Skip to main content

Mistral AI

Mistral AI integration allows your product to connect with Mistral's large language model API to perform natural language understanding, text generation, summarization, code assistance, and reasoning using high-performance open-weight and hosted models such as Mistral Small, Mistral Large, and Mixtral. This integration is ideal for applications that need fast, cost-efficient, and privacy-friendly AI models — especially in enterprise or on-prem environments.

Credentials Needed​

To connect with Mistral's API, you need an API Key associated with your Mistral account.

Required credentials:

  • Mistral API Key (Bearer token) — e.g., sk-xxxxxxxxxxxxxxxx

API keys provide full access to your Mistral account quota — handle them securely and never expose them client-side.

Permissions Needed / API Scopes​

Mistral currently uses a single API key authentication model. There are no granular scopes — the key grants access to all endpoints available in your account.

FunctionalityEndpointDescription
Chat / Completion/v1/chat/completionsGenerate text, code, or reasoning output
Embeddings/v1/embeddingsGenerate vector embeddings for semantic tasks
Model Info/v1/modelsList available Mistral models
Streaming/v1/chat/completions (with stream=true)Stream real-time responses

Creating Users / Access Tokens​

Step 1: Generate an API Key​

  1. Go to the official Mistral platform: https://console.mistral.ai
  2. Log in or create an account
  3. Navigate to API Keys in the left sidebar
  4. Click + New Key
  5. Provide a name for your key (e.g., MistralIntegrationKey)
  6. Copy and securely store the API Key — it will be shown only once

Test Connectivity​

You can test the Mistral API connectivity using curl or any OpenAI-compatible SDK (since Mistral follows a similar schema):

curl https://api.mistral.ai/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <MISTRAL_API_KEY>" \
  -d '{
    "model": "mistral-large-latest",
    "messages": [
      {"role": "user", "content": "Hello, Mistral! Please confirm the integration is working."}
    ]
  }'

If the response returns a JSON message with generated content, your API key and connection are correctly configured.

Save the Results in the Platform and Create Connection​

  1. In your platform's integration setup, securely store:
    • MISTRAL_API_KEY
    • (Optional) Default model name (e.g., mistral-large-latest)
  2. Label the connection as Mistral AI Integration
  3. During setup, test the connection by making a small text generation or reasoning request
  4. Save the verified connection for further use across features like chat, analysis, or generation

Best Practices​

  • Store API keys securely in your secret manager or encrypted configuration (never in client code)
  • Rotate API keys regularly and immediately revoke unused ones
  • Use the right model for the task:
    • mistral-small-latest: cost-effective and fast
    • mistral-large-latest: for reasoning and enterprise workloads
    • mixtral-8x7b: mixture-of-experts for balanced performance
  • Implement rate-limiting and retry logic for 429 or transient network errors
  • Use streaming mode for long responses to improve user experience
  • Log minimal metadata for debugging — never log full prompt or response contents
  • Monitor usage and billing in the Mistral Console to manage costs effectively
  • If building a SaaS product, proxy API requests through your backend rather than exposing the key client-side