Content on Rails
Join Beta
M

Mistral Large 2

AI Assistantsv2407

The new generation of our flagship model.

By Mistral AI•Updated 2025-12-16•Visit Website ↗
Try Mistral Large 2

Overview

123B parameters, offering state-of-the-art performance.

128k token context window for long-context applications.

Strong multilingual support, excelling in over a dozen languages.

Advanced reasoning and mathematical capabilities.

Native function calling and JSON outputting for agentic capabilities.

Visual Guide

✨ Nano Banana Slides
Open Fullscreen ↗

Interactive presentation generated by Manus Nano Banana

Key Features

globe

Supports dozens of languages including English, French, German, Spanish, Italian, Chinese, Japanese, Korean, Portuguese, Dutch and Polish.

code

Trained on over 80 coding languages such as Python, Java, C, C++, JavaScript, and Bash, as well as more specific languages like Swift and Fortran.

robot

Best-in-class agentic capabilities with native function calling and JSON outputting.

brain

State-of-the-art mathematical and reasoning capabilities, with high scores on benchmarks like GSM8K and MATH.

file-text

A large 128k context window, allowing for the processing of extensive documents and complex conversations.

Real-World Use Cases

Complex Code Generation

For

A developer needs to generate a complex algorithm in Python for a data analysis task.

Example Prompt / Workflow

Multilingual Customer Support

For

A global company wants to build a chatbot to handle customer support queries in multiple languages.

Example Prompt / Workflow

Financial Data Analysis

For

A financial analyst needs to analyze a large dataset of financial reports to identify trends and anomalies.

Example Prompt / Workflow

Frequently Asked Questions

Pricing

Model: Pay-as-you-go

Pros & Cons

Pros

  • ✓ State-of-the-art performance in reasoning, code, and math.
  • ✓ Excellent multilingual capabilities.
  • ✓ Large 128k context window.
  • ✓ Open weights under a research license.
  • ✓ Strong instruction-following and conversational abilities.

Cons

  • ✕ Commercial use requires a separate license.
  • ✕ Requires significant computational resources for self-deployment.
  • ✕ No built-in moderation mechanisms.

Quick Start

1

Step 1

Visit the Mistral AI website to access the model via la Plateforme.

2

Step 2

For self-deployment, download the model weights from Hugging Face.

3

Step 3

Follow the provided documentation to set up the model with `mistral_inference` or `transformers`.

4

Step 4

Start building applications by leveraging the model's capabilities through the API.

Alternatives