Mistral Large 4
Mistral AI · France · #7 in our ranking · Best European model
Mistral Large 4 is an open-weight mixture-of-experts model from Mistral AI (France), released Sep 2026, with 1T parameters (52B active per token), a 1M-token context window, and the Mistral Ecosystem Open-Weight License.
Mistral Large 4 is a 1T-parameter mixture-of-experts model with about 52B active parameters, trained from scratch by Mistral AI in France. It combines instruction following, reasoning with three effort levels, function calling, and image input in one set of weights, and Mistral highlights cybersecurity, finance, and law as strengths. Its license is a modified MIT license that is free for internal use but requires a separate agreement to offer it as a hosted model service.
Mistral Large 4 specs
- Provider
- Mistral AI
- Origin
- France
- Architecture
- Mixture-of-experts
- Parameters
- 1T total · 52B active
- Context
- 1M tokens | 786K words
- License
- Mistral Ecosystem Open-Weight License
- Input
- Text, Image
- Released
- Sep 2026
- Weights
- Hugging Face
Sourced from the model card on Hugging Face. Last checked 2026-10-29.
Highlights
- 1T parameters, about 52B active
- Text and image input
- Free for internal use under a modified MIT license
Where it fits for payment teams
A good fit for EU teams that want a European provider in vendor and model risk reviews, for work like compliance alert narratives in several languages.
Harnesses in Runtime
Run Mistral Large 4 through any of these harnesses on Runtime.
Provider-reported benchmarks
| AA Coding Agent Index | 48 |
|---|---|
| AA Intelligence Index v5 | 26 |
Benchmark scores are reported by each model provider on its Hugging Face model card. They are not independently verified by Runtime.
How to run Mistral Large 4
Download the weights
Pull mistralai/Mistral-Large-4-1T-A52B from Hugging Face. Check the Mistral Ecosystem Open-Weight License.
Serve it in your cloud
vLLM or SGLang on your GPUs, or a managed host.
Put it to work in Runtime
Point an agent or skill at it, then compare with evals.
Mistral Large 4 FAQ
What is the context window of Mistral Large 4?
+
Mistral Large 4 supports a 1M-token context window (1,048,576 tokens), roughly 786K words of English text.
How many parameters does Mistral Large 4 have?
+
Mistral Large 4 is a mixture-of-experts model with 1T total parameters, of which 52B are active per token.
Is Mistral Large 4 free for commercial use?
+
Mistral Large 4 is released under the Mistral Ecosystem Open-Weight License, a custom license. Commercial use may come with conditions, so read the license on Hugging Face before you deploy or fine-tune it.
What can Mistral Large 4 take as input?
+
Mistral Large 4 accepts text, image input and generates text.
Can I run Mistral Large 4 in my own cloud with Runtime?
+
Yes. Download the weights from Hugging Face (mistralai/Mistral-Large-4-1T-A52B) and serve them with vLLM or SGLang in your cloud, or use a managed provider that hosts the model. Runtime agents can then use Mistral Large 4 through a harness like OpenCode, with your data staying in your environment.