The race between open and closed AI is on, and Europe is still in the race. On Tuesday, French AI lab Mistral AI launched Mistral Large 4 (ML4), a new large multimodal model aimed at overtaking its American and Chinese rivals – following what French President Macron described as “a third way in AI”.
Amid a growing divide between unplugged closed models and open models often made in China, Mistral is positioning the ML4 as an alternative to both. Nicknamed The Chonk in reference to its 1,000 billion parameters, ML4 is definitely not small; but it is not yet an open weight model. For now, it’s only accessible through a public guardrail endpoint, but Mistral plans to make its weights available in just three weeks, once safety testing is complete.
“In the meantime, we will work with trusted partners and governments to ensure that open source weights can be used to defend, but not to (perform) malicious attacks,” Pierre Stock, vice president of science at Mistral, told TechCrunch.
Security concerns have increased in recent months, particularly among Mistral’s target audience: businesses and institutions. At the same time, an open-weighted model is easier to audit, Stock said.
Another important aspect behind the scenes is that ML4 has been fully trained on Mistral calculation; using just 4,000 Nvidia GPUs, “which is two to three times less than our Chinese competitors, and significantly less than closed-source competitors,” Stock said.
While benchmark results are still pending, Mistral hopes the ML4 will be best-in-class among open-weight models, especially outside of China, but not only that, Stock said. Through targeted training, it could also outperform closed models in specific areas that are critical to its customers and where multimodal capabilities can add value.
According to Stock, ML4’s optimized use cases include cybersecurity and finance, but also chip design, which is central to two of Mistral’s biggest backers: Dutch giant ASML, which led its Series C, and Samsung, which led its Series D last month to a valuation of 21 billion euros (about $24.39 billion).
At the time, the company tried to make clear that its decision to host Chinese models was not a move toward becoming a mere inference provider. With Le Chonk on its side, the company believes it should also be considered a frontier laboratory.
When you purchase through links in our articles, we may earn a small commission. This does not affect our editorial independence.
Gn bussni
