Fast pyrolysis of biomass is an important technology in the conversion of lignocellulosic feedstocks to value-added fuels and chemicals. Significant efforts have been dedicated to modeling of these processes to improve the viability of large-scale operation through reactor design, feedstock selection and processing, and optimization of operating conditions, among others. This work is a review of the current progress in the field of modeling of biomass fast pyrolysis processes across multiple length and time scales. Enclosed are summaries of the current state of the art in atomistic and kinetic modeling of biomass fast pyrolysis toward production of fuels and chemicals. Decomposition of aggregate biomass and its individual components was reviewed for models at various scales, highlighting important considerations. Recent applications of machine learning methods to couple multiscale phenomena with the goal of reducing computational complexity were also included. Historical context was provided for existing models and correlations, highlighting some of those most widely applied. Some of the shortcomings and bottlenecks in existing models were identified as areas for further study. Finally, potential future directions for the field are suggested with the goal of improving the viability and sustainability of pyrolysis processes and the applications of multiscale modeling toward this goal.