I'm no real expert, but DMR allows for definition of attributes than can be nested under a dimensional member. Think of The dimension customer with attributes address,zip and residence. You can add them within Analysis Studio nested under the customer. AFAIK a cube model does not have this functionality.
Prior to version 8.4 DMR did not behave properly with regards to sorting members. Traditional cubes will perform better, but with small datasets, the difference may not be noticable..
Ties Blom