An honest 2026 review of ChatGPT Virtual Try-On, OpenAI's in-chat shopping feature: how real it looks, fit accuracy, privacy, and where it stops for creators.
ChatGPT Virtual Try-On is a genuinely useful shopping feature: upload a photo, tap Try On in a result, and the Images 2.5 model renders a clothing item or accessory on you convincingly. For a shopper deciding whether something suits them, it is fast, easy, and often surprisingly good. Two honest limits define it — OpenAI says the images do not guarantee fit or size, and it is a buyer-side tool with no way to brand, caption, or publish anything, so for a creator or seller it is a research step, not a content tool.
OpenAI began rolling out Virtual Try-On globally on October 1, 2026 as part of its push to make ChatGPT a place people shop. The pitch is simple: when ChatGPT shows you a product, a Try On button lets you upload a selfie or full-body photo and see the item on yourself, and a companion Favorites feature saves products and your try-on images to a Library. It runs on ChatGPT Images 2.5, the image model OpenAI shipped in September 2026.
This review judges it two ways, because two very different people search for it. The first is a shopper asking "does this look good on me before I buy?" The second is a creator or seller asking "can I use this to make content?" Those answers diverge sharply, and the score reflects both — strong for the shopper it was built for, weak for the creator it was not.
Everything here is reconciled against OpenAI's launch coverage as of 2026-10-01. Because this is a brand-new, actively rolling-out feature, treat plan-by-plan availability, catalog coverage, and the exact placement of the Try On button as a moving snapshot, and confirm specifics against OpenAI's own help pages. Where a detail could shift, I generalize rather than invent precision.
ChatGPT Virtual Try-On is an in-chat shopping feature. When ChatGPT returns a shopping result, a Try On button appears; you upload a selfie or a full-body photo and ChatGPT generates an image of how that clothing item or accessory might look on you. You are not limited to results it surfaces — you can upload an item image yourself, such as a screenshot from a web page, and ask ChatGPT to try it on. The companion Favorites feature saves products you like, and your try-on images, into a Library in the app for later. The quality comes from ChatGPT Images 2.5, which brings more natural lighting, richer textures, and more reliable edits — the reason a garment can drape onto a real photo plausibly instead of looking pasted on. What the feature deliberately is not: it is not a size or fit guarantee (OpenAI says so explicitly), and it is not a creation tool for sellers. There is no branding, no catalog connection to your own store, no captioning, and no publishing. It produces a private preview image inside the shopper's chat and stops there.
This is for shoppers — anyone weighing an apparel or accessory purchase who wants to picture the item on themselves before buying, or to compare a few options quickly without a physical try-on. For that person it is a strong, low-effort aid. It is a weak fit for the creator, seller, or brand whose actual need is producing and distributing shoppable content: on-model imagery they own, carousels of angles, short-form video, product posts in a consistent voice across many feeds. Virtual Try-On does none of that; it is, at most, a research or mockup step before the real content work begins elsewhere.
| Dimension | Score | Why |
|---|---|---|
| Try-on realism & image quality | 4.2 / 5 | Running on Images 2.5, garments drape onto a real photo convincingly with natural lighting — a clear step up from pasted-on effects. |
| Ease of use | 4.6 / 5 | Upload a photo, tap Try On in a shopping result; there is almost no learning curve. |
| Flexibility (bring your own item) | 4.1 / 5 | You can upload an item image, such as a web screenshot, rather than being limited to ChatGPT's surfaced results. |
| Item & catalog breadth | 3.5 / 5 | Covers clothing and accessories, but as a new global rollout the real-world coverage across stores and categories is still proving out. |
| Fit & size accuracy | 2.5 / 5 | OpenAI is explicit that the images do not guarantee fit or size — it is a styling preview, not a measurement. |
| Favorites / saving | 3.8 / 5 | A Library saves products and your try-on images for later, which is a sensible, if basic, shopping aid. |
| Privacy of uploaded photos | 3.0 / 5 | You upload photos of your body or face to OpenAI; review current data-handling terms before use, especially for full-body images. |
| Value for shoppers | 4.0 / 5 | Bundled into ChatGPT's shopping experience for logged-in users at no separate charge. |
| Usefulness for creators & sellers | 1.6 / 5 | Buyer-side only — no branding, no catalog connection, no captioning, and no publishing, so it produces nothing a creator can ship. |
There is no separate price for Virtual Try-On. It arrives as part of ChatGPT's shopping experience, which OpenAI has been rolling out to logged-in users across its tiers, so if you are already in ChatGPT the feature is simply there when a shopping result appears. That makes it effectively free to the shopper at the point of use, which is the right model for a purchase-assist feature — charging to preview an item you might buy would kill the behavior it is trying to create.
The real cost, such as it is, is not money. It is the data you hand over — photos of your face or body uploaded to OpenAI — and the honest caveat that a convincing image is not a fit guarantee. A shopper should weigh the preview against the size chart and return policy, not treat the render as a promise. Because plan-level availability and any usage limits are still settling during the rollout, confirm current terms on OpenAI's help pages.
For a creator or seller, the pricing question is the wrong one entirely, because this is not a tool that produces content you can sell with. The spend that matters for that audience is on a content engine that generates and publishes shoppable posts — a different product category, covered in the positioning below.
| Use case | Fit | Why |
|---|---|---|
| Previewing how an apparel item looks on you before buying | Strong | This is exactly what it is built for — fast, easy, and convincing on the Images 2.5 model. |
| Comparing a few styling options quickly | Strong | Trying several items on your own photo and saving favorites is low-effort and genuinely useful. |
| Trying an item you found elsewhere on the web | OK | You can upload an item image, though results depend on the clarity of that screenshot and your photo. |
| Confirming the size that will fit you | Weak | OpenAI states the images do not guarantee fit or size — it is a visualization, not a measurement. |
| Producing on-model product imagery a seller can publish | Weak | The output is a private preview in the shopper's chat with no branding or export-to-publish path. |
| Making social content or ads from a product | Weak | There is no captioning, multi-format fan-out, or scheduling — it is not a content tool. |
To be clear and honest: Kompozy is not a virtual try-on tool, and it does not compete with ChatGPT Virtual Try-On. They sit on opposite sides of the same shopping workflow. Virtual Try-On is buyer-side — it helps a shopper picture an item on themselves. Kompozy is seller-side — it is a content generation and publishing engine that turns a product or a campaign into finished, on-brand posts and ships them across platforms. If what you want is to see a shirt on your body, use ChatGPT; Kompozy does not do that step and will not pretend to.
Where Kompozy matters to anyone selling is the job Try-On raises the bar on but never touches: producing enough strong, consistent shoppable content that you actually get discovered. Kompozy generates scene Photo Posts and Infographic Photo on the same gpt-image family that powers Images 2.5, face-locked Persona Photos for a consistent model, brand-exact Carousel Posts, Clipped Shorts and Listicle Video for short-form, plus a Blog Article and an Email Newsletter — all governed by one Persona Brief and published on autopilot across eight social platforms plus blog and email. Many sellers will use both: Try-On helps their customers decide, Kompozy produces the content that brought those customers in.
For a shopper, yes — it is a fast, low-effort way to picture a clothing item or accessory on yourself before buying, and on the Images 2.5 model the results are convincing. Just treat it as a styling preview, not a fit guarantee. For a creator or seller hoping to make content, no — it produces a private preview image with no way to brand or publish.
It is not a fit or size tool. OpenAI explicitly states the generated images do not guarantee fit or size. Use it to judge style and look, and still rely on the size chart, measurements, and return policy for fit.
There is no separate charge. It arrives as part of ChatGPT's shopping experience for logged-in users. Because availability by plan is still settling during the global rollout, confirm current terms on OpenAI's help pages.
The feature requires uploading a selfie or full-body photo to OpenAI. Whether that is acceptable is a personal data-privacy call; review OpenAI's current data-handling and content terms before uploading, especially for full-body images.
Not meaningfully. It is a buyer-side tool that produces a private preview inside the shopper's chat — there is no branding, catalog connection, captioning, or publishing. For producing and distributing shoppable content, you need a separate engine like Kompozy.
Favorites, launched alongside Virtual Try-On, saves products you discover — and your try-on images — into a Library in the ChatGPT app so you can compare and return to them later.
It runs on ChatGPT Images 2.5, OpenAI's September 2026 image model, which brings more natural lighting, richer textures, and more reliable edits — the quality that makes a garment drape onto a real photo convincingly.
See ChatGPT Virtual Try-On vs Kompozy comparison → · Get Started →