Mistral
Released October 6, 2026

Mistral Large 4 Preview

Mistral Large 4 Preview: a 1T-parameter MoE model with 49B active params, 1M context, 160+ languages, and a cybersecurity focus. Open weights coming.

Inputs
Text
Image
Outputs
Text

Model Overview

Capabilities, design details, and architectural traits

Mistral Large 4 Preview - Mistral's trillion-parameter "le Chonk"

Mistral Large 4 Preview, unofficially called ML4 and jokingly named le Chonk in Mistral's own announcement, is a public preview of a natively multimodal model built on a granular Mixture-of-Experts architecture, with a stated cybersecurity focus and open weights promised but not yet published.

TraitDetail
ArchitectureGranular Mixture-of-Experts with 1.05T total parameters and only 49B active per forward pass
MultimodalityNatively multimodal with a 1.6B vision encoder; accepts text and image input, outputs text
Context window1M tokens listed in Mistral's documentation
Red-teaming approachTested in real-world settings with cybersecurity leaders, vetted partners and state authorities using a version with reduced moderation and expanded cyber capabilities
Multilingual trainingTrained with significant multilingual data spanning more than 160 languages, including every official EU language
ReasoningListed by Artificial Analysis as a reasoning model

Weights promised, not yet published

The preview API is available on Mistral Studio, but the docs page marks both the weights and the license as Coming soon. Mistral says the weights will follow by the end of the month, along with architecture details and its post-training methodology.

Autonomy as the security pitch

Mistral's case for the model leans on where it runs. Open weights and self-deployment are presented as giving organizations "both the capability and the autonomy to run advanced security work under their own policies." The model also uses the same training, customization and RL environment Mistral offers customers through Mistral Forge.

Benchmark Performance

Independent evaluations · Artificial Analysis

26.1%
Intelligence

Compare Models Side-by-Side

Evaluate specifications, pricing, and independent benchmark indices

Model Details
General Info
ProviderMistralAnthropicGoogle
Release DateOctober 6, 2026November 9, 2026October 28, 2026
Knowledge Cutoff---
Context & Limits
Context Window1M1M
1.0M
Best Context Window
Pricing (per 1M tokens)
Input Pricing
$0.70
Best Input Pricing
$8$2
Output Pricing
$2.10
Best Output Pricing
$40$10
Modalities
Inputs
textimage
textimagefile
text
Outputs
text
text
text
Benchmarks (0-100)
Intelligence Index25.3
64.6
Best Intelligence Index
60.0
Coding Index---
Agentic Index---
Mistral Large 4 Preview
Gemini 4 Carbon

Humanity's Last Exam

Extremely difficult logical reasoning and knowledge.

35%
Mistral Large 4 Preview
Humanity's Last Exam
Score: 35%
Mistral Large 4 Preview
N/A
Claude Fable 5.2
60%
Gemini 4 Carbon
Humanity's Last Exam
Score: 60%
Gemini 4 Carbon

Long Context Reasoning

Logical reasoning over long context windows.

81%
Mistral Large 4 Preview
Long Context Reasoning
Score: 81%
Mistral Large 4 Preview
N/A
Claude Fable 5.2
79%
Gemini 4 Carbon
Long Context Reasoning
Score: 79%
Gemini 4 Carbon

SciCode Benchmark

Scientific coding and mathematical modeling.

54%
Mistral Large 4 Preview
SciCode Benchmark
Score: 54%
Mistral Large 4 Preview
N/A
Claude Fable 5.2
64%
Gemini 4 Carbon
SciCode Benchmark
Score: 64%
Gemini 4 Carbon

Independent evaluation data provided by Artificial Analysis. To view the latest benchmarks and full details, visit their official site.

Explore more from Mistral

Other models by Mistral

Top AI Models

Leading alternatives by intelligence score

View all