Skip to main content
7h ago

Mistral Launches 1T-Parameter Model to Bridge Open and Closed AI Divides

The competition between open and closed artificial intelligence models continues, with Europe remaining a significant player. On Tuesday, the French A

2 min read5 views5 tags
Originally reported bytechcrunch

The competition between open and closed artificial intelligence models continues, with Europe remaining a significant player. On Tuesday, the French AI lab Mistral AI unveiled Mistral Large 4, a new large multimodal model designed to surpass both American and Chinese competitors. This launch follows French President Macron's recent description of the technology as representing "a third way in AI."

As a growing divide emerges between models that can be fully detached from their creators and open models that are frequently developed in China, Mistral is positioning its offering as an alternative to both. Nicknamed Le Chonk in reference to its one trillion parameters, the model is quite large, though it is currently not an open-weight release. Instead, it is accessible via a public guardrail endpoint, with Mistral planning to release the model weights in approximately three weeks following the completion of safety testing.

Pierre Stock, Vice President of Science at Mistral, emphasized to TechCrunch that during this interim period, the company will collaborate with trusted partners and governments to ensure the open-source weights are utilized for defense rather than malicious attacks. Security concerns have intensified in recent months, particularly among Mistral's core audience of enterprises and institutions. Stock noted that an open-weight model provides a distinct advantage in this regard because it is easier to audit.

A significant technical detail is that the model, known as ML4, was trained entirely on Mistral's own infrastructure, utilizing only 4,000 NVIDIA GPUs. Stock highlighted that this resource usage is two to three times less than their Chinese competitors and significantly lower than that of closed-source rivals. While benchmark results are currently pending, Mistral aims for ML4 to be the best in class among open-weight models, particularly outside of China, though Stock believes it could also excel in specific areas valued by customers through focused training.

According to Stock, the model's optimized applications include cybersecurity and finance, as well as chip design, a sector that is central to two of Mistral’s major investors. These include the Dutch company ASML, which led its Series C funding round, and Samsung, which led the recent Series D round at a valuation of €21 billion, or approximately $24.39 billion. The company has previously clarified that its decision to host Chinese models does not signify a pivot into becoming a mere inference provider. Stock asserts that with Le Chonk in their arsenal, Mistral maintains its status as a frontier laboratory.

#AI News#Mistral Large 4#1T parameters#NVIDIA GPUs#Chip design
ES
Editorial StaffEditor

The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.

View all posts
Reader feedback

What did you think of this story?

User Comments

Filter:
No comments yet. Be the first to comment!
Continue reading
View all news