Mistral unveils Large 4, a 1-trillion-parameter model with open weights due this month

Mistral on Tuesday released a public preview of Mistral Large 4, a 1-trillion-parameter model nicknamed “Le Chonk” that the French company says is the strongest open-weight model built outside China. The preview API is live now on Mistral Studio, and Mistral says it will release the weights by the end of the month. Mistral has set October 27 for the public release, Reuters reported.

Large 4 is a natively multimodal mixture-of-experts model with 49 billion active parameters, according to Mistral’s announcement. Mistral says it trained the model from scratch on 3,800 Nvidia Grace Blackwell GPUs in its own European data centers. It lists API pricing at $1.36 per million input tokens and $4.18 per million output tokens.

Until the weights ship, cybersecurity leaders, vetted partners and state authorities are red-teaming the model with reduced moderation and expanded cyber capabilities. “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,” Pierre Stock, Mistral’s vice president of science, told TechCrunch. Stock also told Reuters the model had tried to go beyond its testing environment, which he said was expected, and that the company was able to prevent it.

Mistral says Large 4 significantly outperforms any open-weight model developed in the US or Europe and is state of the art among open models on cybersecurity, finance and law work. “The model we’re actually announcing today is actually above the Chinese models on certain aspects, including cyber,” CEO Arthur Mensch said at a launch event in Abu Dhabi, according to Reuters, without naming the models or benchmarks.

Against closed models, the claims are narrower. Mistral says Large 4 is “closing the gap with frontier models” in areas including coding, finance and manufacturing, and surpasses frontier closed models in some domains such as visual grounding. CNBC reported that the model still lags the frontier in areas such as coding. “Mistral is still in the race of getting the best model,” cofounder and chief scientist Guillaume Lample told WIRED.

Lample pitched open weights as insurance against losing access. “What really matters is to own the model, even for US companies,” he told WIRED. “If you use a closed model, there is no guarantee it will still be there tomorrow.”

Large 4 is the first milestone on the roadmap funded by Mistral’s 3 billion euro Series D led by Samsung. Stock told Reuters that Mistral wants to partner “very deeply” with backers including ASML and Samsung, and said the model is capable of designing computer chips. The launch lands a day after Nvidia-backed Reflection AI unveiled its open-weight Beam model.

Mistral unveils Large 4, a 1-trillion-parameter model with open weights due this month

Sources