfie or a full-body photograph. The tool is not restricted to items generated by the chatbot. Users can also upload screenshots from third-party online retailers, photographs taken from fashion websites, or images of celebrity outfits to request virtual fittings on their uploaded reference pictures.

Technical foundation and user workflow

The feature is powered by ChatGPT Images 2.5, an image generation and editing model released by OpenAI on 8 September 2026. OpenAI stated that the updated model operates up to 50% faster than Images 2.0, while producing richer fabric textures, more natural lighting, and higher fidelity to prompt editing instructions. The model also offers improved retention of the original human subject's facial and physical features across edits. Users can interact with the system by entering text descriptions of specific aesthetics, asking the chatbot to identify purchasable pieces matching a particular dress code, or seeking look-alike items from social media trends. The release builds on prior commerce tools introduced throughout 2026, including structured product comparison tables and the Shopping Research interface designed to organize purchase criteria.

Library organization and data policies

Alongside the fitting tool, the Favorites function provides an integrated system for storing discovered products and generated try-on imagery. Users can organize saved items into custom folders inside the ChatGPT Library, the interface area that stores uploaded files and generated visual assets. This folder system allows shoppers to retain outfit options and compare competing items across multiple browsing sessions rather than buying immediately. The try-on tools connect to OpenAI's broader commercial initiatives, which include merchant advertisements in chat streams and transaction fees collected through Instant Checkout. For personal accounts, uploaded user photos remain subject to OpenAI's standard training policy, meaning images can be used to train future models unless an account holder explicitly opts out.

Competition in visual commerce

OpenAI's latest release expands its push into visual product discovery, a sector where search engines and specialized platforms already hold substantial market presence. Google introduced a comparable AI-based virtual try-on tool in Google Search in 2025, while retail firms including Amazon, Walmart, and Zalando have deployed visual fitting systems. Visual inspiration platforms such as Pinterest have also long served as starting points for fashion discovery that convert into third-party retail purchases. OpenAI has adjusted its commercial strategy following mixed results with earlier efforts, such as Instant Checkout, while third-party agent startups in the space, such as Instinct, previously faced user backlash after delivering proactive product suggestions that consumers viewed as intrusive advertising.