Free Mistral AI Releases First Generative AI Model

Paris, France - Mistral, a French AI startup that secured a significant seed funding round in June, has unveiled its inaugural model, asserting its superior performance compared to others of similar scale. Notably, this model, named Mistral 7B, is made accessible at no cost and without any usage limitations.

Mistral 7B represents a further enhancement of existing "small" large language models like Llama 2, delivering comparable capabilities based on certain standardized benchmarks while significantly reducing computational expenses. While foundational models like GPT-4 offer broader capabilities, they come with higher costs and operational complexities, often being provided exclusively through APIs or remote access.

In a blog post accompanying the model's launch, Mistral's team articulated their aspiration to lead and support the open generative AI community, aiming to elevate open models to state-of-the-art performance. They emphasized Mistral 7B's achievements as a testament to the potential of small models when developed with unwavering dedication. The release follows three months of intense work, involving the formation of the Mistral AI team, reconstruction of a high-performance MLops stack, and the meticulous design of a highly advanced data processing pipeline from the ground up.