Mixtral 8x7B Instruct
M

Mixtral 8x7B Instruct

Classic mixture of experts model from Mistral AI using 8 experts with 7B parameter architecture. Achieves efficiency optimization while maintaining high performance through mixture of experts technology, activating only partial experts per inference. Instruction-tuned version excels across multiple tasks, becoming important milestone for MoE models. Suitable for applications requiring efficient multi-task processing, laying foundation for subsequent MoE model development.
Intelligence(Weak)
Speed(Relatively Slow)
Input Supported Modalities
Yes
Is Reasoning Model
32,768
Context Window
-
Maximum Output Tokens
-
Knowledge Cutoff

Pricing

- /M tokens
Input
- /M tokens
Output
¥5.04 /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 8x7B InstructTechnical Parameters
Parameter Count
Not Announced
Context Length
32.77k tokens
Training Data Cutoff
Open Source Category
Open Weights (Permissive License)
Multimodal Support
Text Only
Throughput
Release Date
2023-12-11
Response Speed
83.3,547 tokens/s

Benchmark Scores

Below is the performance of Mixtral 8x7B Instruct in various standard benchmark tests. These tests evaluate the model's capabilities in different tasks and domains.
Intelligence Index
16.98
Large Language Model Intelligence Level
Coding Index
4.67
Indicator of AI model performance on coding tasks
Math Index
14.97
Capability indicator in solving mathematical problems, mathematical reasoning, or performing math-related tasks
MMLU Pro
38.7
Massive Multitask Multimodal Understanding - Testing understanding of text, images, audio, and video
GPQA
29.2
Graduate Physics Questions Assessment - Testing advanced physics knowledge with diamond science-level questions
HLE
4.5
The model's comprehensive average score on the Hugging Face Open LLM Leaderboard
LiveCodeBench
6.6
Specific evaluation focused on assessing large language models' ability in real-world code writing and solving programming competition problems
SciCode
2.8
The model's capability in code generation for scientific computing or specific scientific domains
HumanEval
1.2
Score achieved by the AI model on the specific HumanEval benchmark test set
Math 500 Score
29.9
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
© 2025AIbase