On the road towards high-fidelity quantum computers, it is desirable to obtain a model of the physical system that is detailed enough to understand the limits of gate fidelity. Such a process is arduous and impractical with increasing chip size. Using optimal control, we show a systematic way to evaluate and analyze a model prediction of experimental data and use it to construct a matching goal function. As an application example, we present the characterization of a two-qubit QPU in simulation. Our goal function expresses the likelihood that the noisy calibration data is predicted by a candidate model. We then optimize the parameters of the candidate, like qubit frequency and couplings, to improve the match and enrich the model to include open-system effects, control line distortions and SPAM errors.  Wittler, N., et.al. An integrated tool-set for Control, Calibration and Characterization arXiv:2009.09866 *This work was supported by the European Commission through the Marie Curie ETN QuSCo (Grant Nr. 765267) and the OpenSuperQ project (Grant Nr. 820363), by the Intelligence Advanced Research Projects Activity (IARPA) through the LogiQ (Grant Nr. W911NF-16-1-0114) and by the Germany Ministry of Science and Education (BMBF) through project VERTICONS (Grant Nr. 13N14872).