ChatGPT Adds Virtual Try-On for Apparel and Accessories

OpenAI has added a virtual try-on feature to ChatGPT that lets users render themselves in clothing and accessories before they buy. The tool, embedded inside the existing ChatGPT interface, accepts a live selfie or an uploaded photo.

The feature, per OpenAI's product description: "Users can take or upload a selfie to generate images and see how clothing and accessories look on them prior to purchase."

The mechanic matters more than the novelty. ChatGPT is no longer just an answer engine for retail queries. It now stages the try-on itself, inside the same window where shoppers ask about shipping, returns and sizing. No additional app, plugin or SDK sits between the prompt and the render.

What this changes for marketers

Virtual try-on has historically run through two channels: retailer-owned smartphone apps and AR lenses on social platforms. Both have required shoppers to leave the surface where they were already browsing. ChatGPT removes that step. A consumer mid-conversation about a wedding outfit, a job-interview suit or weekend denim can now generate a styled image in the same chat thread.

For brand and retail marketers, that collapse carries three specific consequences:

• Product exposure is now a conversational outcome. When a shopper prompts ChatGPT for a navy blazer under $200, any brand surfaced in the response earns an impression inside the try-on view. Traditional paid placements and keyword bids do not apply.

• Visual presentation is now model-mediated. The PDP studio shot is no longer the only image a consumer sees before buying. OpenAI's system generates the drape and, to a meaningful degree, the fit. Brand teams cede some control over how the garment appears in the rendered view.

• Merchandising rank moves upstream. Buy signals — what gets tried on, what gets saved, what converts — now form inside the chat thread before any retailer-side analytics records the session. The path-to-purchase funnel that marketers have measured against for decades now starts at a different point.

What to do this quarter

The first workstream is operational. Brands with structured product feeds — clear color, silhouette, occasion and fit attributes — will render better inside generative try-on. Brands operating on sparse feeds will produce weak outputs and lose trial sessions.

The second is measurement. Owned PDP traffic and retailer traffic can be expected to dilute as shopper try-on shifts into the assistant. Last-click attribution models will undercount ChatGPT's contribution to upper-funnel exposure.

The third is competitive audit. Retailers with heavy investment in proprietary try-on apps should benchmark that experience against ChatGPT's zero-download path. The economics of convenience just moved.

What to watch

OpenAI has not disclosed whether try-on images will link directly into retailer checkout, whether sponsored placements will appear alongside the generated renderings, or how user-supplied selfies will be handled for model training. Each answer will reshape e-commerce media planning.

The render itself is now shipped. The marketing playbook around it has not been written.