We know a lot about varying gut microbiome compositions. Yet, how the bacteria affect each other remains elusive. In mammals, this is largely based on the sheer complexity of the microbiome with at least hundreds of different species. Thus, model organisms such as Drosophila melanogaster are commonly used to investigate mechanistic questions as the microbiome consists of only about 10 leading bacterial species. Here, we isolated gut bacteria from laboratory-reared Drosophila, sequenced their respective genomes and used this information to reconstruct genome-scale metabolic models. With these, we simulated growth in mono- and co-culture conditions and different media including a synthetic diet designed to grow Drosophila melanogaster. Our simulations reveal a synergistic growth of some but not all gut microbiome members, which stems on the exchange of distinct metabolites including Tricarboxylic acid cycle intermediates. Culturing experiments confirmed our predictions. Our study thus demonstrates the possibility to predict microbiome-derived growth promoting cross-feeding.