Looksmax AI: Looksmaxxing Analysis With a Real Ceiling

Upload one photo. Get your rating, twelve feature scores, and a potential score calibrated against 17,032 real analyses — including which features actually respond to effort and which do not. Free — one tap to sign in.

Seconds Free, one-tap sign-in Potential measured, not promised

1 / 1Free Scans Remaining Today

Upload a clear photo

Click here or drag and drop your image

Choose one front-facing photo in even lighting

What looksmaxxing actually is

Looksmaxxing is the practice of deliberately improving your appearance, usually with a rating as the scoreboard. At its reasonable core it is nothing controversial: skincare, grooming, a haircut that suits your face, sleep, body composition and learning how to be photographed. Those things work, and we can see them working in the data.

The controversial end is the claim that any face can be rebuilt into any other face — that a 4 becomes an 8 with enough discipline, or enough surgery. That claim is what sells guides, supplements and consultations. It is also a claim we can test, because every analysis we run estimates a realistic potential score alongside the current one, and we have 17,032 of them.

A looksmax AI, in the plainest sense, is a tool that does three things: rates the face, breaks the rating into features, and estimates how far it can move. The third one is where most tools quietly flatter you. This page is what happens when you do not.

What the data says is achievable

The average current score across 17,032 analyses is 5.59. The average potential is 6.9. The gap between them — the amount of improvement the model considers realistic for a given face — averages 1.31 points, with a median of 1.3.

The spread is narrow. The 10th percentile of headroom is 0.7 points and the 90th is 1.7. Only 4.4% of people have two points or more of headroom. 15 people in the entire dataset have three. Nobody has four.

Headroom (potential minus current)	Share of people
Under 0.5 points	2.8%
0.5 – 1.0 points	14.4%
1.0 – 1.5 points	36.7%
1.5 – 2.0 points	41.7%
2.0 + points	4.4%

The single most common outcome is a gap of 1.5 – 2.0 points (41.7% of people). Read against the looksmaxxing pitch, that is deflating. Read as a plan, it is useful: a point and a half is real, visible improvement, and almost everyone has it.

Figures use analyses from January 2026 onward (n=17,032), after the Dec 2025 model recalibration. Earlier analyses used a different scale.

The ceiling tracks the starting point

Current score and potential score correlate at 0.94. In practice that means where you start almost entirely determines where you can end up. The gap is larger in raw points at the bottom of the scale, but the ceiling stays low:

Current score	Share	Avg. headroom	Avg. potential
Below 4.0	0.5%	+2.02	5.17
4.0 – 4.9	24.8%	+1.75	6.47
5.0 – 5.9	49.6%	+1.36	6.66
6.0 – 6.9	17%	+0.87	7.53
7.0 – 7.9	4.8%	+0.73	8.33
8.0 – 8.9	3.1%	+0.39	8.84
9.0 +	0.2%	+0.32	9.51

The number the community does not quote

Of the 4,307 people scoring below 5, 0% have a potential of 8 or higher. The highest potential recorded for anyone in that group is 7.5. Someone in the 4.0 – 4.9 band averages 1.75 points of headroom and lands around 6.47; someone already at 8.0 – 8.9 gains only 0.39 but finishes near 8.84. The people who end up with the highest ceilings are the people who started highest.

Figures use analyses from January 2026 onward (n=17,032), after the Dec 2025 model recalibration. Earlier analyses used a different scale.

Read the full potential study Convert your score to PSL

What responds to effort, and what does not

The potential estimate is an aggregate. The practical question is which of the twelve scored features it is made of. Each feature below is tagged with how much it responds to anything short of surgery, alongside its average score and how often it is a person's weakest feature.

Feature	Improvable?	Avg. score	Weakest feature for
Skin	Yes	5.3	38%
Eyebrows	Yes	5.94	1.9%
Jawline	Partly	5.59	9.1%
Cheeks	Partly	5.42	6.6%
Eyes	Partly	5.97	0.1%
Lips	Partly	5.65	6.2%
Forehead	Partly	5.39	15%
Facial harmony	Barely	5.58	0.5%
Symmetry	Barely	5.62	2.6%
Bone structure	No	5.6	2.6%
Chin	No	5.39	8.4%
Nose	No	5.36	9%

Weakest-feature shares are tie-split: when several features tie for lowest, each gets an equal fraction of the credit. Figures use analyses from January 2026 onward (n=17,032), after the Dec 2025 model recalibration. Earlier analyses used a different scale.

Genuinely improvable

Skin and Eyebrows. Skin is the lowest-scoring feature in the dataset (5.3) and the weakest feature for 38% of people — and it is the one that responds most to sleep, sun protection and consistency. Eyebrows are the opposite case: already high-scoring (5.94), the best feature for 29.3% of people, and fixable in an afternoon.

Partly improvable

Jawline, Cheeks, Eyes, Lips, Forehead. Body-fat percentage changes how the jaw and cheeks read. Hair, brows and how you hold the camera change how the forehead and eyes read. None of these are structural changes; they are changes in what the structure looks like.

Barely improvable

Facial harmony and Symmetry. These are relationships between features rather than features themselves. They move a little when the things around them move, and a lot when the camera angle changes — which is a photography effect, not a face effect.

Not improvable

Bone structure, Chin, Nose. Bone structure has the second-highest correlation with the overall score (0.96) of anything we measure, and it does not respond to mewing, chewing devices or any routine sold as a "bone" fix. This is the category the extreme end of looksmaxxing targets, and it is exactly the category where the free methods do nothing.

Read the full feature study

Softmaxxing vs hardmaxxing, honestly

The community splits its methods into two tiers. Softmaxxing is everything non-invasive: skincare, grooming, haircut, brows, body composition, sleep, posture, and learning your angles. Hardmaxxing is surgical or otherwise invasive: jaw and chin implants, rhinoplasty, orthognathic surgery, and the pharmacological routines that circulate alongside them.

Line those two tiers up against the table above and the picture is simple. Softmaxxing targets the features marked improvable and partly improvable. Hardmaxxing targets the features marked not improvable. The measured headroom — 1.31 points on average, 1.7 at the 90th percentile — lives almost entirely in the first tier, because that is where the improvable features are.

We are not in a position to evaluate surgical outcomes, and we do not pretend to be. What the data can say is narrower and more useful: the transformation that hardmaxxing is sold on — several points, a change of tier, a different life — does not appear anywhere in 17,032 analyses. 0 people have four points of headroom. If someone is offering you four, ask what their number is based on.

None of this is a judgement about wanting to look better. It is a budget. Spend the effort where the return is measurable.

How to track progress over time

A single rating is a snapshot. Looksmaxxing is a process, and the only honest way to know whether it is working is to re-rate the same setup and compare feature by feature. The setup matters more than the interval.

Fix the conditions. Same room, same time of day, same distance from the camera, same neutral expression. Write them down. Lighting alone can move a skin score more than a month of effort.
Compare features, not the headline. The overall score is an aggregate and moves slowly. Skin, eyebrows and jawline are where change shows first, and they are the features you are actually working on.
Use a realistic interval. Four to eight weeks. Skin turns over on roughly a monthly cycle; body composition changes take longer to show in the face. Weekly re-rating measures photo noise, not progress.
Treat the potential score as the target. It is the model's estimate of the realistic ceiling for your face. Closing the gap to it is the whole project; exceeding it is not something the data supports planning around.
Keep the photos. Two scores are less informative than two photos taken under the same conditions. If the number moved and the photos look the same, suspect the number.

Frequently asked questions

What is looksmax AI?

A face analysis tool used for looksmaxxing: it scores your face, breaks the score into individual features, and estimates how far the rating could realistically move. Ours does the same three things, with one difference — the potential estimate is calibrated against real analyses rather than against what a community wants to hear.

How much can looksmaxxing actually improve a face rating?

Across 17,032 analyses the average gap between a current score and the realistic potential is 1.31 points, with a median of 1.3. Ten percent of people have less than 0.7 points of headroom and ninety percent have 1.7 or less. Only 4.4% have two points or more, 15 people in the whole dataset have three, and nobody has four.

Can a 4 become an 8 with looksmaxxing?

Not in our data. Of the 4,307 people scoring below 5, 0% have a potential of 8 or higher, and the highest potential recorded for anyone in that group is 7.5. The gap is largest at the bottom in raw points, but the ceiling still tracks the starting point.

What is the looksmaxxing scale?

Most looksmaxxing communities use either a decile scale (1–10) or the PSL scale, which compresses the same range and treats anything above 6 as exceptional. Our rating uses 1–10 and reports the real share of people at each score, so a number can be read as a percentile rather than as a label. The PSL rating page converts between the two.

Which facial features respond to looksmaxxing?

Of the twelve features we score, two are marked as genuinely improvable: skin and eyebrows. Five respond partly — jawline, cheeks, eyes, lips and forehead, mostly through body composition, grooming and photography. Symmetry and facial harmony barely move. Bone structure, chin and nose do not respond to anything short of surgery.

Does softmaxxing or hardmaxxing work better?

Softmaxxing — skin, grooming, hair, fitness, sleep, photography — is where almost all of the measured headroom lives, because the improvable features are the soft ones. Hardmaxxing targets the features marked as not improvable, and the potential estimate already assumes a realistic ceiling that surgery is marketed as exceeding. Our data cannot evaluate surgical outcomes; it can only tell you the gap is smaller than the pitch implies.

Is the looksmax AI analysis free?

Yes. The rating, the twelve feature scores and the potential estimate come with the free analysis; you sign in first with Google or Apple.

How do I track looksmaxxing progress with AI?

Re-rate the same setup — same room, same light, same distance, same neutral expression — every four to eight weeks, and compare feature scores rather than the headline number. A one-point movement on skin is a real signal; a 0.2 movement on the overall score between two differently-lit photos is not.