Platform

Usage and billing

Understand usage measurements and credit deductions.
View as Markdown

Usage measurement

APIUnitReturned in
TTSInput charactersx-usage-input-characters response header
STTInput audio duration per utterance, in secondsusage.input_audio_seconds in each transcript.completed event

TTS uses input characters for both Free and Partner models. Characters are counted as Unicode code points after trimming surrounding whitespace and before text normalization. Internal spaces and punctuation count: " Hello. " counts as 6 characters.

STT reports the audio duration assigned to each utterance, excluding silence discarded by VAD.

These values report usage for the current request or utterance.

Pricing

Partner models use the rates on Models and pricing:

  • TTS: input characters ÷ 1,000,000 × the model’s price per million characters.
  • STT: input audio seconds ÷ 3,600 × the model’s price per audio hour.

For example, 100,000 input characters cost 0.50∗∗withTTSStandard.Tenminutesofinputaudiocost∗∗0.50** with TTS Standard. Ten minutes of input audio cost **0.02 with STT Fast.

Credits

Successful Partner requests debit credits. Failed and cancelled requests do not. Free-model and Playground requests do not consume credits.

View organization usage and costs on the Usage page, and your credit balance on the Billing page.