Back to all models
Mixtral 8x22B
by Mistral AI
A Mixture-of-Experts model offering excellent efficiency by activating only a subset of parameters per inference.
Technical Specifications
Parameters
176B total (39B active)
Context Window
65K tokens
Category
Open Source
Released
April 2024
Pricing Tier
Free
Capabilities
Text Code Multilingual Function Calling
Pricing Details
Open-weight under Apache 2.0. Self-hosting only.
Pros
- Efficient MoE architecture
- Strong multilingual performance
- Fully open with permissive license
- Good cost-to-performance
Cons
- Smaller context window
- No vision capabilities
- Requires significant compute
- Superseded by Mistral Large 3
Best Use Cases
Multilingual applications
Code generation
Self-hosted enterprise deployment
Research
Quick Info
Need AI Compliance Help?
GovernMy.ai provides expert-led compliance for EU AI Act, Colorado AI Act & ISO 42001.
Learn more