By making this model open source, Mistral joins a small but growing cohort of companies pushing transparency in AI.
French artificial intelligence company Mistral has introduced Europe’s first AI model built specifically for reasoning, positioning itself as a serious contender in the global race dominated by American and Chinese players. The company announced the launch of two new models on Tuesday: the open-source Magistral Small and a more powerful enterprise version called Magistral Medium.
Unlike traditional large language models that rely heavily on vast datasets and computational power, reasoning models employ chain-of-thought techniques, breaking down problems step by step to reach more accurate and contextually aware solutions. Mistral’s move comes as AI developers around the world begin shifting away from brute-force scaling and toward more nuanced, human-like logic in machine learning systems.
“The best human thinking isn’t linear, it weaves through logic, insight, uncertainty, and discovery. Reasoning language models have enabled us to augment and delegate complex thinking and deep understanding to AI,” the company said in a statement.
The launch marks a strategic moment for Mistral, which has attracted high-profile backing, including support from French President Emmanuel Macron. Despite being valued at $6.2 billion, the Paris-based startup has so far lagged behind bigger rivals such as OpenAI, Google, and China’s DeepSeek in terms of market presence and commercial adoption.
Still, a broader shift in the industry could work in its favour. As scaling up existing models runs into performance limitations, the focus on reasoning over raw data volume may give leaner, innovation-focused firms like Mistral a better chance to leapfrog competitors.
The Magistral Small model is now available for download on Hugging Face, and supports reasoning in multiple languages including English, French, Spanish, Arabic, and simplified Chinese. By making this model open source, Mistral joins a small but growing cohort of companies pushing transparency in AI, standing in contrast to the closed model strategies employed by most US firms.
OpenAI and Google introduced reasoning models last year, while Chinese tech players like DeepSeek and Alibaba have gained traction by offering cost-efficient open-source alternatives. Meta, another major AI player, has yet to launch a standalone reasoning model but claims its latest offerings include reasoning capabilities.
