Enough to make an important discovery in math, which gets citations and such. Having read math papers, there is a wide range of ability. Compare the stuff featured in Quanta magazine to your typical pdf in google. 130 is probably the min. to maybe get a master's degree.
The average IQ among mathematicians is somewhere between 130 and 140.
I'm not debating the claim that smart people can learn faster, but like the above I am calling the specifics into question.