Post-Cutoff

Mistral AIMistral Large familyCurrentMultimodal modelReleased 6 Oct 2026

Mistral Large 4

Mistral’s flagship open-weight MoE, successor to Mistral Large 3 (Apache 2.0). The Oct 29 release moved to a custom licence. Self-reported: Terminal-Bench 4.0 41%, AA Coding Agent Index 48% (about level with GLM-5.3), Cognition FrontierCode 2.0 19.7%.

Model idmistral-large-4-0

Price$0.70 in, $2.10 out per 1M tokens

Checked29 October 2026

No published training cutoff

At least1,143events since its release on 6 October 2026

25 historic, 230 major. 47 days of news.

Mistral AI doesn’t publish a training cutoff, so we count from its release date. The real gap is larger.

Five things it has never heard of

How to call it

7 ways to use Mistral Large 4, with the model id each provider expects.

Providers, model ids, endpoints and docs for Mistral Large 4
ProviderModel idEndpointDocs
Mistral API mistral-large-4-0 https://api.mistral.ai/v1/chat/completions docs.mistral.ai
Hugging Face huggingface.co
Hugging Face (NVFP4) huggingface.co
OpenRouter mistralai/mistral-large-4-0
NVIDIA (build.nvidia.com) mistralai/mistral-large-4-1t-a52b
Microsoft Foundry ai.azure.com
Web app chat.mistral.ai

Pricing

Prices for Mistral Large 4
Input$0.70 per 1M tokens
Cached input$0.07 per 1M tokens
Output$2.10 per 1M tokens

As published: “per 1M tokens (USD), Mistral API, from the Oct 29, 2026 weights release”.

Source: mistral.ai, checked 29 October 2026.

Specs

Input
Text and images
Output
Text
Context window
1,000,000 tokens
Max output
Not published
Open weights
Yes
Licence
Mistral Ecosystem Open-Weight License (modified MIT; model-as-a-service businesses need a commercial licence)
Training cutoff
Not published
Released
6 October 2026
Status
Current
Type
Multimodal model

What stands out

What its maker marketed at launch, and what people found later, each with its source.

  • Trillion-parameter open-weight European model

    Granular MoE, ~1T total / 49B active parameters (52B with embeddings), 1M context, 1.6B vision encoder; weights released Oct 29, 2026 in FP8 and NVFP4.

    Source: huggingface.co

  • Cyber defence

    Mistral reports 51% on the Artificial Analysis Cyber Index and 96.6% refusal of malicious cyber requests; a specialised cyber variant is gated to vetted partners.

    Source: mistral.ai

  • Three reasoning levels

    Low, Medium and High (default); reasoning cannot be switched off in the released checkpoint.

    Source: huggingface.co

Notes

Preview from Oct 6, 2026 (entry Mistral releases Mistral Large 4 (“le Chonk”)); further-trained weights and price cut on Oct 29. Artificial Analysis Intelligence Index v5: 26 (High). Self-hosting: NVFP4 needs Blackwell GPUs; 2-bit GGUF fits a 512 GB Apple Silicon Ultra.