In silico metabolic modeling has enabled systematic study of complicated metabolic processes underlying phenotypes of organisms. Modeling of plant metabolism is often hampered by the network complexity and lack of adequate knowledge. The existing metabolic networks of cassava only cover broad metabolism and are not compartmentalized to truly represent metabolism in photosynthetic tissues. To address the aforementioned limitations and develop a robust metabolic network, physiological and genomic data derived from cassava and leaf models of Arabidopsis and rice were to extend the scope of the existing model. The proposed compartmentalized network of metabolism in photosynthetic tissues of cassava, ph-MeRecon (photosynthetic-Manihot esculenta Metabolic Pathway Reconstruction) was developed based on the information resulting of the comparative study of multiple model plants and cassava genome. The ph-MeRecon covers primary carbon metabolism and comprises 461 metabolites, 550 reactions, and 1,037 metabolic genes. Enzymatic genes on the network were validated using RNA-expression data, and the reactions and pathways were compartmentalized into cytoplasm, chloroplast, mitochondria, and peroxisome. To ensure network connectivity, metabolic gaps were filled using gap reactions obtained from literature and metabolic pathway omnibus. In addition, information on plant physiology, including photosynthetic light-dependent reactions, carboxylase and oxygenase activity of RuBisCO enzyme, and phosphoenolpyruvate carboxylase enzyme activity was incorporated into ph-MeRecon to mimic cellular metabolism in cassava leaves. Thus, ph-MeRecon offers a multi-level platform for system analysis of cellular mechanisms underlying phenotypes of interest in cassava. The ph-MeRecon metabolic model is available at http://bml.sbi.kmutt.ac.th/ph-MeRecon/.