Qwen2.5-Coder is the latest series of the Qwen large language models, focusing on code generation, code reasoning, and code repair. Built upon the powerful Qwen2.5 model, this series significantly enhances coding capabilities by adding a training token volume of 5.5 trillion, which includes source code, text code bases, and synthetic data. Qwen2.5-Coder-32B has become the current state-of-the-art open-source large language model for code, comparable in coding ability to GPT-4o. Moreover, Qwen2.5-Coder provides a more comprehensive foundation for practical applications like code agents, enhancing coding abilities while maintaining strengths in mathematics and general capabilities.