The rapid adoption of Large Language Models (LLMs) has generated growing interest in their potential application to Project Management, a domain where decision-making demands precision, operational efficiency, and adaptability. While general-purpose evaluation frameworks such as MMLU and HELM are widely employed, existing literature reveals a notable absence of systematic benchmarks tailored specifically to project management contexts. This thesis addresses this methodological gap by developing a comprehensive set of benchmarks designed to evaluate the reliability, computational efficiency, and practical utility of LLMs in managerial settings. The research is structured around two primary questions: (RQ1) which combinations of datasets, evaluation metrics, and prompting techniques facilitate the construction of meaningful benchmarks for project management tasks; (RQ2) which language model currently achieves the optimal balance among accuracy, computational speed, and economic cost.