Twelve short tasks that look like games. Employers watch how you play, not whether you win. The only useful preparation is knowing what each one records.
OpenThinkr is not affiliated with Pymetrics or Harver. These are unofficial practice versions built to match the publicly described mechanics.
Pymetrics is a hiring assessment made of short games rather than questions. Employers send it out early in an application, usually straight after the CV screen and before anyone human has looked at you properly.
It was built by a New York company of the same name and bought by the hiring software firm Harver in 2022. You will still see it called Pymetrics almost everywhere, including by the employers who use it, which is why this site does too.
The idea behind it is that how you behave under a small amount of pressure says more about fit than what you claim about yourself on a form. So instead of asking whether you are good under risk, it gives you a balloon that pays more the longer you pump it and pops without warning, and watches what you do. Instead of asking whether you are attentive, it makes you press a key for three minutes while a target keeps changing.
Twelve games, roughly 25 to 35 minutes in total according to published employer and university careers guidance. Some employers switch on a further four numerical and logical games, which makes it sixteen.
What it is not is an IQ test, a personality quiz or a knowledge test. There is nothing to revise. That is genuinely the part people find hardest to accept.
None, in the sense most people mean.
There is no question bank, no multiple choice and no answer sheet circulating that you can memorise. Anyone selling you one is selling you nothing. What exists instead is twelve games, and what gets recorded is the behaviour you produce while playing them.
A rough sense of what that means in practice:
There are no universally correct answers to any of this, because the assessment is not scored against a fixed standard. It is scored against a profile the employer built from people already doing the job well. A pattern that fits a trading desk will not fit a research role, and the same run can pass one and fail the other.
The nearest thing to seeing the questions in advance is playing the games, which is what the rest of this page is for.
Every game below is a working example you can play right now, not a screenshot or a description of one. Each is rebuilt from publicly described mechanics, untimed on most, with the instructions on screen the whole way and unlimited retries. Start anywhere. If you want the real rhythm of the assessment instead, the timed run further down plays all twelve back to back.
Pump a balloon for money and bank it before it bursts. The colour hides the odds, so the only way to earn well is to learn each colour as you go. 39 balloons, the same as the real assessment.
Pumping and banking a balloonDigits appear one at a time, then you type them back in order. Get it right and the next sequence is a digit longer. Three mistakes and the run is over.
Digit sequence flashing then recallFive arrows appear. The colour tells you which one to answer. Blue means the middle arrow, red means the outer ones. The rule changes without warning.
Arrow rows switching colourOne button, two minutes. While it is green, tap it as fast as you can. The moment it turns red, stop completely. It changes without warning.
Go cue and rapid tappingFour decks, $2,000 and sixty cards. Every card pays you and some also charge you. Two of the decks lose money over the run, and nobody tells you which.
Drawing from four decksYou get $10 and your partner gets nothing. Whatever you send them is tripled on the way, and they decide how much to send back. They are under no obligation to send anything.
Sending and receiving a transferYou and your partner each start with $5, and one of you gets an extra $5. Whoever holds it decides what to move. Nothing is multiplied here and nothing comes back, so anything you hand over is simply gone.
Splitting a bonus with a partnerTwo minutes to earn as much as you can. Every trial offers an easy task worth $1 or a hard one worth up to $4.35, and tells you the odds it pays out at all before you pick.
Choosing between two tasksA face flashes for a tenth of a second. Decide whether its mouth was short or long, ninety times over, while some correct answers quietly pay you five cents.
Short mouth versus long mouthA photograph of a face and ten emotion words. Pick the one that fits before the clock on that face runs out. Some faces come with a one-line story, and the story can pull against the expression.
Face with emotion word optionsCircles flash one at a time, about one a second. Press space when the circle is red. Do nothing when it is green. Most of them are red, which is the whole problem.
Red and green circles flashingThree towers, five discs and one target arrangement in view the whole time. Rebuild the target in as few moves as you can, inside two minutes. It can be done in nine.
Discs moving between pegsThe games above are practice: untimed on most, retry as often as you like, instructions on screen the whole way. The full run is the rehearsal. All 12 back to back in the order the real assessment most commonly uses them, timed, one attempt each, and no feedback until the end. About 35 minutes, and nothing is saved, so it needs one sitting.
You do not get a score, and neither does the employer. There is no percentile, no pass mark and no number out of a hundred anywhere in the process.
What the employer receives is a match against a profile they built earlier by having their own existing employees play the same games. Fit is the whole output. This is why two people can play almost identically and get opposite outcomes at two companies, and why a rejection here says less about you than a rejection from an exam would.
Most candidates are told very little. A decision, sometimes a short trait summary, rarely a breakdown. Some employers let a result carry across to other roles at the same company for a period, some do not, and the only reliable way to find out is to ask the recruiter.
Practising cannot move a match toward a profile you cannot see. What it can do is remove the surprise, so that the first time a balloon pops on you is not during the real thing.
Nothing you play on this site is scored, saved, or sent anywhere. These games are for understanding the format, not for predicting your result.
The ones worth answering before you start.
Pymetrics and Harver are trademarks of their respective owners and are not affiliated with OpenThinkr. Practice results are for interest and self-reflection, not assessment.
