NAPS, the UBC service we use for Sanger sequencing, charges 2 CAD for purifying a sequencing reaction; especially if submitting several reactions, there is therefore a reasonably big incentive in doing the purification ourselves.
The way that is normally done in the lab is with Sephadex-based size-exclusion columns (I believe Dylan is planning to write a post about that protocol ); they work well, but since you have to pack the columns individually each time, it becomes quite work intensive if you are handling more than a handful of samples. An alternative is to use paramagnetic beads (the ones we use for size selection and cleanups in NGS library preps). After a few attempts, I found a protocol from JGI that works really well (Elkin et al. Biotechniques 2002). The price per sample is less than 2 cents (roughly the same as for Sephadex purification).