The goal of this project is to establish the relative reasoning capabilities of different large language models through a unique and hopefully unbiased benchmark. Chess puzzles are a very challenging ...