As scientific questions become increasingly complex and research collaborations more frequent, multi-principal investigators (M-PIs) model have emerged. However, Chinese science funding agencies are still in the exploratory phase of implementing M-PIs model, and the decision-making process lacks empirical research support. This paper constructs an econometric model based on projects funded by the U.S. National Institutes of Health from 2006 to 2022, combined with publication and patent performance metrics, to assess the output performance differences between M-PIs grants and single-PI grants. Moreover, this paper explores the heterogeneous effects of the M-PIs model across projects with varying levels of funding. The study demonstrates that M-PIs grants surpass single-PI grants in terms of the number of publications, patents, citations, and innovation. Furthermore, there is a complex non-linear relationship between the number of PIs and output performance, with the positive impact of the M-PIs model becoming more pronounced in higher-funded grants. This research provides important empirical evidence for the National Natural Science Foundation of China to promote and implement a M-PIs policy in scientific grants.