The collaborative network within the open source software community plays a crucial role in motivating developers and ensuring the community’s sustainable development. However, current research on open source collaboration network primarily focuses on statistical indicators and there is a significant lack of standardized benchmarks and objective impact evaluations. This gap hinders a comprehensive understanding of the insights within collaboration network. To address this problem, we first provide a public dataset of open source collaboration network in real-world graph which captures their structural and intrinsic characteristics. In addition, we establish an benchmark through the tasks of link prediction, recommendation systems by employing the graph neural network. Finally we evaluate influence of developers and repositories from influence evaluation. The results demonstrate the effectiveness of our evaluation method and provide potential development trends in open source software community. We anticipate that our proposed benchmark and evaluation will serve as a platform for testing and comparing the potential and performance of future open source collaboration network.