Call Name

Tag requests with a label to filter and analyze them in the dashboard

call_name is a label you attach to a request through the metadata field. Once set, it appears in your dashboard activity table and can be used to filter requests by use case — useful for separating production traffic, tracking which part of your app made a request, or measuring cost and latency per feature.


Setting a call name

Pass call_name inside the metadata object:

const response = await client.chat.completions.create({
  model: 'sansa-auto',
  messages: [{ role: 'user', content: '...' }],
  metadata: {
    call_name: 'summarize-email',
  },
});
response = client.chat.completions.create(
    model="sansa-auto",
    messages=[{"role": "user", "content": "..."}],
    metadata={
        "call_name": "summarize-email",
    },
)
  • Must be ≤ 64 characters.
  • Cannot be empty or whitespace-only.
  • Invalid values return 400 with code invalid_call_name.

Using it in the dashboard

Once requests are tagged, the Activity page shows a Call Name column and a filter dropdown. Select a name to scope the table to only those requests.

Clicking any row opens the request detail view, which shows the call name in the Overview section alongside the request ID, status, and timestamp.


Naming suggestions

There's no required format. A few patterns that work well:

PatternExample
Feature namesummarize-email, generate-code, extract-data
Environmentproduction, staging, testing
Customer or tenantcustomer-acme, tenant-xyz
Pipeline stepstep-1-parse, step-2-summarize

Currently, call_name is the only metadata key that Sansa reads. Other keys are accepted for OpenAI SDK compatibility but are not stored.