In Python, why is "sum ([int (digit) for a digit in str (n)" faster than "sum ([int (digit) for a digit in str (n))"?

Question

It was interesting to me:

> %timeit sum([int(digit) for digit in str(n)]) 100000 loops, best of 3: 1.52 us per loop > %timeit sum(int(digit) for digit in str(n)) 100000 loops, best of 3: 2.04 us per loop