Free IQ Test — 4-Domain Cognitive Battery, About 20 Minutes

This edition has 4 sections. Numerical reasoning (12 items, 7 minutes) asks for the rule in a number sequence, and matrix reasoning (10 items, 7 minutes) for the shape that completes a 3×3 grid — 22 items in all, each with 4 options, advancing automatically when time runs out. In working memory you type digits shown one at a time, forward from 4 to 8 and backward from 3 to 7, one try per length; each direction ends at the first miss. Processing speed gives you 60 seconds to judge whether two strings of 4 symbols match. The Korean edition adds verbal reasoning (12 items, 5 minutes), which depends on Korean vocabulary.

Scoring is simple. Each reasoning section turns the share of correct answers into a 0–100 score, and unanswered items count as wrong. Working memory converts the sum of your longest forward and backward spans (5–15) to 0–100, and processing speed does the same with correct minus wrong answers, where 40 earns the full 100. The composite is a weighted average of these domain scores in which each reasoning section carries 2 times the weight of memory or speed. The reasoning items are the same on every attempt, while the memory digits and speed symbols are generated at random each time.

The result screen first shows your composite out of 100 with a ‘Compared with others’ percentile; the detailed report adds the domain scores and an IQ conversion. The comparison group is the distribution of first-attempt scores from Soundary users who saved their results to an account. Retakes are excluded and the top and bottom 2% are trimmed, and the measured distribution is used only once at least 30 people are in it; before that, a provisional estimate stands in. The IQ figure expresses how many standard deviations your composite sits from that group’s mean on a scale with mean 100 and SD 15, capped between 40 and 160.

So this number is not the IQ you would get in a clinic. The Wechsler Adult Intelligence Scale (WAIS-IV) compares you with a large standardization sample split into age bands from 16 to 90; here the reference group is self-selected online users, pooled across ages and across language editions with different section sets. Because abilities such as processing speed slow on average across adulthood (Salthouse, 1996), your position among people your own age could differ. With the composite capped at 100 and the sample still small, the upper range is also compressed. Read it as a rough position among people who took this test here.

The numerical and matrix items (plus the verbal items in the Korean edition) are identical on every attempt, so a second try is easily lifted just by remembering the questions. A meta-analysis of retest effects on cognitive ability tests (Hausknecht et al., 2007) found larger gains when the identical form was repeated, and a later meta-analysis pooling 122 studies (Scharfen et al., 2018) found that scores stopped rising after the third administration. If your reasoning scores go up from the second attempt on, consider familiarity with the items before concluding that your ability has changed.

The memory digits and speed symbols, by contrast, are generated at random every time, so there is no item memory to inflate them, although familiarity with the task format can still nudge them up. To track change, compare these two domains on the same device. Processing speed is correct minus wrong answers in 60 seconds, so it also depends on whether you tap a phone or click a mouse and on how quickly your screen responds. The shape of your domain profile, and how it differs from your own earlier records taken under similar conditions, tells you more than any single composite.