Meta-Prompting is an effective prompting technique aimed at augmenting the capabilities of language models (LMs). This method transforms a single LM into a multifaceted commander, adept at managing and synthesizing multiple independent LM queries. Through the use of high-level instructions, Meta-Prompting guides the LM to decompose complex tasks into smaller, more manageable subtasks. These subtasks are then handled by different 'expert' instances of the same LM, each operating under specific customized instructions. At the heart of this process lies the LM itself, acting as the commander, ensuring seamless communication and effective integration between the outputs of these expert models. It also leverages its inherent critical thinking and robust validation processes to refine and verify the final results. This collaborative prompting approach enables a single LM to function as both a comprehensive commander and a diverse team of experts, resulting in a substantial performance boost across a wide range of tasks. Meta-Prompting's zero-shot and task-agnostic nature greatly simplifies user interaction, eliminating the need for detailed task-specific instructions. Furthermore, our research indicates that external tools, such as Python interpreters, can seamlessly integrate with the Meta-Prompting framework, expanding its applicability and utility. Through rigorous experimentation with GPT-4, we demonstrated that Meta-Prompting outperforms traditional prompting methods: averaging across all tasks, including 24-point game, One-Step Checkmate, and Python programming challenges, Meta-Prompting with Python interpreter functionality achieved a 17.1% improvement over standard prompting, a 17.3% improvement over expert (dynamic) prompting, and a 15.2% improvement over few-shot prompting.