M

Mistral: Mixtral 8x7B Instruct

Mistral AI
Vision
32K
Context
$0.5400
Input /1M
$0.5400
Output /1M
16K
Max Output
Run API Call Open in Chat

Mistral: Mixtral 8x7B Instruct

Description

Mixtral 8x7B Instruct is a pretrained generative Sparse Mixture of Experts model with 8 experts totaling 47 billion parameters. It has been fine-tuned by Mistral AI specifically for chat and instruction-following applications.

The model uses a Mixture of Experts architecture, which means that while it has 47B total parameters across all experts, only a subset of parameters are activated for each token, making it more computationally efficient than a dense model of equivalent size.

Technical Specifications

Context & Token Limits

Parameter Value
Context Window 32,000 tokens
Context Length 32,768 tokens
Max Completion Tokens 16,384 tokens
Total Parameter Count 47 billion (8 experts)

Supported Parameters

Parameter Description
max_tokens Maximum number of tokens to generate
temperature Controls randomness (0.0 - 2.0)
top_p Nucleus sampling threshold
top_k Top-k sampling parameter
stop Stop sequences
frequency_penalty Penalize frequent tokens
presence_penalty Penalize tokens based on presence
repetition_penalty Penalize repeated tokens
seed Random seed for reproducibility
min_p Minimum probability threshold
response_format Output format specification
tools Tool/function definitions
tool_choice Tool selection strategy

Modalities

Direction Supported Types
Input Text only
Output Text only

Pricing

Type Price
Input Tokens $0.54 per 1M tokens
Output Tokens $0.54 per 1M tokens

Cost Examples

Use Case Input Tokens Output Tokens Total Cost
Short query 100 500 $0.000324
Medium conversation 1,000 2,000 $0.00162
Long document processing 10,000 5,000 $0.0081
Full context usage 32,768 16,384 $0.0265

Capabilities

  • Instruction Following: Fine-tuned for following complex instructions
  • Conversational AI: Optimized for multi-turn chat interactions
  • Tool/Function Calling: Supports tool definitions and function calling via tools and tool_choice parameters
  • Multilingual: Supports multiple languages (French, German, Spanish, Italian, English)
  • Code Generation: Capable of generating and understanding code

Limitations

  • Does not support reasoning mode
  • Text-only (no image or audio processing)
  • No native vision capabilities

Use Cases

  1. Customer Support: Automated response generation and query handling
  2. Content Generation: Blog posts, articles, creative writing
  3. Code Assistance: Code generation, debugging, and explanation
  4. Translation: Multi-language translation tasks
  5. Summarization: Document and text summarization
  6. Data Extraction: Structured data extraction from unstructured text

From Mistral AI

Model Description
mistralai/mistral-7b-instruct Smaller 7B parameter instruct model
mistralai/mistral-medium Medium-sized Mistral model
mistralai/mistral-large Larger Mistral model
mistralai/mixtral-8x22b Larger MoE model with 8x22B architecture

Similar MoE Models

Model Parameters Provider
Mixtral 8x22B 141B Mistral AI
Grok-1 314B xAI

Model Identity

Property Value
Model Name Mistral: Mixtral 8x7B Instruct
Model ID mistralai/mixtral-8x7b-instruct
Creator Mistral AI
Release Date December 10, 2023
Architecture Sparse Mixture of Experts (MoE)

Provider Information

DeepInfra

Property Value
Provider DeepInfra
Quantization FP8
Status Page https://status.deepinfra.com/

API Usage

LangMart Endpoint

curl https://api.langmart.ai/v1/chat/completions \
  -H "Authorization: Bearer $LANGMART_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "mistralai/mixtral-8x7b-instruct",
    "messages": [
      {
        "role": "user",
        "content": "Hello, how are you?"
      }
    ]
  }'

With Tool Calling

curl https://api.langmart.ai/v1/chat/completions \
  -H "Authorization: Bearer $LANGMART_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "mistralai/mixtral-8x7b-instruct",
    "messages": [
      {
        "role": "user",
        "content": "What is the weather in Paris?"
      }
    ],
    "tools": [
      {
        "type": "function",
        "function": {
          "name": "get_weather",
          "description": "Get current weather for a location",
          "parameters": {
            "type": "object",
            "properties": {
              "location": {
                "type": "string",
                "description": "City name"
              }
            },
            "required": ["location"]
          }
        }
      }
    ],
    "tool_choice": "auto"
  }'

Access Points

Resource URL
LangMart Chat https://langmart.ai/chat
Model Comparison https://langmart.ai/model-docs
Hugging Face Weights https://huggingface.co/mistralai/Mixtral-8x7B-Instruct-v0.1

Benchmarks & Performance

The Mixtral 8x7B model has demonstrated strong performance across various benchmarks:

  • Competitive with or exceeding GPT-3.5 on many tasks
  • Strong multilingual capabilities
  • Efficient inference due to MoE architecture (only ~13B parameters active per forward pass)
  • Good balance of quality and cost-effectiveness

Data sourced from LangMart.ai on December 23, 2025