How pretty am I? An honest measurement, not a verdict.

Four measurements carry the headline score, and you are shown all four rather than a single unexplained number:

• Symmetry. Every paired feature - eyes, brows, nose, mouth, jaw - is compared against its mirror image across the centre line of your face, and the deviation becomes a score out of 100. The jaw carries the most weight, because it is the pair the eye notices most.
• Golden ratio. The height of your face against its width, scored by how close it sits to the classical 1.618 target.
• Facial thirds. Hairline to brow, brow to the base of the nose, nose to chin. An even split scores highest. The lower third is the one most often out of balance.
• Facial fifths. The width of your face divided into five eye-widths, which tests eye spacing and the proportion of the outer face.

Alongside those, the panel reports your face shape, jaw angle, width to height ratio and how square your head was to the camera. The method page shows the formula, the weights and where the reference numbers came from.

Almost every face-scoring site online is selling something on the back of a number - a subscription, an app, a procedure. This one sells one thing, a written report on the same measurements, and it never sells a fix, which makes it much easier to be blunt about what a score like this can carry.

It cannot tell you whether people find you pretty. Facial geometry is one input into that judgement and a fairly weak one. Expression, eye contact, skin, hair, clothes, voice, posture and simple familiarity all move it more. Two faces that measure four points apart are routinely rated the other way round by actual humans.

It also cannot tell you what to change about your face, and this site never will: there is no procedure list and no link to a clinic anywhere on it. The paid report is about presentation - hair, light, grooming, skincare, habits, and how to photograph this face - and its images style your photo without ever touching the face itself. A symmetry deviation of four percent is not a problem waiting to be solved. It is a fact about a face in the same way that a shoe size is a fact about a foot.

If a score from any tool lands badly, the genuinely useful response is to close the tab. A measurement has no authority over how you feel about your own face.

• Face the camera square. Turning even 10° drops the symmetry score.
• Hold the camera at eye level. Shooting from above compresses the lower third of your face; from below, it stretches it.
• Neutral expression, mouth closed. Smiling changes four ratios at once.
• Even light from the front. Side shadow reads as asymmetry.
• Hair off your forehead, no glasses, no filters, no beauty mode - a smoothing filter changes the landmark positions.
• One face in frame, filling most of it.

The panel grades your head angle and tells you when it is off, so you can retake the photo before reading anything into the numbers.

Q:How pretty am I - can a website actually tell me?

It can tell you something narrow and precise, and nothing broad. A tool like this measures the geometry of your face in one photo: how closely your left and right sides mirror each other, how evenly the face divides into thirds and fifths, and how the main distances compare to classical proportion targets. Those are real, repeatable measurements. But prettiness is a judgement made by people, and people weigh expression, skin, hair, styling, colouring, movement and familiarity far more heavily than geometry. So the honest answer is: a website can measure your face, and it cannot tell you how pretty you are.

Q:Is this how pretty am I test free?

Completely. There is no account, no email, no trial and no credits, and the whole measurement is JavaScript running in your own browser tab, so it costs nothing to run and nothing to give away. The one thing for sale on the site is an optional written report on the same measurements, offered under the result; the test never needs it and never holds anything back to sell it.

Q:Is my photo uploaded, stored or shared?

Not for the test. The face-landmark model is downloaded to your browser, and your browser reads the photo locally with it. The image is never sent to a server, never written to a database and disappears the moment you close or refresh the tab. The single exception is the optional paid report: if you choose to buy one, that one photo is sent, with your consent, to the two AI services that draw and write it, and deleted from our database the moment the report is built. Nothing happens to your photo unless you choose that.

Q:How pretty am I on a scale of 1 to 10?

This test deliberately does not give a mark out of 10. A 1-10 rating is the format every face-rating site uses, and it is the format most likely to be read as a verdict on a person rather than a measurement of a photo. The score here is out of 100 and it is labelled facial harmony, because that is what is being measured. If you want to know how the familiar 1-10 pretty scale works and where its numbers actually come from, that is explained on the pretty scale page.

Q:What is a good facial harmony score?

The median measured face scores 74 out of 100. Half of all faces land between 65 and 80, a score of 83 or above sits in the top 15%, and 86 or above in the top 5%. Scores cluster tightly because human faces genuinely do cluster tightly on symmetry and proportion, so a few points apart is a real difference in geometry rather than a rounding artefact. This site never reports how far down that distribution a face sits.

Q:Why did my score change between two photos?

Because the tool measures the photo, not you. Turning your head even ten degrees compresses one side of your face and drops the symmetry score. Shooting from above shortens the lower third of the face. A smile widens the mouth and moves several ratios at once. The tool flags a bad head angle when it detects one - a square, neutral, evenly lit photo is the only fair comparison between two shots.

Q:How accurate is an AI prettiness test?

The measuring is accurate and the interpretation is not. The same photo returns the same numbers to the decimal, every time. What is unreliable is the leap from those numbers to attractiveness: published research consistently finds symmetry and proportion explain only a modest part of how attractive people rate a face. Treat the score as an accurate measurement of a narrow thing, not an accurate verdict on a broad one.

Q:Does the test work the same for every skin tone and ethnicity?

The landmark detection works across skin tones and face shapes. The proportion targets do not travel as well. Facial thirds, fifths and the golden ratio come from a European-descended aesthetic tradition, and what counts as ideal spacing genuinely differs across populations. A lower proportion sub-score can mean your face differs from that one tradition, not that anything about it is wrong. Every ratio-based face test shares this limitation, including this one.

Q:How old do I need to be to use this?

This site is intended for adults. If you are a teenager wondering how pretty you are, the most useful thing on this page is not the score - it is the fact that facial geometry has been measured for a century and still predicts very little about how people respond to a face. A number from any website, this one included, has no authority over how you feel about yourself.

Q:Can I use it on a photo of someone else?

Please don't. Measuring someone else's face without them knowing is the obvious way to misuse a tool like this. Use your own photo.

• Am I pretty? What a photo can and cannot answer
• The 1-10 pretty scale, and where its numbers really come from
• Pretty or ugly? Why the question has no answer
• Pretty privilege: what the research actually found
• How the score is worked out, formula and all