Models / Mistral AI

Mistral AI

Deploy Mistral's model family on Together AI. State-of-the-art performance, Apache 2.0 open weights, and native multilingual support.

Why Mistral AI on Together AI?

Designed for production workloads that need 
consistent performance and operational control.

State-of-the-art at 8× lower cost

Mistral models deliver enterprise-grade performance at a fraction of the price. Mistral Medium 3 benchmarks at state-of-the-art while cutting costs by 8× versus closed-source alternatives.

Multilingual and transparent by design

Native support across English, French, Spanish, German, Italian, and more. Magistral's reasoning chain is fully visible — transparent thinking you can follow and verify across languages.

From frontier to edge, open licensed

Apache 2.0 licensing, on-premises deployment, and fine-tuning on proprietary data. SOC 2 Type II certified and HIPAA compliant on Together AI's US-based infrastructure.

Meet the Mistral AI family

Explore top-performing models across text, image, video, code, and voice.

new

Chat

Ministral 3 8B Instruct 2512

new

Chat

Ministral 3 14B Instruct 2512

new

Chat

Ministral 3 3B Instruct 2512

Chat

Mistral Small 3

New

Transcribe

Voxtral-Mini-3B-2507

Chat

Mistral Instruct

New

Code

Devstral Small 2505

New

Code

Magistral Small 2506

Chat

Mistral

Chat

Mixtral 8x7B Instruct v0.1

Chat

Mixtral 8x7B v0.1

Chat

Mistral (7B) Instruct v0.2

Breakthrough technical innovations

Explore all the game-changing architectural advances that make Mistral AI models shine.

  • Mixture of Experts (MoE)

    Sparse expert routing activates only 37B out of 671B parameters for each token in V3. Advanced load balancing without auxiliary losses maintains performance while reducing computational cost.

  • Group Relative Policy Optimization

    New RL approach that removes separate value networks in RLHF, using grouped relative advantage estimation to cut compute requirements while maintaining training stability.

  • Native Reasoning Transparency

    First reasoning model to expose complete thinking process in <think> tags. Native reasoning capabilities built into model foundation through large-scale reinforcement learning.

  • MetaP Training

    First successful implementation of FP8 mixed precision training on a 671B parameter model. Pioneering reinforcement learning approach without supervised fine-tuning as preliminary step.

  • Multi-Head Latent Attention

    Innovative attention mechanism that reduces KV-cache memory requirements while maintaining modeling performance. Optimized for efficient inference deployment.

  • Multi-Token Prediction

    Novel training objective that allows the model to predict multiple tokens simultaneously. Enhanced performance and efficiency through advanced training techniques.

Deployment options

Run models using different deployment options depending on latency needs, traffic patterns, and infrastructure control.

  • Serverless

  • Inference

Serverless Inference

Real-time

A fully managed inference API that automatically scales with request volume.

Best for

Variable or unpredictable traffic

Rapid prototyping and iteration

Cost-sensitive or early-stage production workloads

Batch

Process massive workloads of up to 30 billion tokens asynchronously, at up to 50% less cost.

Best for

Classifying large datasets

Offline summarization

Synthetic data generation

Dedicated Inference

Dedicated Model Inference

An inference endpoint backed by reserved, isolated compute resources and the Together AI inference engine.

Best for

Predictable or steady traffic

Latency-sensitive applications

High-throughput production workloads

Dedicated Container Inference

Run inference with your own engine and model on fully-managed, scalable infrastructure.

Best for

Generative media models

Non-standard runtimes

Custom inference pipelines