# Whisper large v3 Turbo speech to text

`groq/whisper-large-v3-turbo` · OpenAI · audio

Whisper large v3 Turbo runs on Groq and prioritizes transcription speed. It turns multilingual recordings into text in the original language.

[Model page](https://sociaro.com/models/groq/whisper-large-v3-turbo) · [JSON](https://sociaro.com/models/groq/whisper-large-v3-turbo.json) · [Model pricing](https://sociaro.com/pricing?model=groq%2Fwhisper-large-v3-turbo) · [Documentation](https://docs.sociaro.com/models/groq-whisper-large-v3-turbo/)

## Use cases

- Produce first-pass transcripts for a podcast backlog
- Turn a newly uploaded voice note into searchable text

## Technical details

- API model ID — groq/whisper-large-v3-turbo
- Provider — OpenAI
- Modality — Audio
- Input — Audio file
- Output — Transcript
- Endpoint — POST /v1/audio/transcriptions
- Response formats — json · verbose_json · text

## Request parameters

- `file` (audio file (multipart), required) — The recording to transcribe, sent as a multipart form field.
  - flac · mp3 · m4a · mpeg · mpga · ogg · wav · webm
- `language` (string, optional, default `detected`) — The spoken language. Naming it improves accuracy and speed; leave it out to let the model detect it.
  - an ISO-639-1 code, e.g. en, de, ru
- `prompt` (string, optional) — Context or spelling to steer the transcript, such as names and terms, in the audio's language.
  - text
- `response_format` (string, optional, default `json`) — verbose_json adds segments with timestamps.
  - json · verbose_json · text
- `temperature` (number, optional, default `0`) — Higher is more varied; 0 is the most deterministic.
  - 0 – 1

## Pricing

Rates in USD. Catalogue timestamp 2026-11-21T02:23:44Z.

- USD 0.048 per hour of input audio · active · billed at $0.000013333333 a second

- Runs on Groq.
