A genome-scale community model reveals conserved metabolic cross-feedings in epipelagic bacterioplankton communities. Marine microorganisms form complex communities that influence oceanic ecosystem functions such as primary production and nutrient cycling. Identifying the mechanisms controlling their assembly and activities is a major challenge in microbial ecology. Here, we integrated Tara Oceans meta-omics data to predict genome-scale community interactions within prokaryotic assemblages in the euphotic ocean. A global genome-resolved co-activity network revealed a significant number of inter-lineage associations across diverse phylogenetic distances. Identified co-active communities include species with smaller genomes but encoding a higher potential for quorum sensing, biofilm formation, and secondary metabolism. Community metabolic modelling reveals a higher potential for interaction within co-active communities and points towards conserved metabolic cross-feedings, particularly of specific amino acids and group B vitamins. Our integrated ecological and metabolic modelling approach suggests that genome streamlining and metabolic auxotrophies may act as joint mechanisms shaping bacterioplankton community assembly in the global ocean surface. Marine microbes constantly interact with each other and their environment, forming complex and dynamic networks. These communities and their interactions play crucial ecological and biogeochemical roles on our planet, forming the basis of the marine food web, sustaining biogeochemical cycles in the ocean, and regulating climate. Complex networks of trophic interactions, mediated through metabolic cross-feeding and ecological successions, can influence the nature of microbial interactions (e.g., mutualism or competition), in space and time, and thus significantly shape microbial community assembly. Expanding our understanding of microbial trophic interactions is fundamental given their capacity to modulate ecological niches, constrain microbial biogeography, drive microbial diversification, and modulate the eco-evolutionary dynamics of microbial communities. While species co-occurrence networks are useful tools to model the large-scale structure of microbial communities and to resolve biome-specific ecological associations, these approaches are inherently limited since correlation metrics do not provide evidence for direct biotic interactions, and do not allow to disentangle true biotic interactions from environmental preferences (niche overlap). Thus, we still lack a comprehensive and mechanistic understanding of biotic and abiotic interactions shaping community assembly of microbial communities. Ecosystem modelling approaches are therefore needed to capture and predict emergent properties resulting from complex interactions within microbial communities, such as resilience, niche space, and biogeography, that shape microbial communities and ecosystems. Recent experimental work has demonstrated the significant impact of underlying cross-feeding metabolic networks in shaping community assembly and ecological successions in synthetic microbial communities. Using microbial community assembly experiments in soil, coupled with a simple resource-partitioning model, functional convergence was shown to be mainly driven by emergent metabolic self-organization, while taxonomic divergence seemed to arise from multi-stability in population dynamics. In another system, coculture experiments of a marine microbial community able to degrade chitin demonstrated the hierarchical preferencesA genome-scale community model reveals conserved metabolic cross-feedings in epipelagic bacterioplankton communities. Marine microorganisms form complex communities that influence oceanic ecosystem functions such as primary production and nutrient cycling. Identifying the mechanisms controlling their assembly and activities is a major challenge in microbial ecology. Here, we integrated Tara Oceans meta-omics data to predict genome-scale community interactions within prokaryotic assemblages in the euphotic ocean. A global genome-resolved co-activity network revealed a significant number of inter-lineage associations across diverse phylogenetic distances. Identified co-active communities include species with smaller genomes but encoding a higher potential for quorum sensing, biofilm formation, and secondary metabolism. Community metabolic modelling reveals a higher potential for interaction within co-active communities and points towards conserved metabolic cross-feedings, particularly of specific amino acids and group B vitamins. Our integrated ecological and metabolic modelling approach suggests that genome streamlining and metabolic auxotrophies may act as joint mechanisms shaping bacterioplankton community assembly in the global ocean surface. Marine microbes constantly interact with each other and their environment, forming complex and dynamic networks. These communities and their interactions play crucial ecological and biogeochemical roles on our planet, forming the basis of the marine food web, sustaining biogeochemical cycles in the ocean, and regulating climate. Complex networks of trophic interactions, mediated through metabolic cross-feeding and ecological successions, can influence the nature of microbial interactions (e.g., mutualism or competition), in space and time, and thus significantly shape microbial community assembly. Expanding our understanding of microbial trophic interactions is fundamental given their capacity to modulate ecological niches, constrain microbial biogeography, drive microbial diversification, and modulate the eco-evolutionary dynamics of microbial communities. While species co-occurrence networks are useful tools to model the large-scale structure of microbial communities and to resolve biome-specific ecological associations, these approaches are inherently limited since correlation metrics do not provide evidence for direct biotic interactions, and do not allow to disentangle true biotic interactions from environmental preferences (niche overlap). Thus, we still lack a comprehensive and mechanistic understanding of biotic and abiotic interactions shaping community assembly of microbial communities. Ecosystem modelling approaches are therefore needed to capture and predict emergent properties resulting from complex interactions within microbial communities, such as resilience, niche space, and biogeography, that shape microbial communities and ecosystems. Recent experimental work has demonstrated the significant impact of underlying cross-feeding metabolic networks in shaping community assembly and ecological successions in synthetic microbial communities. Using microbial community assembly experiments in soil, coupled with a simple resource-partitioning model, functional convergence was shown to be mainly driven by emergent metabolic self-organization, while taxonomic divergence seemed to arise from multi-stability in population dynamics. In another system, coculture experiments of a marine microbial community able to degrade chitin demonstrated the hierarchical preferences