AIbase
EN
Home
AI Models Finder
Mixtral 8x22B Instruct
M

Mixtral 8x22B Instruct

Large-scale mixture of experts model from Mistral AI using 8 experts with 22B parameter architecture. Achieves good balance between parameter efficiency and performance through mixture of experts technology, dynamically activating relevant experts based on task type. Instruction-tuned version excels in instruction following and complex task processing, suitable for high-performance yet efficiency-conscious enterprise applications, especially multi-domain and complex task scenarios.
Intelligence(Relatively Weak)
Speed(Relatively Slow)
Input Supported Modalities
Yes
Is Reasoning Model
65,384
Context Window
-
Maximum Output Tokens
-
Knowledge Cutoff
Go Compare

Pricing

- /M tokens
Input
- /M tokens
Output
￥21.6 /M tokens
Blended Price

Quick Simple Comparison

Input

Output

Devstral
￥0.1
Devstral Medium
￥0.4
Devstral Small (May '25)
￥0.1

Basic Parameters

Mixtral 8x22B InstructTechnical Parameters
Parameter Count
Not Announced
Context Length
65.38k tokens
Training Data Cutoff
Open Source Category
Open Weights (Permissive License)
Multimodal Support
Text Only
Throughput
Release Date
2024-04-17
Response Speed
64.34,568 tokens/s

Benchmark Scores

Below is the performance of Mixtral 8x22B Instruct in various standard benchmark tests. These tests evaluate the model's capabilities in different tasks and domains.
Intelligence Index
26.16
Large Language Model Intelligence Level
Coding Index
16.78
Indicator of AI model performance on coding tasks
Math Index
27.23
Capability indicator in solving mathematical problems, mathematical reasoning, or performing math-related tasks
MMLU Pro
53.7
Massive Multitask Multimodal Understanding - Testing understanding of text, images, audio, and video
GPQA
33.2
Graduate Physics Questions Assessment - Testing advanced physics knowledge with diamond science-level questions
HLE
4.1
The model's comprehensive average score on the Hugging Face Open LLM Leaderboard
LiveCodeBench
14.8
Specific evaluation focused on assessing large language models' ability in real-world code writing and solving programming competition problems
SciCode
18.8
The model's capability in code generation for scientific computing or specific scientific domains
HumanEval
72.1
Score achieved by the AI model on the specific HumanEval benchmark test set
Math 500 Score
54.5
Score on the first 500 larger, more well-known mathematical benchmark tests
AIME Score
-
An indicator measuring an AI model's ability to solve high-difficulty mathematical competition problems (specifically AIME level)
AIbase
Empowering the Future, Your AI Solution Knowledge Base
English简体中文繁體中文にほんご
© 2025AIbase