Mistral AI

Mistral Small 3

Mistral's efficient 14B dense model. State-of-the-art for its size class. Apache 2.0 license. Great for edge deployment and cost-sensitive applications.

4.2/ 5.0
Free Tier
128K context
Try Mistral 3

Pricing

Free
Limited features
API Pricing
Input$0.1/1M tokens
Output$0.3/1M tokens

Ratings

Overall
4.2
Coding
4.2
Writing
4.0
Reasoning
4.0
Speed
4.9
Value
5.0

Pros

  • Extremely affordable ($0.1/$0.3 per MTok)
  • Fast inference — 14B dense model
  • Apache 2.0 open-weight license
  • Runs on consumer hardware
  • Great for edge/on-device deployment
  • Vision support included

Cons

  • Not competitive with frontier models on complex tasks
  • Limited reasoning depth
  • Smaller context understanding than larger models

Best For

Edge deployment, high-volume low-cost processing, local inference

Quick Info

Provider
Mistral AI
Version
3.0
Released
Dec 2025
Context
128K tokens

Features

Vision/Images
Audio
Video
PDF Processing
Code Generation
AI Agents

Ready to try Mistral 3?

Start with the free tier today.

Get Started