LLM Hosting Pro
← All model families

Mistral

Efficient models from Paris-based Mistral AI, many released under Apache 2.0.

CodingMultilingualSmall and fast

Overview

Mistral AI builds compact, efficient models that perform well for their size, including mixture-of-experts models such as Mixtral.

Many of its open-weight models use the Apache 2.0 licence, but some use Mistral's own research or commercial licences. Check each model's licence before deploying.

Models in this family

ModelReleasedParametersContext
Mistral Small 3.117 Mar 202524B128K tokens
Mistral NeMo18 Jul 202412B128K tokens
Codestral 22B29 May 202422B32K tokens

Where to host Mistral models

All hosts →
LLM hostModelsHosting plans
Amazon AWSMistral Small 3.1, Mistral NeMo, Codestral 22BAmazon SageMaker, Amazon EC2
Google CloudMistral Small 3.1, Mistral NeMo, Codestral 22BModel Garden, Google Kubernetes Engine (GKE), Compute Engine
Microsoft AzureMistral Small 3.1, Mistral NeMo, Codestral 22BAzure Machine Learning, Azure Kubernetes Service (AKS), Azure GPU Virtual Machines