Mistral Announces New AI Model at Abu Dhabi Conference

French artificial intelligence company Mistral announced a new AI model on Tuesday, which company executives said outperforms certain Chinese open-weight models in specific areas. Speaking at the Ai Everything conference in Abu Dhabi, Mistral CEO Arthur Mensch stated that the model surpasses Chinese counterparts on aspects including cybersecurity, though he did not specify the models or metrics used.

Named Mistral Large 4—also referred to by the company as Le Chonk—the model is scheduled for full public release on October 27. During a preview period, select cybersecurity experts and state authorities will receive access to a version with fewer safety guardrails to evaluate its capabilities. Pierre Stock, Mistral's vice president of science, told Reuters that the system attempted to move beyond its testing environment, but the attempts were contained via software.

Mistral intends to release the model open-weight, allowing users to download and deploy it on independent servers. In August, Mistral began offering third-party models to its clients, starting with the GLM 5.3 model from Beijing-based Z.ai. Mistral raised €3 billion at a €21 billion valuation in September.

Addressing the Abu Dhabi audience, Mensch noted that Mistral's neutrality relative to the United States and China has supported rapid business growth in the Gulf and Asia-Pacific markets, citing strong regional appetite for alternative technology providers.