What does your minimum actually cost?
Everyone on this page is available — until you start having standards. Set yours and watch the crowd thin. Real distributions, not vibes.
Watch the background while you drag.
At 1, nobody misses. There is no first no.
AI composites, not real people. Your taste is allowed to disagree.
Where the numbers come from
Everything here runs on a fixed simulation of 60,000 people of the gender you picked, inside your age range. Each simulated person gets an age, a relationship status, a consensus attractiveness score, a height, an income, and an education level, drawn from published distributions and correlated the way those traits correlate in real life (better-looking people earn somewhat more, income tracks education, and so on). Your filters then eliminate people one stage at a time. Same seed every time, so the same settings always give the same answer. The faces in the background are a 1-in-1,000 sample of that simulation: each portrait stands for one simulated person, and it dims when your filters cut them.
Scores are holistic — face, grooming, and style together, which is also how the example portraits are drawn. Pick "either" and the pool blends men and women 50/50, each rated on their own curve.
The looks curves are the important part, and they are not symmetric. When men rate women's attractiveness, the ratings form a roughly normal bell curve centered near the middle. When women rate men, the curve shifts hard to the left: in OKCupid's dataset, women rated about 80% of men as below-average looking. We model both curves accordingly, which is why "a minimum of 7" costs dramatically more when you're shopping for men than for women.
| Minimum score | Top % of women (as men rate them) | Top % of men (as women rate them) |
|---|
The reciprocity step, when you turn it on, uses what online-dating studies keep finding: people mostly match near their own level and aim about 25% above it. A candidate two points above you is possible but unlikely; four points above you is a rounding error. Self-ratings run about a point above how strangers score the same face, hence the honesty checkbox.
| Source | Used for |
|---|---|
| OKCupid data blog, 2009 | The two rating curves; the 80%-below-average finding |
| Pew Research Center, 2022 | Share of people single by age and gender (63% of men under 30; 34% of women) |
| US Census / ACS, 2023 | Population by age; individual income; education levels |
| CDC NHANES | Male height distribution (mean 5'9", SD about 3") |
| Bruch & Newman, Science Advances, 2018 | How far above their own desirability people aim (~25%) |
| Epley & Whitchurch, 2008 | Self-rating inflation of about one point |
The portraits, including the background crowd, are AI-generated composites drawn to stand roughly where consensus ratings put each score. No real person appears on this page, and reasonable people will disagree with any given face's placement — that disagreement is itself the point of the caveat below.
Honest caveats: these are consensus scores, and raters disagree with each other almost as much as they agree — the pool of people you'd personally find attractive is bigger than any consensus cutoff. Income and education filters use national shares adjusted for age, not your city. Singleness is treated as independent of looks, which is generous to high bars. None of this measures charm, humor, or whether they text back.