Days after Germany’s Aleph Alpha release Kolibri, France’s Mistral has responded.
French AI lab Mistral has released Mistral Large 4, a 1-trillion-parameter model that the company is calling the best open-weights model from the US or Europe. Nicknamed “Le Chonk” by the company, the model is natively multimodal and is available through Mistral’s API starting today, with the open weights to follow at the end of October.
Large 4 is a mixture-of-experts model with 1 trillion total parameters, of which 49 billion are active at any given time. That’s a significant jump in scale from its predecessor, Mistral Large 3, which launched in December 2025 with 675 billion total parameters and 41 billion active.
Mistral Large 4 Claims on benchmarks
Mistral says Large 4 is the strongest open-weights model from the US or Europe on aggregated benchmarks. The framing is deliberate: the open-weights leaderboard has been dominated by Chinese labs such as DeepSeek, Moonshot’s Kimi and Alibaba’s Qwen, and Mistral’s claim carves out the Western half of that field rather than the whole of it.
Beyond the aggregate claim, Mistral says the model is state-of-the-art on “critical workloads”, including cyber defense, manufacturing and finance. It also says Large 4 surpasses closed frontier models on visual grounding, the task of locating and identifying specific objects or regions within an image. As with most launch announcements, these are the company’s own claims, and independent evaluations will show how well they hold up.
The cybersecurity angle appears to be a major focus. Mistral CEO Arthur Mensch told a conference in Abu Dhabi that the company’s newest model outperforms Chinese models on cybersecurity, according to Reuters. Mistral says it is working with cybersecurity partners privately on the model.
“Forged in Europe”
Sovereignty is a central part of the pitch. Mistral says Large 4 was built end-to-end in Europe and can be deployed from Europe on its own Mistral Cloud infrastructure, an appeal to enterprises and governments that want to keep their AI workloads, and their data, within the continent. Mistral has become Europe’s flagship AI lab, and its $2 billion funding round led by chip-equipment giant ASML was widely seen as a sign of how much the region wants a homegrown alternative to US and Chinese AI. Mistral also tops most lists of the leading AI companies in Europe.
Availability
Large 4 is available to all through Mistral’s API today. The open weights are slated to arrive at the end of October, and the company hasn’t yet said what license they’ll ship under. Mistral’s recent releases have been mixed on that front: Large 3 and Small 4 came with Apache 2.0 licenses, while Mistral Medium 3.5 shipped under a modified MIT license.
Running a 1-trillion-parameter model locally will require serious hardware, so for most developers the API is likely to remain the practical way to use Large 4, at least initially. The open release will still matter to enterprises that want to self-host, fine-tune or audit the model.