French AI startup Mistral to launch AI reasoning model that thinks in multiple languages

French AI startup Mistral on Tuesday announced it’s launching its first reasoning model to compete take rivals like OpenAI and China’s DeepSeek.

Unlike the other reasoning models, Mistral’s new model will bring something fresh to the table that its competitors don’t have: the ability to think through problems in multiple languages.

“We’re announcing in a couple of hours our new reasoning model, which is very much competitive with all the others and has the specificity of being able to reason in multiple languages,” CEO Arthur Mensch told CNBC in an interview during a fireside chat at London Tech Week.

Mistral is launching its First Reasoning Model to Challenge OpenAI and DeepSeek

The launch puts Mistral head-to-head with some serious competition. OpenAI’s o1 model has been available to ChatGPT users since late last year, while DeepSeek’s R1 represents China’s entry into the reasoning model space. But Mistral thinks it has found a gap in the market.

Reasoning models work differently from regular AI chatbots. Instead of just spitting out answers, they work through complex problems step by step, showing their logical thought process. Think of it like watching someone solve a math problem by writing out each step rather than just giving you the final answer.

However, Apple’s latest research paper suggests that AI reasoning capabilities may not live up to the hype. The company’s researchers published a paper called “The Illusion of Thinking” that’s shaking up assumptions about these models. They found that popular AI systems “face a complete accuracy collapse beyond certain complexities,” particularly when dealing with unfamiliar problems.

But Mensch says Mistral’s new model “is great at mathematics [and] great at coding.” But the real differentiator? Language support that goes beyond English and Chinese.

“Historically, we’ve seen U.S. models reason in English and Chinese models reason in Chinese,” Mensch explained. Mistral’s upcoming Magistral reasoning model breaks this pattern by supporting European languages.

Why This Matters

This matters more than you might think. Most AI development has centered around English, with Chinese models serving that market. European languages have often been afterthoughts. For developers and businesses across Europe, having an AI that can reason fluently in French, German, Spanish, or Italian opens up new possibilities.
Mistral has built its reputation on open-weight large language models.

These systems make their underlying parameters publicly available, letting developers peek under the hood and modify the core knowledge. This approach saves developers from the expensive and time-consuming process of training models from scratch.

With backing from Microsoft, Mistral has the resources to challenge industry giants. But the real test will be whether Magistral can match the performance of OpenAI’s o1 while delivering on its multilingual promises.

The AI reasoning model space is heating up fast. Each company is trying to solve the same basic problem: how to make AI systems that can think through complex problems logically rather than just pattern-matching from training data.

Mistral bets that language diversity will be its winning card in this high-stakes game.

Founded in 2023 by Arthur Mensch (CEO), Guillaume Lample, and Timothee Lacroix, Mistral’s meteoric rise reflects its steadfast commitment to pushing the boundaries of AI innovation. As the company steers toward fully operational status in early 2024, the winds of change continue to propel Mistral toward its lofty aspirations in the AI domain.

Mistral Team