The efficiency and sustainability of carbon dioxide (CO2) storage in deep geological formations crucially depends on the integrity of the overlying cap-rocks. Existing oil and gas wells, which penetrate the formations, are potential leakage pathways. This problem has been discussed in the literature, and a number of investigations using semi-analytical mathematical approaches have been carried out by other authors to quantify leakage rates. The semi-analytical results are based on a number of simplifying assumptions. Thus, it is of great interest to assess the influence of these assumptions. We use a numerical model to compare the results with those of the semi-analytical model. Then we ease the simplifying restrictions and include more complex thermodynamic processes including sub- and supercritical fluid properties of CO2 and non-isothermal as well as compositional effects. The aim is to set up problem-oriented benchmark examples that allow a comparison of different modeling approaches to the problem of CO2 leakage.