Agent-based vegetation models are a widely used tool in ecology, for example, to understand and predict the response of vegetation to environmental change. Models are based on well-established descriptions of processes such as vegetation establishment, growth and mortality. However, they are often developed from scratch, which can be inefficient. Here we present pyMANGA, a free and open-source platform for plant growth modelers. pyMANGA's modular design allows for the combination of different concepts and theories of how plants establish, grow or compete in response to above- and below-ground resource availability. New or alternative modules describing, e.g., competition or facilitation, can be easily added. The interchangeability of modules supports the systematic testing of different hypotheses, e.g., on dominant processes in soil-plant feedback loops. Here we further present the thorough benchmarking strategy to maintain the platform and how pyMANGA can be used to compare models with different levels of abstraction and complexity.