Prompting and multiple-choice (MCQ) benchmarks to evaluate reasoning capabilities of LLMs using Mastermind.