How the AI attractiveness test works
- Your browser checks the photo first. Before anything is uploaded, the page measures brightness and sharpness. Very dark or blurry photos are stopped right there, because no rating of them would be fair. The photo is also shrunk to 1,024 pixels and re-saved, which strips hidden data such as GPS location.
- A vision AI looks at it once. The photo goes to Grok, an image-understanding model from xAI, with a fixed set of instructions. It first checks that the picture shows one real adult face. Group photos, cartoons, screenshots of photos, and anyone who might be under 18 are politely declined.
- It scores against the general population. The AI is told that 5.0 is a typical adult and that most people land between 4 and 6.5, so scores aren't inflated. It rates the face separately from the photo quality, then gives you three strengths and three tips.
- The photo is gone. Nothing is saved on our side, and the AI provider runs with zero data retention. You can read the details in the privacy policy.
What your score means
The scale is calibrated like a percentile, not like a school grade. A 6 isn't a failing mark. It's comfortably average, which is where the largest group of people is.
| Score | What it means | How common |
|---|---|---|
| 9.0–10 | Exceptional, professional-model level | about 1 in 100 |
| 8.0–8.9 | Very attractive | about 1 in 20 |
| 7.0–7.9 | Clearly above average | about 1 in 7 |
| 4.0–6.9 | The broad average range, where most people land | most adults |
| Below 4.0 | Rare, and usually a sign of a hard photo rather than a face | uncommon |
“How common” is the target the AI is asked to follow, not a measured survey of real people.
The five sub-scores
- Facial harmony: how well proportions and symmetry fit together. Psychologists have found that average-shaped and symmetrical faces tend to be rated attractive across cultures (Rhodes, 2006).
- Features: the eyes, nose, lips, and jawline taken one at a time.
- Skin: how clear and even the skin looks in this photo. Lighting changes this a lot.
- Grooming: hair, facial hair, and styling, the part that is easiest to change.
- Expression: warmth, confidence, and how relaxed you look. A small genuine smile usually helps.
You also get a separate photo quality score for light, focus, and angle. If it is low, a better photo is the quickest way to a higher result.
How to take a photo that's fair to you
- Don't hold the phone too close. A camera about 12 inches from your face makes the nose look roughly 30% wider than at 5 feet (Ward et al., 2018). Stretch your arm out fully, or better, ask a friend or use a timer from 4–5 feet and zoom in a little.
- Face a window. Soft daylight from the front evens out skin and fills shadows. Ceiling lights straight above cast shadows under the eyes and nose.
- Camera at eye level. Shooting from below exaggerates the chin and neck. Slightly above eye level is usually the most flattering.
- Relax your face. Breathe out, think of something you like, and let a small smile happen rather than forcing one.
- One face, no filters. Beauty filters and heavy edits can make the AI decline the photo or score it unpredictably.
What an AI score can't tell you
Attractiveness in real life is much more than a still photo: how you move, talk, laugh, and treat people matters, and tastes differ between people and cultures. An AI model also learns from images collected online, so it can reflect the beauty standards and biases of that data. Use the score as a fun experiment and the tips as a way to take better pictures, not as a verdict on yourself. If thoughts about your appearance are making you unhappy, talking to someone you trust or a professional can help much more than any rating.
AI attractiveness test FAQ
How accurate is an AI attractiveness test?
It is consistent, but it isn't the truth about your looks. In our testing the same photo scored within half a point when rated again, so the AI is steady. What it measures, though, is one model's opinion of one picture. Different photos of the same person can differ by a point or more, and people disagree with each other about faces too.
Why did I get a lower score than I expected?
Most often it's the photo: overhead or harsh light, a close-up selfie that distorts the nose and face, a low camera angle, or a tired expression. Check the photo quality number and the tips under your result, then try a photo taken at arm's length or further away, at eye level, facing a window.
Is my photo stored or used to train AI?
No. The photo is resized in your browser and sent once to the AI service. Our server never writes it to disk or logs it, and our account with the AI provider (xAI) has zero data retention switched on, which means they don't store requests either. The site checks that confirmation on every single request and refuses to show a result if it's missing. Nothing is used to train models.
Can I rate someone else's photo?
Only your own, or someone who has said yes. Rating other people's faces without asking is unkind and, in some places, against the law. Photos of anyone under 18 are not allowed and are rejected.
Why isn't the test available in Illinois, Texas, and Washington?
Those states have specific laws about facial and other biometric information. We don't create face templates, but we've chosen to keep the tool unavailable there rather than risk getting it wrong.
How many free tests do I get?
Three per day. The limit keeps the site free for everyone, because each rating costs us a small fee for the AI. It resets at midnight UTC.
Does the AI judge race, age, or weight?
It is instructed never to guess or comment on race, ethnicity, religion, health, or body weight, and to rate the face against adults in general. Like every AI model, it may still carry biases from the data it learned from, which is one more reason to treat the score lightly.
Please note
This tool is for adults (18+) and for entertainment only. Upload only your own photo. Scores are generated by an AI, can be wrong or inconsistent, and aren't advice of any kind. See the terms.
Sources
- Rhodes, G. (2006). The evolutionary psychology of facial beauty. Annual Review of Psychology, 57, 199–226. doi:10.1146/annurev.psych.57.102904.190208
- Ward, B., Ward, M., Fried, O., & Paskhover, B. (2018). Nasal distortion in short-distance photographs: the selfie effect. JAMA Facial Plastic Surgery, 20(4), 333–335. doi:10.1001/jamafacial.2018.0009
- Scoring model: Grok (grok-4.20) by xAI, image understanding via the xAI API with zero data retention.