M

Mistral 7B Instruct

Mistral AI
Vision
32K
Context
$0.0280
Input /1M
$0.0540
Output /1M
16K
Max Output
Run API Call Open in Chat

Mistral 7B Instruct

Source: LangMart Last Updated: 2025-12-23

Description

A high-performing, industry-standard 7.3B parameter model, with optimizations for speed and context length. Mistral 7B Instruct has multiple version variants, and this endpoint is intended to be the latest version (currently v0.3).

Specifications

Specification Value
Context Window 32,000 tokens
Parameters 7.3B
Context Length 32,768 tokens
Max Completion Tokens 16,384 tokens
Quantization bf16 (DeepInfra)
Instruct Type Mistral format

Pricing

DeepInfra Provider

Type Price (per million)
Input Tokens $0.028
Output Tokens $0.054

Supported Parameters

The model supports the following parameters:

  • temperature - Controls randomness (default: 0.3)
  • top_p - Nucleus sampling parameter
  • top_k - Top-k sampling parameter
  • max_tokens - Maximum tokens to generate (max: 16,384)
  • stop - Stop sequences
  • frequency_penalty - Penalize frequent tokens
  • presence_penalty - Penalize already-present tokens
  • repetition_penalty - Penalize repetition
  • seed - Random seed for reproducibility
  • min_p - Minimum probability threshold
  • response_format - Output format specification
  • tools - Function/tool definitions
  • tool_choice - Tool selection mode

Use Cases

The Mistral 7B Instruct model is well-suited for:

  1. General Q&A - Answering questions across various domains
  2. Text Generation - Creating articles, stories, and content
  3. Code Assistance - Helping with programming tasks
  4. Summarization - Condensing long documents
  5. Translation - Converting text between languages
  6. Chat Applications - Conversational AI assistants
  7. Function Calling - Tool-augmented workflows

API Usage Example

LangMart API

curl https://api.langmart.ai/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $LANGMART_API_KEY" \
  -d '{
    "model": "mistralai/mistral-7b-instruct",
    "messages": [
      {
        "role": "user",
        "content": "Explain quantum computing in simple terms."
      }
    ],
    "temperature": 0.3,
    "max_tokens": 1024
  }'

LangMart Gateway

curl -X POST https://api.langmart.ai/v1/chat/completions \
  -H "Authorization: Bearer sk-your-api-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "mistralai/mistral-7b-instruct",
    "messages": [
      {
        "role": "user",
        "content": "Explain quantum computing in simple terms."
      }
    ],
    "temperature": 0.3,
    "max_tokens": 1024
  }'

Other Mistral AI models available on LangMart:

  • mistralai/mistral-tiny - Smaller, faster variant
  • mistralai/mistral-small - Balanced performance
  • mistralai/mistral-medium - Higher capability
  • mistralai/mistral-large - Most capable Mistral model
  • mistralai/mixtral-8x7b - Mixture of Experts architecture
  • mistralai/mixtral-8x22b - Larger MoE model

Model Identification

Property Value
Name Mistral: Mistral 7B Instruct
Model ID mistralai/mistral-7b-instruct
Organization Mistral AI
Created May 27, 2024
Latest Version v0.3
Hugging Face mistralai/Mistral-7B-Instruct-v0.3

Modalities

Type Supported
Input Text
Output Text
Vision No
Audio No

Default Configuration

Parameter Default Value
Temperature 0.3
Stop Sequences [INST], </s>

Features & Capabilities

Feature Supported
Tool/Function Calling Yes
Tool Choice Parameter Yes
Response Formatting Yes
Reasoning Mode No
Text Training Yes
Image Training No

Provider Information

DeepInfra (Primary Provider)

Property Value
API Base https://api.langmart.ai/v1/openai
Chat Completions Supported
Standard Completions Supported
Abortable Requests Enabled
Data Training Not permitted
Prompts Retained No

Data Policy

Through the DeepInfra provider:

  • No training use of prompts
  • Prompts are not retained
  • Data cannot be published
  • Governed by DeepInfra's terms of service

Usage Statistics

Recent usage analytics demonstrate significant adoption:

Date Requests
December 22, 2025 51,790
December 8, 2025 137,801

The model consistently processes billions of tokens daily across varying request volumes.

Version History

Version Notes
v0.3 Latest version (current endpoint default)
v0.2 Previous stable version
v0.1 Initial release

Specific version variants can be accessed via separate model IDs:

  • mistralai/mistral-7b-instruct:v0.1
  • mistralai/mistral-7b-instruct:v0.2
  • mistralai/mistral-7b-instruct:v0.3

Instruction Format

The model uses Mistral's specific instruction format:

[INST] Your instruction here [/INST]

For multi-turn conversations:

[INST] First user message [/INST] First assistant response </s>[INST] Second user message [/INST]

Notes

  • The model weights are openly available on Hugging Face
  • Optimized for both speed and extended context handling
  • Supports tool/function calling for agentic workflows
  • Industry-standard performance for its parameter class