Performance management in university-based scientific research institutions is essential for driving reform, advancing education quality, and fostering innovation. However, current performance evaluation models often focus solely on research indicators, neglecting the critical interdependence between the education and research systems. This oversight leads to inefficiencies in resource allocation and an underestimation of overall institutional performance, particularly in universities with varying development levels. To address these deficiencies, this study introduces two innovative two-stage data envelopment analysis models: the group-frontier and meta-frontier evaluation models. The findings are validated using data from 144 universities across China. They demonstrate that the proposed models effectively mitigate the underestimation of efficiency in traditional models and accurately reflect the intertwined nature of university subsystems and the disparities in university development. These results offer valuable insights for improving the performance of scientific research institutions and informing the strategic decisions of university administrators and government education departments.