Being comprehensive knowledge bases of cellular metabolism, Genome-scale metabolic models (GEMs) serve as mathematical tools for studying cellular flux states in various organisms. However, analysis of large-scale (human) GEMs, still presents considerable challenges with respect to objective selection and reaction flux constraints. In this study, we introduce a model-based method, ComMet (Comparison of Metabolic states), for comprehensive analysis of large metabolic flux spaces and comparison of various metabolic states. ComMet allows (a) an in-depth characterisation of achievable flux states, (b) comparison of flux spaces from several conditions of interest and (c) identification and visualization of metabolically distinct network modules. As a proof-of-principle, we employed ComMet to extract the biochemical differences in the human adipocyte network (iAdipocytes1809) arising due to unlimited/blocked uptake of branched-chain amino acids. Our study opens avenues for exploring several metabolic conditions of interest in both microbe and human models. ComMet is open-source and is available at https://github.com/macsbio/commet .