LLaMA Pro is a model designed for large-scale natural language processing. By leveraging the extended Transformer modules, the model can effectively utilize new data sources to enhance its knowledge without forgetting previous information. LLaMA Pro demonstrates outstanding performance across general tasks, programming, and mathematics. It is a general-purpose model initialized based on LLaMA2-7B. Both LLaMA Pro and its instruction-tuned counterpart (LLaMA Pro-Instruct) achieve state-of-the-art performance on various benchmark tests, showcasing its immense potential for reasoning and handling diverse tasks within intelligent agents. This model provides valuable insights into integrating natural and programming languages, laying a solid foundation for the development of advanced language agents capable of operating effectively in various environments.