It is true. Multiple attempts have been made to measure the IQs of clearly gifted people. They tend to be 2-3 standard deviations out.
IQ is imperfectly correlated with the intellectual abilities needed to be a good mathematician. So top mathematicians are, in the abilities that they need to have, indeed probably several standard deviations over the mean. But in the other abilities (I say abilities loosely here - the luck of how you are feeling that day is surprisingly significant) tested in an IQ test which they don't need for what they do, they are all over the map. Average those two factors together on an IQ test, and they consistently have good IQs, but not stellar ones.
IQ is imperfectly correlated with the intellectual abilities needed to be a good mathematician. So top mathematicians are, in the abilities that they need to have, indeed probably several standard deviations over the mean. But in the other abilities (I say abilities loosely here - the luck of how you are feeling that day is surprisingly significant) tested in an IQ test which they don't need for what they do, they are all over the map. Average those two factors together on an IQ test, and they consistently have good IQs, but not stellar ones.