Skip to main content
Fonoster measures what every call, test and eval uses: minutes, transcription, speech, language-model tokens and the rest. The Usage page and the API show it in quantities only, never money. That’s the same on Fonoster Cloud and on your own install.

Open the Usage page

In the dashboard, open Usage in the sidebar. Pick a month at the top right; the page shows:
  • Calls: minutes from the moment a call is answered until it ends. Ringing isn’t counted.
  • Browser tests: minutes of calls you made from the dashboard, apart from real calls.
  • Noise cancelling, Recordings: minutes each ran.
  • Machine detection: how many outbound calls were checked for an answering machine.
  • AI usage: transcription minutes, speech characters, language-model and knowledge-base tokens, by vendor and model. Your key means the vendor was called with your own vendor key, so that vendor bills you directly. Fonoster key means the platform’s key was used.
  • Storage: the most your call records and recordings took up during the month.
Usage shows up once a call ends. A month with nothing in it says so.

Get it from the API

Usage/GetUsage returns totals for a range of days, split by meter and by any of day, workspace, provider, model, key_source and origin:
Each row has a meter, its unit and a quantity. Workspace owners can pass allOwned: true to get every workspace they own. See the SDK reference for every meter.

How it’s counted

Evals count only their language-model tokens and show up with origin eval. Calls in a private region are counted the same way; only quantities leave the region, never call content.