From audio to text.
You send a file to POST /v1/audio/transcriptions. Short recordings come straight back, long ones run in the background and you poll the status.
How long can a file be?
The limit depends on whether you wait for the response. Beyond the marked range, you need background=true.
up to 600 s
In the background up to 28,800 s
up to 24 MiB
In the background up to 200 MiB
Beyond that, the limits of the model apply. Results stay available for 24 hours.
These models are available.
The table reads the same catalog as the pricing page. Prices are in euros per minute, from the lowest-priced mode.
| Model | File from€ per minute | Timestamps |
|---|---|---|
groq/whisper-large-v3-turbo
Groq
|
0.0008 | yes |
ionos/openai/whisper-large-v3
IONOS
|
0.0032 | yes |
openai/gpt-transcribe
OpenAI
|
0.0051 | no |
Silence is a successful, billed result with empty text. There is no provider switch here either. FlowModels does not currently offer speaker labels.
A run in the background.
A background job answers immediately with 202. You then poll status and progress with GET /v1/audio/transcriptions/{id}. There are no webhooks for this route.
curl https://api.flowmodels.one/v1/audio/transcriptions \
-H "Authorization: Bearer $FLOWMODELS_API_KEY" \
-H "Idempotency-Key: recording-2026-11-09-001" \
-F 'model=groq/whisper-large-v3-turbo' \
-F 'language=en' \
-F 'timestamps=true' \
-F 'background=true' \
-F 'file=@recording.wav'
{
"id": "018f5678-1234-7abc-9def-0123456789ab",
"object": "audio.transcription",
"model": "groq/whisper-large-v3-turbo",
"status": "processing",
"progress": 0.42,
"flowai_request_id": "01HXYZ8M3F6V2KQ"
}
Status of a job
queuedprocessingcompletedfailedcancelled
Cancel and delete
POST /v1/audio/transcriptions/{id}/cancel starts no further segments; what is running still finishes. DELETE only works on finished jobs. FlowModels stores transcripts encrypted.