mistralai

Mistral: Mixtral 8x22B Instruct

Mistral: Mixtral 8x22B Instruct is an AI model from MistralAI with a context length of 65,536 tokens and the ability to handle text and file inputs. It supports tools integration but lacks reasoning capabilities or structured output options. At $2.0 per million input tokens and $6.0 per million output tokens, it offers a specific pricing structure; however, currently, there is no benchmark coverage for this model. Given its cost and functionality, Mistral could be suitable for users requiring file processing alongside basic text handling who are willing to pay the higher rate for outputs but do not prioritize reasoning or structured data formats.

Open weights 141B parameters, about 78.8 GiB at Q4_K_M: fits comfortably on 0 of 17 consumer graphics cards in the PicksByCard catalog. See which cards run it →
Quality Score
79/100
price + capability + benchmarks
Input Price
$2.00
per 1M tokens · checked 2026-09-08
Output Price
$6.00
per 1M tokens · checked 2026-09-08
Context Window
65,536
tokens
Model ID
mistralai/mixtral-8x22b-instruct
Vendor
mistralai
Released
April 2024
Tokenizer
Mistral
Input Modalities
text, file
Output Modalities
text
Max Output
52,428 tokens
Tool Calling
✓ supported
Structured Output
✓ supported
Reasoning Mode
not supported
Vision
text only
Audio
no
Moderated
no

What it costs in practice

Computed from the current $2.00/M input and $6.00/M output rates. Run your own numbers →

JobTokensCost
Summarize a 50-page report 30k in / 1.5k out $0.07
Classify 1,000 customer emails 500k in / 50k out $1.30
A month of a busy support chatbot 5M in / 2M out $22.00

Similar models

Quick answers

How much does Mistral: Mixtral 8x22B Instruct cost?
$2.00 per million input tokens and $6.00 per million output tokens.
What is Mistral: Mixtral 8x22B Instruct's context window?
65,536 tokens, roughly 98 pages of text in a single request.
Does Mistral: Mixtral 8x22B Instruct support tool calling?
It supports tool calling, structured output.
Can Mistral: Mixtral 8x22B Instruct process images?
No, it is text-only on the input side.