Evaluating reasoning capabilities of LLMs using the game of Mastermind (paper is coming)