Am I Pretty? An Honest Answer, With Real Numbers
Upload one photo and get a score out of 10 — then see exactly where it sits among 17,032 real analyses, which is the part every quiz leaves out. No cruelty, no flattery, no credit card.
Upload a clear photo
Click here or drag and drop your image
Choose one front-facing photo in even lighting
What the score actually measures
Before the number, the honest framing. The test reads the geometry of your face from one photo — symmetry, the proportions between features, skin texture, the line of the jaw and the shape of the eyes, twelve features in all — and places the result on a ten-point scale calibrated against real analyses. That is a consistent AI estimate, and it is a narrow one. It sees a still image. It does not see your expression when you laugh, how you carry yourself, your voice, or anything at all about who you are.
We say that at the top rather than in a footnote because the people typing "am I pretty" into a search box are often not really asking about geometry. They are asking whether they are okay. A facial-geometry score cannot answer that question and we will not pretend it can. What it can do — and what a quiz cannot — is tell you plainly where your face sits among 17,032 others, so that the number you get comes with a sense of scale instead of a feeling.
Where most people actually land
The average score across 17,032 analyses is 5.59, and the median is 5.3. If you scored a 5, you are exactly average — and average is not a consolation prize here, it is the most common thing there is. 49.6% of everyone lands between 5.0 and 5.9. Add the fours and that is 74.4% of all faces in the middle two bands.
| Score | People | Share |
|---|---|---|
| Below 4.0 | 77 | 0.5% |
| 4.0 – 4.9 | 4,230 | 24.8% |
| 5.0 – 5.9 | 8,450 | 49.6% |
| 6.0 – 6.9 | 2,891 | 17% |
| 7.0 – 7.9 | 824 | 4.8% |
| 8.0 – 8.9 | 524 | 3.1% |
| 9.0 + | 36 | 0.2% |
Notice how thin the top is. Scores of 8 and above are 3.3% of the dataset. A 9 is 0.2%. Most of the faces you think of as exceptionally pretty would land in the sevens on this scale, and the sevens are less than one person in twelve. If you came in expecting a 9 and got a 6, the scale was always going to do that, to almost everyone.
Figures use analyses from January 2026 onward (n=17,032), after the Dec 2025 model recalibration. Earlier analyses used a different scale.
Your score as a percentile
This is the table that turns "am I pretty" into an answer. Read it as: if you scored this or higher, this is the share of people you are level with or ahead of.
| You scored at least | You are in the top |
|---|---|
| 5 | 74.7% |
| 6 | 25.1% |
| 7 | 8.1% |
| 8 | 3.3% |
| 9 | 0.2% |
A 6 puts you ahead of three in four people. A 7 puts you in the top 8.1%. The 90th percentile is a score of 6.8, the 95th is 7.8, and the 99th is 8.8. Those thresholds are lower than most people would guess, which is exactly why a raw number without the table underneath it is so easy to misread.
If you scored in the fives
You are where half of everyone is. Not below the bar — the bar. The model rated 8,450 faces in that band, more than every other band combined, and the average potential for that group is 6.66 — meaning the typical five has more than a point of headroom in the things that can actually change.
Figures use analyses from January 2026 onward (n=17,032), after the Dec 2025 model recalibration. Earlier analyses used a different scale.
Why you judge your own face harder than anyone else does
People who ask this question usually already have an answer in mind, and it is usually lower than the one the test gives them. That gap is not a mystery. There are a few well-understood reasons your own read of your face is the least reliable one available.
- The mirror is backwards. You have spent your whole life looking at a reversed image of yourself, and you have grown comfortable with that version. A photo shows the true orientation, which is what everyone else sees — and because it is unfamiliar to you, it reads as slightly off. It is not off. It is just not the one you are used to.
- The phone lens is unkind up close. A selfie at arm's length uses a wide field of view. Whatever is nearest the lens gets enlarged — the nose, the chin — and the sides of the face fall away. Move the camera back a metre and the same face measures as more balanced, because the distortion is gone.
- You look for flaws; others look for expressions. When you study your own photo, you go straight to the feature you worry about. Other people process faces as a whole and mostly register expression and warmth. The pore you can see from across the room does not exist for them.
- You compare against a highlight reel. The faces you see most are lit, angled, filtered and selected. Comparing your unedited bathroom-mirror face against that is not a fair test, and the model does not make it — it compares you against 17,032 other unedited uploads.
- Mood moves the reading. The same face looks different to you on a good day and a bad one. The geometry has not changed. That alone is a reason to want a measurement that does not care how you feel today.
One practical consequence: if your own estimate and the test disagree, the test is the more reliable of the two, in either direction. That includes when it is kinder than you were expecting.
What genuinely changes a score
Every analysis estimates a potential score alongside the current one — what the same face could reach with the changeable things changed. Across 17,032 analyses the average current score is 5.59 and the average potential is 6.9. The gap averages 1.31 points; most people sit between one and two. 4.4% have two or more points of headroom, and not one person in the dataset has four.
So: you can move, and there is a ceiling. Which features are doing the moving is the useful part. Each analysis scores twelve, and the one people score worst on is not the one they expect.
| Feature | Average | Weakest feature for | Improvable |
|---|---|---|---|
| Jawline | 5.59 | 9.1% | Partly |
| Bone structure | 5.6 | 2.6% | No |
| Skin | 5.3 | 38% | Yes |
| Facial harmony | 5.58 | 0.5% | Barely |
| Chin | 5.39 | 8.4% | No |
| Symmetry | 5.62 | 2.6% | Barely |
| Cheeks | 5.42 | 6.6% | Partly |
| Eyes | 5.97 | 0.1% | Partly |
| Nose | 5.36 | 9% | No |
| Lips | 5.65 | 6.2% | Partly |
| Eyebrows | 5.94 | 1.9% | Yes |
| Forehead | 5.39 | 15% | Partly |
Skin is the weakest feature for 38% of people — far more than any other — and it has the lowest average of the twelve at 5.3. It is also the feature that responds most to what you do: sleep, sun protection, a consistent routine, and simply not photographing it under a bathroom bulb. Of everything on the list, only skin and eyebrows are marked fully improvable. Eyebrows are already the most common best feature (29.3%), and they are a half-hour fix. Bone structure, at the other end, is the weakest feature for just 2.6% and is not something a routine can touch.
The honest summary is that the levers are real and they are modest. If you are a 5 with a skin score dragging the average down, a year of boring, consistent care can plausibly show up in the number. It will not turn a 5 into an 8 — nobody in the data under a 5 has a potential above 7.5 — and anyone selling you that outcome is selling.
Figures use analyses from January 2026 onward (n=17,032), after the Dec 2025 model recalibration. Earlier analyses used a different scale.
What the score cannot tell you
It cannot tell you whether people like looking at you, because that depends on how you move, speak and pay attention — none of which is in a still photo. It cannot tell you whether you are attractive to any particular person, because individual taste varies far more widely than any average. It cannot see the things that make a face memorable rather than merely proportional, and plenty of the most striking faces in the world would not score in the eights here.
And it says nothing about worth. That sentence is on every page of this site because it is true, not because it is polite. A score is a model’s estimate of geometry from one photo, on one day, from one angle. It is useful for the same reason a bathroom scale is useful: it is consistent, it does not flatter, and it tells you whether the things you are doing are changing anything. It is not useful as a judgement, and it was never built to be one.
If you take the test and the number stings, the most accurate thing we can tell you is this: look at the distribution table again. The bulk of people are in the same place, they are getting on with their lives, and the people who love them did not check first.
Frequently asked questions
Am I pretty, according to the test?
The test gives you a number out of 10 and tells you where it sits among 17,032 real analyses. The average is 5.59 and the median 5.3. If you score a 5 you are exactly average, and average is the single most common outcome — 49.6% of everyone lands between 5.0 and 5.9. A 6 puts you in the top 25.1%, a 7 in the top 8.1%. Whatever the number is, it describes how one model reads the geometry of one photo. It is not a verdict on you.
Is the "am I pretty" test free?
Yes. Upload one photo and you get your overall score, your percentile and twelve feature scores after a one-tap Google or Apple sign-in. Nothing is posted anywhere and you decide who sees it.
What is a good score?
Anything above 5.59 is above average. Only 8.1% of people score 7 or higher, 3.3% score 8 or higher, and 0.2% reach 9. If you came in hoping for an 8 and got a 6, it is worth knowing that a 6 already beats three quarters of everyone measured.
Why do I look worse in photos than in the mirror?
Three separate things. The mirror shows you a reversed image, and you have looked at that version every day of your life, so the true orientation in a photo reads as slightly wrong. A phone held at arm's length uses a wide lens that enlarges the nose and chin and narrows the face. And you study your own photos far more closely than anyone else ever will. None of those change your face. They change what you are comparing it to.
Can I become prettier?
Within a range, yes, and the data is specific about the range. Every analysis estimates a potential score alongside the current one. Across 17,032 analyses the gap averages 1.31 points, 4.4% of people have two or more points of headroom, and nobody has four. The biggest lever is skin — it is the weakest feature for 38% of people and the one that responds most to what you do.
Does the score mean people find me unattractive?
No. It means a model assessed the geometry of one photograph and placed it on a scale. The people who like you are responding to your expressions, your voice, how you treat them, and a hundred other things a still image does not carry. Attractiveness in real life is not a face score, and the face score does not pretend to be more than it is.
Why did I get two different scores from two photos?
Angle, lighting and expression, in roughly that order. A face turned a few degrees off the camera measures as less symmetric; harsh light exaggerates skin texture; a wide smile changes the lower face. Take two or three well-lit, straight-on photos and treat the range as your real answer rather than the single lowest one.
Is this the same as PrettyScale?
It answers the same question with a different method. Ours scores twelve features, symmetry among them and then tells you where that score sits among real analyses, so the number comes with context. If you want to know whether a 6 is good, you can see exactly how many people score a 6.
Related tools
Attractiveness Test
What each score means, band by band.
Rate My Face
All 12 feature scores in one rating.
ChatGPT Face Rating
Why chat models tell everyone they are a 7.5.
Face Symmetry Test
The single strongest predictor of a score.
Face Shape Detector
Your shape, and how common it is.
AI Hairstyle Try-On
One of the levers that actually moves.
Research
Every number on this page, sourced.