Mistral just launched its first reasoning model, aiming to rival OpenAI and China’s DeepSeek. The French AI firm rolled out this new system Tuesday with a focus on European languages.
CEO Arthur Mensch said the model handles math and coding with strong logic, stepping up over simpler chatbots by reasoning through complex tasks step-by-step.
"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," Arthur Mensch told CNBC at London Tech Week.
The new model joins OpenAI’s o1 and DeepSeek’s R1 in the growing space of reasoning AI, designed to think through problems rather than just spit out text.
Backed by Microsoft, Mistral specializes in open-weight large language models. This means their models’ core data is public, making it easier and cheaper for developers to adapt without building from scratch.
Mensch pointed out why the company stands out:
"Historically, we’ve seen U.S. models reason in English and Chinese models reason in Chinese."
For now, Mistral targets European languages but plans to add more in the future.
DeepSeek shocked the AI world earlier this year by launching R1, a reasoning model that promised to match OpenAI’s performance at a lower cost, heating up global competition.
Mistral’s new model is set to shake up the reasoning AI game by focusing on multi-language capabilities, starting with Europe’s diverse tongues.
