Custom Properties - MetrixLLM

Custom properties allow you to tag requests passing through the gateway with arbitrary metadata. This is extremely useful for A/B testing, attributing costs to specific users or customers, and filtering logs.

Sending custom properties

To attach metadata to a request, add HTTP headers prefixed with metrix-property-. For example, if you want to tag a request with a specific user_id and environment:

response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Hello!"}],
    extra_headers={
        "metrix-property-user_id": "usr_12345",
        "metrix-property-env": "production"
    }
)

const response = await client.chat.completions.create({
  model: "gpt-4o",
  messages: [{ role: "user", content: "Hello!" }],
}, {
  headers: {
    "metrix-property-user_id": "usr_12345",
    "metrix-property-env": "production"
  }
});

curl https://gateway.metrixllm.com/openai/v1/chat/completions \
  -H "Authorization: Bearer mtx_..." \
  -H "metrix-property-user_id: usr_12345" \
  -H "metrix-property-env: production" \
  -d '{"model": "gpt-4o", "messages": [{"role": "user", "content": "Hello!"}]}'

To ensure high performance, the gateway strictly limits custom properties to a maximum of 10 properties per request. Additional properties are ignored.

Viewing metadata in the dashboard

All valid properties passed via metrix-property-* headers are extracted and saved alongside the request log in a structured JSON format. In the MetrixLLM dashboard, you can:

View the properties in the request details panel on the Request Logs page.
Filter the logs table by specific property keys and values.
Group spend by custom properties in the Cost Analytics view (e.g. “Show me the total cost for user_id = usr_12345”).