A size-composition resolved aerosol model (SCRAM) is coupled to the Polyphemus air quality platform and evaluated over Greater Paris. SCRAM simulates the particle mixing state and solves the aerosol dynamic evolution taking into account the processes of coagulation, condensation/evaporation, and nucleation. Both the size and mass fractions of chemical components of particles are discretized. The performance of SCRAM to model air quality over Greater Paris is evaluated by comparison to PM2.5, PM10, and Aerosol Optical Depth (AOD) measurements. Because air quality models usually assume that particles are internally mixed, the impact of the mixing state on aerosols formation, composition, optical properties, and their ability to be activated as cloud condensation nuclei (CCN) is investigated. The simulation results show that more than half (up to 80% during rush hours) of black carbon particles are barely mixed at the urban site of Paris, while they are more mixed with organic species at a rural site. The comparisons between the internal-mixing simulation and the mixing state-resolved simulation show that the internal-mixing assumption leads to lower nitrate and higher ammonium concentrations in the particulate phase. Moreover, the internal-mixing assumption leads to lower single scattering albedo, and the difference of aerosol optical depth caused by the mixing state assumption can be as high as 72.5%. Furthermore, the internal-mixing assumption leads to lower CCN activation percentage at low supersaturation, but higher CCN activation percentage at high supersaturation. © 2016. American Geophysical Union. All Rights Reserved.