Free IQ Test — Online Cognitive Assessment

Yes. The 12-item short form is a free IQ test with no payment step, no trial, and no card details — you can start it anonymously without creating an account or giving an email address, and you see your full result at the end rather than a blurred preview. The 60-item full form is free too. Premium adds the AI-written cognitive narrative, but nothing about the result is withheld to sell it: both forms report the same percentile, estimated IQ and six-domain profile, the short form just with a wider margin of error.

Yes — you can test your IQ here anonymously. The session starts immediately, and your answers are held against a temporary anonymous session rather than an account. If you decide afterwards that you want to keep the result, signing in attaches that finished session to your account automatically, so nothing is lost by starting without one. Signing in also lets you retake the test later and compare the two profiles.

On the clinical mean-100/SD-15 scale used by instruments like the WAIS-5, 100 is average by construction, and roughly 68% of people fall between 85 and 115. This test reports where your performance sits among people who took the same form here, as a percentile, and re-expresses that position on the same metric as an estimated IQ — so the 50th percentile corresponds to ≈ 100. It is an estimate of your position in this site’s sample, not an age-normed clinical IQ. A result near the middle is the single most common outcome, not a poor one.

The two phrases are used interchangeably in everyday speech, and both usually mean a test of general reasoning ability. They are not the same as a multiple-intelligences questionnaire, which is a different idea entirely — Gardner's framework describes eight kinds of strength (linguistic, spatial, musical, interpersonal and so on) and is a self-report strengths-finder, not a measure of reasoning. If you want the reasoning measure, this is the page for it; if you want the strengths framing, take the Multiple Intelligences test linked below instead.

No. A clinical IQ test (WAIS-5, Stanford-Binet 5) is administered 1-on-1 by a trained psychologist, takes 60–90 minutes, and produces age-normed standard scores meaningful for clinical, educational, and legal decisions. This screener measures similar cognitive constructs using self-administered multiple-choice items and expresses results as a percentile against our own sample, re-expressed as an estimated IQ (≈ N) — not an age-normed standard score. It is useful for personal exploration and practice; it is not a clinical instrument.

The range (e.g. 45th–69th percentile on the short form) shows how much your position could plausibly move. The short form carries ±12 percentile points and the full form ±7. Both are deliberately wide placeholders set from the literature on short, uncalibrated item banks, not a standard error measured on our own data; they will be replaced by measured values once enough sittings exist.

A single overall score averages across domains that can vary significantly within one person — high verbal with low spatial, or strong fluid with slow processing speed. The six-domain profile often reveals more actionable information than the overall percentile. The overall composite, and the estimated IQ derived from it, are still shown for a quick reference, weighted by each domain's correlation with general cognitive ability (g-loading).

Yes, a little, and mostly through familiarity with the item format rather than a change in ability: a meta-analysis of retesting found average gains of about a quarter of a standard deviation, roughly 4 IQ points (Hausknecht et al., 2007). There is no retake cooldown on this test, so leave a few weeks between attempts if you want a cleaner comparison. Evidence that brain-training games improve general reasoning is weak (Simons et al., 2016).

Intra-individual variation across cognitive domains is common and normal. The profile skew (high verbal + low spatial, or strong fluid + slow processing speed) often reflects educational history, occupation, and practice history more than fixed capacity. The subdomains are correlated but distinct — that correlation is what g-loading measures.

Processing speed is only meaningfully scored in timed sessions (Relaxed or Exam mode). In untimed sessions, processing-speed items are included but the speed component is not applied. In timed sessions, response time vs. the expected median for each item type applies a ±25% symmetric adjustment — fast and accurate scores higher; slow and accurate scores lower.

All cognitive screeners have some cultural and linguistic loading. Verbal items depend on English-language fluency; pattern-recognition and numerical items are more culturally neutral. The screener is translated into 6 languages, and each translation is checked by an AI reader against the English, not by native-speaker reviewers. Scores represent performance on this specific battery against this specific sample — not an absolute measure of innate cognitive capacity.

Scores at or below the 25th percentile are common by definition (one in four people score there). The screener identifies which specific domains are lower, and the action plan section suggests where targeted practice pays the highest return. Cognitive screener results should not be used to make high-stakes decisions about yourself or others without professional evaluation.

The 12-item short form is free and can be started anonymously — you do not need to create an account or hand over an email address to begin. Both the 12-item and 60-item forms are free. Creating an account lets you save your result and come back to it; Premium adds the AI-written cognitive narrative and retake trends. An anonymous session attaches to your account automatically if you choose to sign in afterwards.

The short form is 12 items and takes about 5 minutes. The full 60-item form takes roughly 15–25 minutes, depending on how long you spend on each item. You can pause and resume an unfinished session — your progress is checkpointed as you go.

There is no pass mark. The 50th percentile is exactly typical — half of test-takers score above it and half below — so a mid-range result is the single most common outcome, not a weak one. Higher percentiles are rarer by definition (the 90th means you performed above roughly nine in ten people in our sample), but the more useful read is your subdomain profile, which shows where your reasoning is relatively stronger or weaker. Because this is a percentile against our own sample — and the estimated IQ is that same position re-expressed, not a normed clinical IQ score — treat it as a snapshot for self-insight rather than a fixed label.

A self-administered online screener is necessarily less precise than a clinical instrument given one-on-one by a psychologist — which is why every figure, the estimated IQ included, comes with a range. The short form gives a directional estimate (±12 percentile points); the full form narrows it to ±7. Those ranges are placeholders, not a measured standard error, and no reliability study has been published for this test. Accuracy also depends on you: a quiet environment, no interruptions, and an honest, un-assisted attempt give the truest read. Use it for self-exploration and practice, not as a diagnostic or high-stakes measure.