Mistral releases 1-trillion-parameter model to rival closed, open AI labs

French AI lab Mistral AI released Mistral Large 4 (ML4) on Tuesday, a new large multimodal model the company says aims to leapfrog both American and Chinese rivals — following what French President Macron has described as “a third way in AI.”

Amid a growing divide between closed models that can be unplugged and open models increasingly made in China, Mistral is positioning ML4 as an alternative to both. Nicknamed “Le Chonk” for its 1 trillion parameters, the model is not small, but it is not yet open-weight. For now it can only be accessed through a public guardrail endpoint; Mistral plans to make its weights available in three weeks, once safety testing is complete.

“In the meantime, we’ll work with trusted partners and governments to make sure that the open source weights can be used to defend, but not to [perform] malicious attacks,” Mistral VP Science Pierre Stock told TechCrunch.

Security concerns have been mounting in recent months, particularly among Mistral’s core audience of enterprises and institutions. At the same time, Stock said, an open-weight model is easier to audit.

Another notable aspect, according to Stock, is that ML4 was trained entirely on Mistral’s own compute, using 4,000 Nvidia GPUs — “two to three times less than our Chinese competitors, and significantly less than the closed source competitors.”

With benchmark results still pending, Mistral hopes ML4 will rank best-in-class among open-weight models, especially outside China, though not exclusively there, Stock said. Thanks to focused training, the company also believes the model could outperform closed rivals in specific areas important to its customers, including where multimodal capabilities add value. Stock pointed to cybersecurity and finance as optimized use cases, along with chip design — a focus tied to two of Mistral’s main backers, Dutch semiconductor-equipment maker ASML, which led its Series C, and Samsung, which led its Series D last month at a €21 billion valuation (about $24.39 billion).

At the time of that round, Mistral had sought to convey that hosting Chinese models on its platform wasn’t a pivot toward becoming a mere inference provider. With Le Chonk now in its lineup, the company says it believes it should still be considered a frontier lab as well.

Why it matters

Mistral is betting that a model trained on far less compute than its Chinese and closed-source rivals, paired with a staged open-weight release, can keep Europe relevant in a frontier-AI race increasingly split between those two camps.