The first way is better as it more clearly conveys the intended scope of the variable and prevents errors from accidentally using an object outside of the intended scope.
One reason for wanting to use the second form is if you want to break out of the loop and still have a reference to the object you last reached in the loop.
A bad reason for choosing the second form is performance. It might seem at first glance that the second method uses fewer resources or that you are only creating one object and reusing it. This isn’t the case here. The repeated declaration of a variable inside a loop doesn’t consume any extra resources or clock cycles so you don’t gain any performance benefit from pulling the declaration outside the loop.