Speech-to-text API plans compared
9 plans from 5 providers, checked 2026-11-22 against the official pages.
Cheapest at 100 audio hours a month
The cheapest plan of each provider for 100 audio hours a month, before tax, from the providers' own pricing pages. 9 of 9 plans state a monthly cost at this volume; 0 cannot serve it and 0 give no cost for it.
Monthly workload
More conditions on the full search · Price statistics.
Side by side
Yes, no and not stated hold for every plan of the provider; "Differs by plan" means its plans differ. A cell that says "not stated on the pages we read" is not a no. Each plan page has the official sentence and the date we read it.
Price at 100 audio hours a month 1 row
| Feature | AssemblyAI | Google Cloud | Speechmatics | Deepgram |
|---|---|---|---|---|
| Cheapest plan | $15 · Pay as you go, pre-recorded (Universal-2) | $18 · V2 API Dynamic Batch (pay as you go) | $24 · Self-serve pay as you go, pre-recorded (Melia 1 base rate) | $25.80 · Pay As You Go, pre-recorded (Nova-3 Monolingual) |
Conditions 6 rows
| Feature | AssemblyAI | Google Cloud | Speechmatics | Deepgram |
|---|---|---|---|---|
| Speaker diarization is stated | Differs by plan | not stated on the pages we read | Yes | Yes |
| Billed by the second with no minimum per request is stated | Differs by plan | Yes | Yes | not stated on the pages we read |
| Korean is supported | Differs by plan | Yes | Yes | Yes |
| Streaming (real-time) is stated | Differs by plan | not stated on the pages we read | not stated on the pages we read | not stated on the pages we read |
| Customer content is not used to train the vendor's models (by default) | No | Yes | Yes | No |
| The provider holds a SOC 2 report | Yes | Yes | Yes | not stated on the pages we read |