I default to the construction of the reals I learned in my first semester Real Analysis course. This is that they are the set of equivalence classes of convergent Cauchy sequences of rationals. I am now wondering where in basic Real Analysis the Axiom of Choice is relied upon. Is that construction of the reals independent of its usage?
If not, can we obtain different models of the real numbers if it is dropped?
(Added) Let me give a specific example of my concern: When I go to prove something like the reals are closed under addition, I say things like: Suppose $x_n$ and $y_n$ are representative cauchy sequences of two reals, then I go on to prove that $x_n + y_n$ is cauchy. Didn't I just use the axiom of choice there? Isn't a choice of representative from each of the equivalence classes (which I made to prove the theorem) a usage?