Aiming at solving the problem of predicting in real-time the dynamic evolution of a transit system, this paper presents DYBUS2, a modeling framework proposed to overcome some limits of existing models by considering: (i) dynamic real-time simulation, (ii) modeling of travel behavior in the sphere of real-time information, by using a travel strategy approach, (iii) real-time prediction of on-board crowding, the information system should give to travelers, (iv) mesoscopic approach also on the demand side, by grouping travelers in packets. The first part of the paper gives an overview both of general transit network modeling and of DYBUS2 architecture, the latter presents the main modeling components, focusing on the new proposal for the real-time dynamic path choice modeling.