Mistral 7B Instruct is a fine-tuned version of the Mistral 7B base model, specifically optimized for conversation and instruction-following. It was released by Mistral AI as a high-performance open-weight model designed to balance efficiency with competitive performance against larger language models. The model utilizes a transformer architecture incorporating Grouped-query attention (GQA) for faster inference and Sliding Window Attention (SWA) to handle sequences effectively. These architectural choices allow it to maintain a smaller memory footprint while achieving higher throughput. It is capable of various natural language processing tasks including summarization, classification, and code generation. The instruct version is trained using supervised fine-tuning to better align with user prompts.