Groq (BYOK)

Groq runs Whisper on its own hardware, which makes it the fastest of the BYOK options in practice. The key is yours and Groq bills you directly.

Setup

1. Create a key at console.groq.com/keys
2. In Speakly, open Settings → Transcription and choose the Your API Key engine
3. Pick Groq in the Provider row
4. Paste the key into Groq API Key — it starts with gsk_
5. Click Test Connection; "API key is valid" means the key is stored
6. Pick a model under Whisper Model

The model list comes from your account

Speakly does not ship a fixed Groq model list. Once the key is present it asks Groq which transcription models the account can use and fills the Whisper Model dropdown from the answer, so new models show up without a Speakly update. Without a key the field reads "Add a valid API key to load available models."; while fetching, "Loading models…"; if the account returns nothing, "No models found — check that your API key is valid." The default selection is whisper-large-v3-turbo.

Deepgram and Mistral model lists are fetched the same way. Google is the exception: its Speech-to-Text models are a curated list inside Speakly, because it has no usable model-listing API.