Meta has released Code Llama, an open-source LLM built on Llama 2, which has been fine-tuned for code-related tasks. It claims to outperform other LLMs in code tasks. Code Llama comes in three variants: a base model, a model specifically tailored for Python, and a model designed for natural language instructions. The models feature enhanced coding capabilities, support multiple programming languages, and offer stable generation. Code Llama is suitable for tasks such as code completion, debugging, and real-time code completion.