A hierarchical modeling approach is presented for describing soot growth dynamics, encompassing reactive Molecular Dynamics (MD) simulations and a monodisperse particle dynamics model. Reactive MD is employed to investigate nucleation of soot nanoparticles during isothermal acetylene pyrolysis at 1200–1800 K. A “lumped” soot nucleation rate is determined by tracking the rate of formation of soot clusters at various fuel concentrations, following a power law dependency with the initial acetylene concentration. The MD-obtained soot nucleation rate is incorporated in a monodisperse particle dynamics model describing soot formation in laminar premixed methane flames. The soot volume fraction predicted by the monodisperse model with the MD-derived nucleation rate is in good agreement with measurements in a methane nucleation flame (φ = 1.95), showing significant improvement (3 orders of magnitude) compared to a semi-empirical nucleation rate. The MD-derived nucleation rate also performs well in the methane sooting flame (φ = 2.32), yielding soot volume fractions comparable to experimental measurements.