Deepgram Nova-3 at $0.0043 a minute: an hour of audio, and Sume STT

Deepgram's pricing page lists Nova-3 pre-recorded speech-to-text at $0.0043 per minute for monolingual audio and $0.0052 per minute for multilingual. An hour is $0.258 and $0.312. Sume STT 1.0 is listed at $0.01 per audio minute, which is $0.600 for an hour.

Deepgram's prices are from its pricing page, read 2026-10-01; Sume's from the rate card. This post compares prices only, not accuracy.

What is an hour of audio on each?

Per-minute price times 60.

How does Sume reserve the cost?

The request takes an optional duration_seconds from 1 to 600. If you omit it, Sume reserves one minute, so pass the real length for a correct reservation. One request covers at most 10 minutes of audio, so an hour is six or more requests.

What does a request look like?

A public HTTPS audio URL, with the length and a language hint.

curl -X POST https://api.sume.com/v1/stt-1.0/transcribe \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "audio_url": "https://example.com/interview.mp3",
    "duration_seconds": 540,
    "language_code": "en"
  }'

Is the cheapest rate the right pick?

Only if the transcript meets your needs. Sume always returns word timings and can add sentence segments; see the API reference. Compare accuracy on your own audio before choosing on price, at these rates a sample costs almost nothing.