Uh, before everyone starts bragging about his insane IQ, you need to be aware that any IQ test can only measure reliably within a range of +/- 3 standard deviations. And, for reasons I never could find out why, one SD sometimes means +10, and sometimes +15. The curve is a Gauss' Bell.
Then, anything not exactly +/- xSD isn't accurate (if your test contains several hundred questions, it starts working).
The population mean is by definition 100. Let's look at a test with SD=15 (most of the popular ones; the more refined medical ones use 10 frequently).
All we can get here with a short test are groups, in the range of 55 - 145. Usuall Gauss stuff; ~68% are within 85-115 etc.
And, 0.3% are out of that range, with half of them below, other half above.
Now, the technical problem is: It is almost impossoble to measure the IQ of someone more than 2 SDs below 100...so while tests with a range of 3 SD are mathematically fine, they do not make much sense. In other words, any publical test you make that contains a range of more than +/-2SDs (=below 70/above 130) is BS.

So, how do you get those astronomic IQs you can read about? You need to test to a preselected group. If the mean in the test resembles IQ130, you can now confirm IQs from 100 to 160, and so on.
It's a lot harder to confirm IQs below 70, of course.
The link doesn't name the SD, and it gives pseudo-exact values (133 in my case, but I had problems to get the graphics loading anyway, so needed to guess some of the questions).
FWIW, if you wanted to study medicine in Germany, you needed to make a specific test until recently, and this test is nothing but an elaborate IQ test without looking for the ability of expression. But unlike that online test, it also includes memory capacity and fast recognition. Takes a whole day. SD here is 10, and the mean is seen as resembling +1 SD compared to general population (since you only test highschool absolvents). Well, I scored above the scale

.
That's more than 4 SD above average, or above 160 for the most common tests...