Mistral just unveiled Mistral Large 4, a model its CEO says beats Chinese models on cyber. Here's what you need to know. The story started on Oct 6 in Abu Dhabi. Mistral CEO Arthur Mensch told a conference, per Reuters, that the model being announced today is "above the Chinese models on certain aspects, including cyber." He added, "The narrative that Europe cannot compete is something that is not true." He did not name which Chinese models he meant, and he gave no specific benchmarks. The claim was only that it wins in "certain" areas. A post from AiBattle on X shared the teaser and drew 34K views. Later on Tuesday, Mistral made it official. Reports say Large 4, nicknamed "Le Chonk", is a 1-trillion-parameter, text-output, open-weight model. Mistral calls it the most capable open model outside of China. It opens in API preview now, and the weights are due Oct 27. Early benchmarks reportedly lead in coding and vision. Reaction online has been mixed. Some call a strong European open model a serious statement and say it could shake up the AI market. Others note that open weights were unconfirmed at the time of the teaser. Some point out that the Chinese models it was compared with date from June and July. Many users said agentic coding is the area that matters most. Key numbers: - 1 trillion parameters - API preview live Oct 6, 2026 - Open weights due Oct 27 - Claim: ahead of Chinese models in "certain" areas, including cyber Mistral is positioning Large 4 as proof that Europe can compete at the frontier of open AI.