Experimental investigations of tumors often result in data reflecting very complex underlying mechanisms. Computer models of such phenomena enable their analysis and may lead to novel and more efficient therapy strategies. We present a generalized finite element mechano-chemical model of a solid tumor and assess its suitability for predicting therapy outcome. The model includes hosting tissue, tumor cells (vital and necrotic), nutrient, blood vessels and a growth inhibitor. At a certain time instant of the tumor development virtual therapies are performed and their outcomes are presented. The model is initiated with parameters either obtained directly from the available literature or estimated using multi-scale modeling. First results indicate the usefulness of multi-physics tumor models for predicting therapy response.