Skip to main content

Transcribe speech audio

POST 
/v1/audio/transcriptions

OpenAI-compatible speech-to-text endpoint. The requested model must already be placed and running as a speech-to-text model. stream=true returns typed SSE events for cards declaring audio.supports_streaming=true; response_format=ndjson retains progressive NDJSON framing.

Request​

Responses​

Successful Response