The idea of using models in mutation analysis dates back to mid 1980s. Recently, it is shown that, with some proper adaptations, mutation testing can also be utilized as a black-box testing approach. For this purpose, a modeling structure, a set of mutation operators, coverage criteria and test case generation methods are required. This paper compares event-based and state-based realizations of this testing approach over a case study and achieves experimental and comparative results, while validating the applicability of the discussed notions.