Looking at the astropy documentation, the Sun's position is calculated in GCRS co-ordinates these are intern expressed in J2000.0 and not JNow.
The difference is therefore due in part to precession between J2000.0 and JNow.
You will also find that the time of the equinox is not exactly 4:30 but has been rounded. The algorithm used to get to 4:30 itself may also have precision limitations due to all sorts of reasons. The same will be true of the calculated position of the Sun at a given time.
These errors will all add up to give a slightly incorrect figure.
Cartes du Ciel (a well known planatarium application) calculates the Sun's location at 4:30 as being:
So it looks like most off the error is due to limits of the Astropy library.