Skip to main content
The Usage screen shows every call to your endpoints. Filter by key, endpoint, model, status, and time.
Valkyrie Usage screen

Usage: a one-line summary, per-key rollups, and a filterable request log.

​
Summary

A single line at the top sums up Total requests, Total tokens, and Avg latency for the selected time range.

​
Usage by key

The Usage by key table breaks down requests, error rate, total tokens, average and p95 latency, and when each key was last used. Click a column header to sort. Calls made with a dashboard session (for example from the Playground) rather than an API key are grouped as No key (JWT / anonymous).

​
Requests

The Requests table lists individual calls with their time, API key, deployment, model, HTTP status, whether the response was streamed, latency, and token count. Filter by:
  • API key, deployment, and model
  • Status: All, 2xx, 4xx, or 5xx
  • Streamed: All, Streamed, or Non-Streamed
  • Date range, plus a Replayable only toggle
Choose the Order, then click Apply. Clear resets the filters.
Use the status filter to find failing calls fast: filter to 4xx or 5xx and check the deployment column to see which endpoint is erroring. See Errors.