This project makes foundation grantmaking data more accessible and understandable to scientists and the public by building an open access database tracking foundation grants to nonprofit organizations in five U.S. metropolitan areas over time. Philanthropic foundations play a vital role in supporting essential programs in areas such as human services, health, education, and the arts, yet the low amount of research on foundation grantmaking is largely due to limited access to reliable, structured grants data. By showing where foundation dollars go, who benefits, and what issues receive support over time, this project enables interdisciplinary research and advances public understanding of philanthropic behavior and its societal impact. The project addresses a notable gap in the scientific understanding of foundation grantmaking through two activities. First, it curates a Longitudinal Foundation Grants Database (LFGD) by extracting and structuring data from foundation tax filings, focusing on grantmaking in five major U.S. metropolitan areas from 2020 to 2023. While the release of Form 990 makes foundation tax data publicly available, its nested XML format and the complexity of funder-nonprofit grants render it difficult to parse and access for research. This project stands out not only for the scale and longitudinal depth of its dataset, but also for its novel use of large language models (LLMs) to classify unstructured grant descriptions into structured variables such as pu