How the test works
IQ Octopus is a 25-puzzle reasoning test that gives you an estimated IQ with an honest range around it. This page explains how the puzzles are built, how your answers become a number, and what that number cannot tell you.
The quickest way to understand the method is to see it. 25 puzzles, about 12 minutes, no sign-up.
What you do
You solve 25 puzzles in five sections, always in the same order. A short title card introduces each section. Within a section the puzzles start easy and get harder. Most people finish in about 12 minutes. If you reach 20 minutes, the test submits whatever you have answered.
| Section | Puzzles | What you do |
|---|---|---|
| Pattern Recognition | 5 | Pick the tile that comes next in a row. |
| Abstract Reasoning | 9 | Complete a 3x3 grid with one cell missing. |
| Spatial Reasoning | 5 | Tell a rotation from a mirror image; find the odd one out. |
| Quantitative Reasoning | 3 | Find the next number in a series. |
| Logical Reasoning | 3 | A is to B as C is to which figure? |
You get no feedback on whether an answer was right. That keeps one puzzle from affecting how you approach the next. Once you confirm an answer the test moves on, and there is no back button. On a phone, the first tap selects an option and the second tap (or the Next button) confirms it, so a slip of the thumb does not count.
If you want to see each puzzle type before you start, the practice page has one worked example per section.
Why shapes and numbers instead of words
Every puzzle on IQ Octopus is made of shapes, positions and numbers. There is no vocabulary, no reading passage and no general-knowledge question. That choice follows a long tradition in testing: Raven's Progressive Matrices, first published in 1938, use nonverbal matrix puzzles and are widely used as a measure of fluid, culture-reduced reasoning (Wikipedia: Raven's Progressive Matrices).
Psychologists separate fluid intelligence, which is reasoning with new material, from crystallized intelligence, which is what you have learned. Raymond Cattell drew that line in 1963 (Wikipedia: Fluid and crystallized intelligence). A test built from unfamiliar visual rules leans toward the fluid side. It is not perfectly free of culture or schooling, and no test is. We explain the limits in our guide to culture-fair IQ tests.
Every test is generated fresh
There is no fixed set of 25 questions. Each puzzle comes from a rule template, such as “the shape turns a quarter turn and gains one dot at each step”, and a random seed picks the shapes, the rules and the wrong options for your attempt. Two people never see the same test, and a retake gives you new puzzles.
This matters for three reasons:
- No answer key exists. Nobody can post the answers online, because the answers change every time.
- Your browser never receives the answer. Puzzles are built on our server and sent to you without the correct option. Scoring happens on the server after you finish.
- Wrong options are built to be plausible. Most wrong options follow all of the rules except one, so you cannot find the answer by spotting the odd tile out.
Try one · Logical Reasoning
Which tile completes the second pair?
Tap the option you think is right, or show the answer.
Show the answer
The missing panel shows a small solid triangle pointing right, with small solid shapes: a circle at the bottom and a circle at the left side (option 2).
How to solve it
- Compare the first two panels and name exactly what changed.
- The whole figure turns a quarter turn counterclockwise.
- The large shape becomes solid.
- Nothing else changes.
- Apply the same changes to the third panel: a small hatched triangle pointing down, with small solid shapes: a circle at the left side and a circle at the top.
- So the missing panel shows a small solid triangle pointing right, with small solid shapes: a circle at the bottom and a circle at the left side.
That was 1 puzzle. The full test has 25.
How your answers become a number
IQ Octopus uses a Rasch model, also called a one-parameter item response model. It is simpler than it sounds. Each puzzle has a difficulty value. You have an ability value on the same scale. The model says that the chance you answer a puzzle correctly depends on the gap between the two: well above the puzzle's difficulty, you will usually get it right; well below, you will usually miss it; right at it, it is close to a coin flip.
After you finish, we find the ability value that best explains your exact pattern of right and wrong answers. Getting a hard puzzle right tells us more than getting an easy one right, so two people with the same number of correct answers can receive different estimates. Unanswered puzzles count as wrong.
The estimate starts from a neutral assumption centered on average and moves as far as your answers justify. That keeps a run of lucky guesses, or a perfect score on a short test, from producing an extreme number the evidence cannot support.
From ability to IQ
Modern IQ scales are set so the mean is 100 and the standard deviation is 15 (Wikipedia: Intelligence quotient). We place your ability estimate on that scale: an ability one standard deviation above average becomes 115, two below becomes 70. Your percentile, the share of people expected to score below you, comes straight from the normal curve. The IQ Octopus scale runs from 65 to 160; results beyond either end are reported at the end of the range.
Where difficulty values come from
Each template has difficulty settings, such as how many rules are combined, and each setting starts with a difficulty value set by design. We store every response, and as data accumulates we refine those values from how people actually perform. The method stays the same; the numbers it uses get better over time.
The confidence range
No test measures anyone exactly. Even clinical full-scale IQ scores carry a standard error of a few points, which is why test manuals report a confidence interval, often around plus or minus 5 points at 95%, instead of a single exact value (Wechsler, 2008).
Your IQ Octopus result does the same. Next to your estimate you see the range your true score most likely falls in, at 95% confidence. A shorter test has a wider range than a long clinical battery, and we would rather show you that than hide it behind a single confident number.
The five-area breakdown
The full result also scores each section on its own: Pattern Recognition, Abstract Reasoning, Spatial Reasoning, Quantitative Reasoning and Logical Reasoning. Each gets a band from 1 to 10 with a plain label and one sentence about what it means. Sections have between 3 and 9 puzzles, so these bands are rougher than your overall estimate. Read them as a profile of where you were strongest, not as five separate scores.
What about speed?
We record how long you spend on each puzzle and describe your pace in the result. Pace does not change your estimated IQ. Only which puzzles you got right does. The time limit is generous so that careful thinkers are not punished, but unanswered puzzles count as wrong, so it is worth answering every one.
Why the result costs money
Taking the test is free and needs no sign-up. The full result costs $4.95 one time. Scoring and hosting cost money, and we do not sell your data or run ads, so the people who want their result pay for it. There is no trial that turns into a subscription, no account and nothing that renews. If you are unhappy, ask within 7 days and we refund you in full. See pricing and the refund policy.
We do not show a free rough score before payment. A partial number would be less accurate than the full result, and we would rather not hand you a figure we do not stand behind.
What it cannot do
- It is not a clinical or diagnostic instrument. It is not given by a psychologist and cannot diagnose anything.
- It cannot be used for official decisions. That includes Mensa admission, school placement, disability assessment and employment screening. Mensa, for example, requires an approved, supervised test (Mensa International).
- It does not apply age norms. Clinical IQ scores compare you with people your own age. IQ Octopus does not ask your age and is designed for adults.
- We do not quote an accuracy figure. We have not run a study comparing IQ Octopus scores with clinical tests, so we make no claim about how closely they agree.
- It cannot control your conditions. You take it on your own device, unsupervised. Noise, interruptions, tiredness and a small phone screen can all pull a score down.
- It cannot tell you what your life will hold. Research links IQ to outcomes like school results across large groups, but a group correlation says little about any one person. Our guide to what an IQ test measures covers what the research does and does not show.
Getting the most accurate estimate
- Pick a quiet place where you can spend 12 to 20 minutes without interruption.
- A laptop or tablet shows the puzzles larger, but a phone works fine.
- Work carefully rather than quickly. Pace does not raise your score.
- Answer every puzzle. A considered guess is better than leaving one blank.
- Take it once, seriously. Retakes tend to score a little higher because of practice, by about a quarter of a standard deviation on average (Hausknecht et al., 2007).
Sources
- Intelligence quotient. Wikipedia. en.wikipedia.org/wiki/Intelligence_quotient
- Raven's Progressive Matrices. Wikipedia. en.wikipedia.org/wiki/Raven%27s_Progressive_Matrices
- Fluid and crystallized intelligence. Wikipedia. en.wikipedia.org/wiki/Fluid_and_crystallized_intelligence
- 68-95-99.7 rule. Wikipedia. en.wikipedia.org/wiki/68%E2%80%9395%E2%80%9399.7_rule
- Wechsler, D. (2008). WAIS-IV Technical and Interpretive Manual. Pearson.
- Hausknecht, J. P., Halpert, J. A., Di Paolo, N. T., & Moriarty Gerrard, M. O. (2007). Journal of Applied Psychology, 92(2), 373-385. doi.org/10.1037/0021-9010.92.2.373