How billing works
Voice services use the unit the API exposes for that operation: characters for synthesis and minutes for transcription.
- Text-to-speech is charged per 1K characters.
- Transcription is charged per minute of audio.
- The two services do not share a token rate.
Price in practice
Text-to-speech and transcription are intentionally shown together because both are voice features, but they are not interchangeable billing units.
For predictable spending, count characters for speech output and minutes for audio transcription before sending the request.
Rate card
ServicePriceHow it is applied
Gemini 3.1 Flash TTS Preview8 CrPer 1K characters
Gemini 3.5 Flash transcription2 CrPer minute
Questions
Why are the units different?
Speech synthesis is measured from text volume, while transcription is measured from audio duration.
What unit is used for speech synthesis?
Text-to-speech is billed per 1,000 characters.
What unit is used for transcription?
Transcription is billed per minute of audio.