Which Celebrity Do I Look Like? An Honest AI Look-Alike Test (No Face Recognition)
Friends say you look like someone famous — but who? We explain how our look-alike test picks three celebrities from a description of your features, never by matching faces.

Contents
Everyone has heard “you look just like…” at least once — usually followed by a different name every time. The celebrity look-alike test on ImageGPT settles it with one selfie: three famous people with a similar look, a score for each, a line on what you share, and a shareable card for your story or group chat.
First, the honest part. This is a game, not face recognition. The AI never compares your photo with photos of celebrities and never tries to find out who you are. Here's what it actually does.
How the look-alike test works
- A vision model looks at your photo and describes your look in words: face shape, eyes, brows, nose, lips, jaw, hair, facial hair, glasses, expression and overall vibe. It's instructed never to identify the person, even if you happen to look famous.
- From that text alone — no photo — it picks three internationally known actors, musicians or athletes of the same gender presentation and a similar age range whose typical look fits the description best.
- Each match gets a look-alike score and one short sentence on the shared features. Skin color, ethnicity and nationality are never used as the explanation.
You also get your screen type — the kind of role your look would be cast in, like “the charismatic leading man”. Only adults who became famous as adults are picked; child stars are excluded.
Real result cards


On the first card the shared features were a rounded face, short curls, expressive brows, trimmed facial hair and a broad smile, and the type was “the charismatic leading man”. On the second: an oval face, expressive round eyes, full brows and a warm smile — “the thoughtful young lead”.
How to read the score
The percentage is the AI's estimate of how close a celebrity's typical look is to yours. Nobody measured it against the celebrity's photos — it exists to rank the three picks against each other. The small print on every card says the same: similarity of type, not face recognition.
Why a different photo gives different celebrities
The description depends on the photo. A big smile versus a serious face, hair up versus down, a beard versus clean-shaven, makeup, angle and light all change what the model sees — and so the matches. That's part of the fun: try a few photos and see which names keep showing up.
What photo to use
- Just you, facing the camera, your whole face visible.
- Good light, no heavy filters or face masks.
- Your everyday hair, beard and glasses — they shape your type more than you'd think.
- Drawings, pets and photos without a face won't work: the analysis doesn't run and the credits come back.
Step by step
- Open AI Face Analysis in the Photo section.
- Under What to find out, choose Celebrity look-alike.
- Upload a selfie and click Analyze my face.
- Download the card or find it later in your history, along with the text result.
The card is 1080×1350 — the 4:5 portrait size that fits a feed post. It's drawn in the language of the site you're using: English, Spanish, German, French, Portuguese, Italian and several more; for scripts the font doesn't cover, the card is in English.
Fun ways to use it
- A group-chat game: everyone uploads a selfie and you compare cards.
- A Halloween or theme-party costume idea: dress as your top match.
- Haircut inspiration: borrow the cut of a celebrity with your type and try it with the haircut tool first.
Pricing
One analysis costs 3 credits, card included. If there's no face in the photo or the analysis fails, the credits come back. New accounts get 6 free credits.
Upload a selfie and find your celebrity look-alikes.
Find my look-alikeMore examples and answers about the test.
Celebrity look-alikeTry everything from this article in minutes
All the models and tools from this article are in one place. You get 6 free credits after sign-up, and every price is shown before you run it.
Find my look-alike

