A mixture-of-experts LLM from Mistral AI with 8x7B parameters, offering high performance for text generation and reasoning.