Forty-five picture puzzles, about twenty minutes, nine per cognitive domain. The most precise of our three culture-fair forms: an IQ score with a ±4 interval and a profile of your five domains.
Same puzzles, same five domains. More questions means a narrower interval around your score, and the Full form adds a per-domain profile.
Most online IQ tests reuse one fixed set of pictures, so the answers circulate and the score drifts upward. This test builds its puzzles from a random seed each time you start: the rules stay the same kind (rotation, fill, count, shape, position), the figures never repeat.
Every item has six choices and exactly one correct answer. Each wrong choice breaks the rule in one specific way, and your report tells you which way — so you learn something from every item you miss.
No vocabulary, no general knowledge, no reading beyond a one-line instruction. Your result depends on reasoning alone, which is why the same test is served in twenty languages without any translation of the puzzles.
Answer keys never leave the server, the time bank rewards steady work over rushing, and the result states its own precision: a range for the Quick form, a confidence interval for Normal and Full.
Nine questions per domain in the Full form, enough for a score per domain next to your overall IQ.
3×3 matrices and figural analogies: find the rule that completes the grid.
Figure series and odd-one-out: see what changes from step to step, or what does not belong.
Picture completion: pick the piece that fits the gap in a tiled pattern exactly.
Number grids and sequences: the arithmetic rule behind the numbers, no formulas needed.
A row of figures is shown for three seconds; then you answer a question about it from memory.
From start to result in about twenty minutes
Quick, Normal or Full: 15, 30 or 45 questions. Every length covers the same five domains.
Six choices per item, one correct. You start with 40 seconds and earn extra time with each answer.
The puzzles are generated from a seed and scored where the answer key lives, so results cannot be gamed.
IQ with its interval, percentile, and for every item the rule and why each wrong choice is wrong.
Each item is worth its difficulty (1 to 5 points), so a hard matrix counts for more than an easy series. Your weighted raw score is converted to a deviation IQ with a mean of 100 and a standard deviation of 15, then to a percentile.
We report the precision with the score. A test's reliability sets the width of its confidence interval: about ±9 IQ points for the 15-item Quick form, ±5 for the 30-item Normal form and ±4 for the 45-item Full form. The Quick form therefore shows a range rather than a single number.
The norms are provisional: they follow the typical difficulty of this item type and are replaced by norms from our own completed runs once each length has enough of them. Runs that are far too fast, or below chance level, are flagged rather than scored.
For an assessment with verbal and knowledge items too, see our standard IQ test or the Stanford-Binet assessment. The scoring method is described on the methodology page.
This culture-fair test is the non-verbal one. The others follow a named model and include verbal and knowledge items.
Yes. All three lengths are free and you see your result immediately. Nothing is hidden behind a payment, and no e-mail address is needed to see your score.
The score is as precise as the length allows and we say so: the Quick form gives a range of about ±9 IQ points, the Normal form ±5 and the Full form ±4 (95 % intervals). Norms are provisional until each form has enough completed runs; no online test replaces an individually administered assessment by a psychologist.
A culture-fair test measures reasoning, not vocabulary or schooling. Matrices, series, odd-one-out and missing-piece items work the same in every language and are what fluid-reasoning tests such as Raven's matrices rely on.
Yes. Every run is generated fresh from a new seed, so you never see the same puzzles twice and there is nothing to memorise. Practice still helps a little, so the first attempt is the most representative.
Short on time or just curious: Quick. Want a score you can quote with its interval: Normal. Want the most precise score and a profile of your five domains: Full. You can move to a longer form from your results page at any time.
The Standard, Mensa-style, Stanford-Binet, CHC and WAIS assessments each follow a named model and include verbal and knowledge items. This test is the non-verbal, generated one: the right choice when you want a result that does not depend on language, and the quickest to retake.