Microsoft-backed AI lab Mistral is launching its first reasoning model in challenge to OpenAI

Mistral, a French AI company, is set to introduce its first reasoning model today, marking a significant move in the competitive landscape against established players like OpenAI and China's DeepSeek. During a discussion at London Tech Week, CEO Arthur Mensch revealed to CNBC's Arjun Kharpal that this new model is designed to handle complex tasks through a logical step-by-step reasoning process. "We're announcing our new reasoning model, which is highly competitive and unique in its ability to reason across multiple languages," Mensch stated. The reasoning model excels in mathematics and coding, highlighting its versatility in various applications. Currently, notable reasoning models include OpenAI's o1, which became available to ChatGPT users late last year, and DeepSeek's recently launched R1 model. Backed by Microsoft, Mistral focuses on open-weight large language models, allowing developers to access and adjust the model's core parameters. This approach significantly reduces the costs and time often associated with developing AI systems from the ground up. Mensch emphasized that the distinguishing feature of the new Magistral model is its capability to reason in European languages. He noted, "Historically, U.S. models have primarily focused on English, while Chinese models are tailored for Chinese language processing." Although the initial release will focus on European languages, Mistral plans to expand language support in the future. Earlier this year, DeepSeek's R1 model entered the market, drawing attention for its competitive performance and affordability compared to OpenAI's offerings, adding further excitement to the evolving AI sector.