Code Llama is a new, cutting-edge language model designed to transform people's experience with coding. This artificial intelligence tool has the ability to generate and discuss code using text prompts, potentially speeding up the process for developers and making it easier for coding novices to get started. Developers can leverage Code Llama's capabilities to write more sturdy, better-documented software, enhancing productivity and learning opportunities. Moreover, recognizing the value of an open AI approach, Code Llama is being made accessible for both research and commercial usage. This unique tool is developed as a refined version of Llama 2 and is optimized for coding tasks. Code Llama can perform a plethora of tasks, ranging from generating code from prompts, completing code, to debugging. It is competent in supporting popular coding languages such as Python, C++, Java, PHP, Typescript (Javascript), C#, Bash, and more. Code Llama is available in three sizes each catering to different operation needs, and is trained with extensive tokens of code and related data. Additionally, specialized models 'Code Llama – Python' and 'Code Llama – Instruct' are provided with unique fine tuning, demonstrating enhanced utility and output alignment.