Sequences
Find the rule a series is following and continue it — including series that describe themselves.
The reasoning test
A score you can actually check.
Most in-app IQ tests ask a model to guess a number, which is why the number changes if you take the same test twice. This one does not work that way.
How it is built
A fixed bank of questions with known answers. Each sitting samples fifteen of them, two from every domain and the rest weighted to the harder end, then shuffles the options. Scoring is arithmetic: correct answers weighted by difficulty, a small speed factor, mapped onto a familiar 100-centred scale.
- 42questions in the bank, 15 in a sitting
- 6domains — sequences, verbal, deduction, pattern, numeric, case reasoning
- 0guesswork: the same answers always produce the same result
- 1:1every question you miss comes back with the reasoning spelled out
It is called a reasoning index, not an IQ. A clinical score is standardised against a population under supervised conditions; fifteen questions on a phone is not that, and saying otherwise would be dishonest.
The six domains
What each one is measuring.
Verbal reasoning
Analogies where the obvious near-synonym is the wrong answer and the relation is the right one.
Logical deduction
What follows and, more often, what does not — necessary versus sufficient, and alibis that only look complete.
Pattern recognition
Codes, grids and spatial questions: the shifted cipher, the magic square, the painted cube.
Numerical reasoning
Arithmetic with a trap in it — the bat and the ball, the lily pads, the average speed that is not the average.
Case reasoning
A short scenario with numbers in it: pricing, worker-days, overlapping sets, and a base rate most people ignore.
Seen in the app
Fifteen questions, one at a time.
Each question is timed by difficulty. Leave halfway and it resumes where you stopped; finish and the report is immediate, because nothing has to be asked of a model.
Take it once, then again next month.
Every sitting samples a different fifteen, so a second run is a new test.