Mistral Large 4

Europe's largest open-weight model: 1T-parameter multimodal mixture-of-experts with 1M context and a modified MIT license.

In short. Mistral Large 4 is the biggest open model from the French company Mistral, trained from scratch in Europe. It reads text and images, speaks more than 160 languages and is tuned for coding and cybersecurity. At about a trillion parameters it needs a data-center server, and its license is generous except for companies that resell hosted access.

Summary

Mistral AI previewed Mistral Large 4, nicknamed Le Chonk, on its API on 6 October 2026 and published the weights on 29 October 2026 after three further weeks of reinforcement learning. It was trained from scratch on 3,800 NVIDIA Grace Blackwell GPUs in Mistral's own European data centers.

It is a natively multimodal mixture-of-experts model with about 1 trillion parameters in total and 49 billion active per token (52 billion counting embeddings and the output layer). There are 128 routed experts with 4 active per token, plus 1 shared expert. It accepts text and images, writes text, and has a 1-million-token context window. Mistral says it speaks over 160 languages, including every official language of the European Union. A Pixtral-style vision encoder handles images of different sizes, and three reasoning-effort settings (low, medium and high) trade speed for depth. The checkpoint includes a built-in draft head for faster speculative decoding.

Mistral emphasizes coding and defensive cybersecurity. Its own report lists 19.7 percent on Cognition's private FrontierCode 2.0 and 41 percent on Terminal-Bench 4.0, which are vendor-reported. The official checkpoint is about 1.06 TB; an NVFP4 version of about 610 GB targets NVIDIA Blackwell GPUs with vLLM and SGLang.

The license is the Mistral Ecosystem Open-Weight License, a modified MIT license. Anyone may use and sell products built on the model, but a business that offers hosted model access to third parties needs a separate commercial license from Mistral, with no revenue threshold.

Good for

  • Sovereign or on-premises deployments, especially in Europe, where data must stay under your control
  • Coding agents and defensive security work
  • Multilingual products that need strong European-language coverage
  • Analyzing documents and images with a 1M-token context

Watch out for

  • About 1 TB of weights means a multi-GPU node; the NVFP4 build needs Blackwell-class GPUs
  • Hosting the model as a paid service for others requires a separate commercial license, however small you are
  • Mistral's benchmark claims and cybersecurity comparisons are self-reported
  • It writes text only

Performance claims come from the developer and have not been independently verified.

How much memory does it need?

Estimates for the Large 4 NVFP4 variant. A mixture-of-experts model must hold all of its experts in memory even though only some are used for each token, so total parameters decide the memory and active parameters decide the speed.

Variants Parameters BF16 FP8 / INT8 4-bit (Q4) Fits on
Large 4 1.05T 52B active 2.42 TB 1.21 TB 725 GB GPU server
Large 4 NVFP4 NVFP4 1.05T 52B active 2.42 TB 1.21 TB 671 GB GPU server

Estimates: weights plus about 15% overhead. Real usage grows with context length and depends on the software.

VRAM calculator

The license in plain words

A modified MIT license. You can use and build on the model freely, but any business that sells hosted access to it needs a separate commercial license from Mistral, with no revenue threshold.

  • Separate deal needed to sell hosted access above a threshold

More about Mistral Open-Weight

See also

Frequently asked questions

What is Mistral Large 4?
Mistral Large 4 is the biggest open model from the French company Mistral, trained from scratch in Europe. It reads text and images, speaks more than 160 languages and is tuned for coding and cybersecurity. At about a trillion parameters it needs a data-center server, and its license is generous except for companies that resell hosted access.
Can I use Mistral Large 4 commercially?
Yes, with conditions. Mistral Large 4 is released under the Mistral Open-Weight, which allows commercial use but adds extra rules: Separate deal needed to sell hosted access above a threshold. Read the license text before you ship a product.
Is Mistral Large 4 open source?
Mistral Large 4 is open-weight, not open source in the strict sense. You can download the model files, but the Mistral Open-Weight adds conditions that the Open Source Initiative definition does not allow.
Can I run Mistral Large 4 on my own computer?
Not on a desktop. The smallest variant, Large 4 NVFP4, needs about 671 GB at 4-bit precision, which means a GPU server with several data-center cards.
When was Mistral Large 4 released?
Mistral Large 4 was released by Mistral AI on October 29, 2026.
How long a context window does Mistral Large 4 have?
Mistral Large 4 supports a context window of up to 1M tokens.

Official links

Sources

Specifications come from these primary sources. Benchmark claims, where mentioned, are reported by the developer.

Release date and facts last verified November 8, 2026.

Related models

Model Released Parameters Context License Smallest variant needs
Muse Spark Open Meta 816B 42B active 1M Muse Model License (Conditional) 563 GB GPU server
Inkling Thinking Machines Lab 276B – 975B 41B active 1M Apache 2.0 (Permissive) 188 GB Big-memory desktop
Command A+ Cohere 218B 25B active 128K Apache 2.0 (Permissive) 145 GB Big-memory desktop
Voxtral TTS Mistral AI 4.09B – CC BY-NC (Non-commercial) 9.2 GB Laptop or phone