Exploring the Boundaries and Capabilities of Large Language Models