Tag
This paper introduces GENSTRAT, a benchmark that uses procedurally generated strategic environments to evaluate LLMs' strategic reasoning across multiple axes, addressing limitations of fixed game suites.