The Groq steps run on your company's own Groq key. Ask a model answers a prompt with text or named fields, using the open models Groq hosts. Transcribe audio turns a recording — a voice note, a call, a customer's voicemail — into text with Whisper.

Who can do this

Workspace Admins and Editors, on every plan.

Before you start

A Groq connection in Connections — see Connect Groq.

Steps

Ask a model:

  1. Select + where the question should be asked. In the step picker, open AI, then Groq, and select Ask a model.
  2. In Connection, choose the connection with your key.
  3. Optional: in Model, choose a model. The list is read from Groq with your key; left empty, the step uses the general model set in Bizomate. To use a model the list does not show, select Use a value and type its name.
  4. Optional: in Instructions, say who the model is and how to answer.
  5. In Prompt, ask the question, with the item's values in {{ }} — Summarise this delivery note in one line: {{ $json.text }}
  6. Under Give me back, choose Fields (the default) or Text. For Fields, add a row per answer: the name, and what goes in it.
  7. Select Run this step, then check Output. Pin it while you build — see Pin a step's output.

Transcribe audio:

  1. In the step picker, under Groq, select Transcribe audio, and choose the Connection.
  2. Under File, select Add a file and drag in the recording from an earlier step — MP3, M4A, WAV, OGG or WebM. See Pass files between steps.
  3. In Model, keep the Whisper model shown, or choose the other from the list.
  4. In Language, keep Detect (the default), or choose the language spoken. Choosing it helps with short recordings.
  5. Optional: in Prompt, write words the recording uses that a model might mishear — Consulace, Northwind Logistics, NW-20417.
  6. Select Run this step.

Output

Step Keys
Ask a model, Fields each field by its name, and usage
Ask a model, Text text and usage
Transcribe audio text, language, duration (seconds of audio), segments (each with start, end and text), usage

usage holds model, tokensIn, tokensOut and seconds — for audio, audioSeconds in place of tokens. Output shows them as the usage strip.

Good to know

  • Each item is a separate call, and a separate cost.
  • Recordings up to 25 MB. Split a longer call, or send it compressed (MP3 or M4A), before it reaches the step.
  • Rate limits are low on Groq's free tier. Set If this step fails to try again — see Make a step try again when it fails.
  • A request Groq refuses — an unknown model, a file it cannot read — is not tried again.

If something goes wrong

What the run says Why What to do
Groq refused the API key. Replace it on Connections. The key is wrong or revoked. Replace the key on Connections.
Groq says too many requests, or the account is out of credit. Set this step to try again, or check the account. Groq's rate limit. Set the step to try again, or change tier in Groq.
Groq answered 400: … Groq refused the request — an unknown model, or audio it cannot read. Read Groq's reason; check the model and the file.
File is empty. Choose a file from an earlier step. File has no file. Drag in a file from Data from earlier steps.
… is over 25 MB, the most Groq transcribes in one go. … The recording is too long. Split it, or send it compressed.
Groq refused: … (Transcribe audio) Groq refused the request — audio it cannot read, an unknown model, or a key it does not accept. Read Groq's reason; for the key, replace it on Connections.
Groq did not give back the fields asked for. Try again, or make the instructions plainer. The answer was not in the fields asked for. Make Instructions plainer, or the fields fewer.
Could not reach Groq: … / Groq did not answer in time. Groq did not answer. Try again.