Deepgram · Audio
Nova-4 API pricing
deepgram/nova-4
Transcription. Nova-4 by Deepgram, called through one INFRO key alongside every other model you run. We price it 27% below Deepgram's direct rate — the same model, the same weights, reached through a different integration.
Pricing
Deepgram direct
$0.26
per audio hour
INFRO
$0.19
per audio hour
You save
27%
on every unit
Illustrative launch rates, reconciled against published provider pricing on 2026-08-24. Live rates come from GET /v1/models once your key is active.
Calling Nova-4
POST /v1/audio/speechAt launch — Committed to the first release. In build now, not usable yet.import os
from infro import Infro
client = Infro(api_key=os.environ["INFRO_API_KEY"])
transcript = client.audio.transcriptions.create(
model="deepgram/nova-4",
file=open("meeting.mp3", "rb"),
)What this modality supports
- Text to speech
- Music generation
- Transcription
- Diarization
Text-to-speech with streaming playback, full-track music generation, and transcription with timestamps and diarization. Billed per character generated, per track, and per hour transcribed.
Alternatives for transcription
Same modality, nearest in price. All of them share the request shape above, so comparing them is a one-word change.
Work out what Nova-4 would cost you
Put your current spend on this model into the calculator and see the same workload priced here — or have us read your real usage.