Model comparison
Mixtral 8x22B Instruct vs Codestral
Capability radar, value bars, and side-by-side posture.
Mixtral 8x22B Instruct
Not scored
value score
Codestral
77
value score
Mixtral 8x22B Instruct ctx
—
Codestral ctx
33K
Capability radar
Value · context · multimodal · openness · speed posture
Value head-to-head
- Codestral77
Attribute tape
| Field | Mixtral 8x22B Instruct | Codestral |
|---|---|---|
| Developer | Mistral AI | Mistral AI |
| Context | — | 33K |
| Modalities | Text, Code | |
| Openness | — | Open Weights |
| Speed | — | — |
| Price | — | — |
| Value score | — | 77 |
| Summary | Mixtral 8x22B Instruct is Mistral AI's open-weight instruction-tuned language model using a Sparse Mixture-of-Experts architecture with 141B total parameters and 39B active per token. It is designed for chat, reasoning, coding, and multilingual generation with a 64K context window and native function calling support. | Code-specialized Mistral model. |