Where this sits against Qoves, Umax and LooksMax AI.
We wrote this the way we wish they'd written theirs: every cell dated and sourced, and their genuine strengths stated. Qoves' ethnicity-stratified norms and clinical register are good, and we copied the good parts. What we won't copy is the 28-day wait, the hidden price, the server-side photo, or the metrics that can't be measured from a photograph.
| Aspect | Canthion | Qoves | Umax | LooksMax AI |
|---|---|---|---|---|
| What it is | Client-side measurement engine + evidence-graded protocol | Human-reviewed analysis membership, delivered as a report portal | Automated face-rating app | Automated face-rating app |
| Turnaround | Seconds | "Up to 28 days" stated; 1–5 days commonly reported1 | Seconds | Seconds |
| Price | $29 one-time Full · $59 one-time Complete (preview free) | ~$150/yr, shown only at checkout; +$80 expedite, +$25 haircut add-on2 | $3.99–$9.99/wk (≈$207–520/yr); consumables3 | $3.99–$9.99/wk; $39.99/yr reported; boost packs3 |
| Before paying | Real overview + headline measurements on your photo, unblurred | None | Scan free, result blurred until paid/referrals | Scan free, result blurred; 3-friend referral wall |
| Where the photo goes | Nowhere — never uploaded | Their servers; policy allows ML training on it; no stated retention period4 | Their servers; up to 1-year retention5 | Their servers; 30-day deletion stated6 |
| Same photo twice | Byte-identical, with fingerprint | Human-in-loop; not reproducible | Different scores reported on re-upload7 | Different scores reported on re-upload7 |
| Uncertainty shown | ±1σ on every value | No | No | No |
| Reference population | User selects (6 + pooled); never inferred | Ethnicity-stratified norms — genuinely implemented8 | Single template | Single template7 |
| Bony / profile / mm metrics | Given as their photographic equivalents (jaw taper & definition, chin–lip direction, dorsum, forehead slope), method stated, error bar shown | Reported (gonial angle 119°, E-line, "157 mm")8 | Sub-scores incl. "jawline" | Canthal tilt, thirds, symmetry, "harmony" |
| Overall score | Yes — 0–100 with a grade, construction printed: harmony 40 · skin 30 · definition 15 · symmetry 15; every proportion graded A–F | No single score; per-feature masculinity | Yes (0–100), method unpublished | Yes (0–100), method unpublished |
| Masculinity / femininity axis | Per feature and overall, in SD from your group's midpoint, with the typical man/woman marked | Per feature ("very masculine") | "Masculinity" sub-score | No |
| Face shape · feature cut-outs | Yes — shape with confidence; brows, eyes, nose, lips, jaw, cheeks, ears cut from your photo, front and ¾ | Yes | No | No |
| Citations | DOI per metric; versioned methodology | ~30 references in a 22-page PDF8 | None | None |
| Human review | By design, no — so it's instant, reproducible and reviewer-bias-free | Yes ("reviewed by specialists"; 2021: "written by doctors")9 — adds days and subjectivity | No | No |
| Tailored non-surgical plan | Hair, brows, facial hair / makeup, lips, skin — matched to your measurements, evidence-graded | General advice; add-on haircut for +$252 | Generic tips | Generic tips |
| Last product update | Live | Active (2026 changelog) | iOS 28 Jun 2024; umax.app now a domain-for-sale page10 | iOS 4 Apr 202410 |
| Age gate | 18+ on entry | 18+ in terms | 17+ iOS / "Everyone" Play5 | 12+ iOS10 |
| Refund | 14 days, no reason, use doesn't void it | Only if photos not yet submitted4 | App-store policy | App-store policy |
vs Qoves, specifically
Where they're better: a seven-year content library and a 1M-subscriber YouTube channel; ethnicity-stratified norms that customers explicitly appreciate; a clinical, non-judgmental register that reviewers say made them feel better; a named advisory board of surgeons; a human reads every report. If you want a person in the loop and don't mind waiting, that has value.
Where we're better: instant; on-device; every value with an error bar; bone-level metrics given as what a photo can actually support (jaw taper and definition instead of a gonial angle, chin–lip direction instead of an E-line, ratios instead of millimetres) with the method stated; price on the page; a refund that survives using the product; and a methodology page where their own site currently gives 512 and 521 landmarks, 100 and 160 and 200 tests, 12 and 14 and 18 sections. We copied their vocabulary on purpose. It isn't proprietary and it's the right register.
vs Umax
Umax was the category's viral moment (~$500k/month at peak, Fortune, July 2024). Its last iOS release was 28 June 2024; its founder moved on to a calorie-tracking app; umax.app now redirects to a domain broker. It still charges ~$4/week, blurs the result, returns different scores for the same face, and its own privacy policy claims it doesn't knowingly serve under-18s while being rated "Everyone" on Google Play. There is nothing to compare methodologically because it publishes no methodology.
vs LooksMax AI
More technical vocabulary than Umax (canthal tilt, thirds, symmetry) applied through a single template regardless of population, with the same blur-gate, a 3-friend referral wall, and hands-on testers reporting different results on re-upload. Its privacy policy is better than Umax's (30-day deletion) but has no children's section despite a 12+ rating. Last iOS release 4 April 2024.
Sources
- Qoves homepage FAQ ("up to 28 days"); Trustpilot reviews Apr 2026 reporting 1–2 and 5-day delivery without expedite.
- Qoves' own /insights/qoves-vs-umax page ("150 USD per year"); add-ons documented in Uncynical (Substack, Apr 2026) and Qoves T&Cs (Express Delivery clause).
- Apple App Store in-app-purchase lists for Umax (id 6471026798) and LooksMax AI (id 6474518292), Aug 2026; Real World Appeal pricing tests 2026.
- Qoves Privacy Policy and Terms & Conditions, qoves.com/legal, read Aug 2026.
- Umax privacy policy (umaxapp.wordpress.com/privacy); Google Play listing.
- looksmax.ai/privacy.
- Independent testing summarised at medium.com/@lennarthewriter (Jul 2026): four photos × three runs each.
- Leaked Qoves Comprehensive report (Sept 2022, 22 pp) — TOC, glossary, values; Qoves' Scribd sample reports V1–V3.
- MIT Technology Review, 5 Mar 2021 ("written by doctors"); qoves.com comparison pages 2026 ("reviewed by specialists").
- Apple iTunes lookup API, queried Aug 2026: Umax v1.4.3 (2024-06-28), LooksMax AI v1.0.21 (2024-04-04).
Competitor facts change. If any cell is out of date, email hello@canthion.com with a source and we'll fix it and note the change.