Optimod'Lyon (www.optimodlyon.com) is a project initiated by the Grand Lyon to improve mobility in Lyon's agglomeration. A first goal is to centralize all mobility data and provide a service of prediction of traffic conditions during the day. Another important goal (called Smart Deliveries) is to optimize planned tours of transporters in a global, time-dependent and reactive way. The optimization is global in the sense that all mobility demands are optimized at a whole; it is time-dependent as predicted travel times vary during the day according to predicted traffic conditions; and it is reactive as it dynamically adapts tours to unexpected events like car accidents. In this paper, we compare different Constraint Programming models for addressing the Smart Deliveries routing problem.