Models
Mistral develops open-weight and commercial large language models. Explore the full lineup, compare benchmarks, and find the right model for your use case.
Latest models
OCR 3
Our OCR service powering our Document AI stack.
Voxtral Mini Transcribe Realtime
An efficient audio input model, fine-tuned and optimized for live transcription purposes.
Devstral 2
Our frontier code agents model for solving software engineering tasks.
Mistral Large 3
A state-of-the-art, open-weight, general-purpose multimodal model.
Explore
Compare specifications and benchmarks, find the right model for your task, and learn how to deploy across cloud providers.
Overview
View all models with specifications, benchmarks, and feature comparison.
Model selection guide
Choose the right model based on task requirements, latency constraints, and cost targets.
Deployment
Deploy models to cloud providers and Mistral Compute.
Best practices
Prompt engineering, sampling parameter tuning, evaluation methods, and fine-tuning workflows.