Deepgram vs Whisper API on price
Deepgram is cheaper for both recorded and live audio at every volume from 10 to 10,000 hours a month. At 100 hours a month, for recorded audio, Deepgram's Nova-3 Monolingual costs $0.258 an hour against $0.36 for OpenAI's Whisper (whisper-1), 28% less; for live audio, Deepgram's Nova-3 Monolingual costs $0.462 an hour against $1.02 for OpenAI's gpt-realtime-whisper, 55% less.
Each side's cheapest option at each volume, at regular pay-as-you-go prices; bold marks the cheaper. Price only: accuracy, languages and latency differ.
Recorded audio (batch)
| Audio a month | Deepgram | Whisper API |
|---|---|---|
| 10 hours | $2.58 ($0.258/h) Nova-3 Monolingual | $3.60 ($0.36/h) Whisper (whisper-1) |
| 100 hours | $25.80 ($0.258/h) Nova-3 Monolingual | $36 ($0.36/h) Whisper (whisper-1) |
| 1,000 hours | $258 ($0.258/h) Nova-3 Monolingual | $360 ($0.36/h) Whisper (whisper-1) |
| 10,000 hours | $2,580 ($0.258/h) Nova-3 Monolingual | $3,600 ($0.36/h) Whisper (whisper-1) |
Live audio (streaming)
| Audio a month | Deepgram | Whisper API |
|---|---|---|
| 10 hours | $4.62 ($0.462/h) Nova-3 Monolingual | $10.20 ($1.02/h) gpt-realtime-whisper |
| 100 hours | $46.20 ($0.462/h) Nova-3 Monolingual | $102 ($1.02/h) gpt-realtime-whisper |
| 1,000 hours | $462 ($0.462/h) Nova-3 Monolingual | $1,020 ($1.02/h) gpt-realtime-whisper |
| 10,000 hours | $4,620 ($0.462/h) Nova-3 Monolingual | $10,200 ($1.02/h) gpt-realtime-whisper |
What else changes the bill
| On the bill | Deepgram | Whisper API |
|---|---|---|
| Free tier or credit | Every new account gets $200 of free credit, which doesn't expire until used. | No free allowance or credit is listed on OpenAI's pricing page. |
| Billing | Billed per second, with no rounding up. | OpenAI's pricing page doesn't say how partial minutes are counted. |
| Speaker labels, recorded audio | Included. | Through the separate gpt-4o-transcribe-diarize model, at the same estimated $0.006 a minute as gpt-4o-transcribe. |
| Speaker labels, live audio | +$0.12 an hour. | Not listed on the pricing page. |
Promotions running now: Deepgram Nova-3 Monolingual (streaming) at $0.0048 a minute instead of $0.0077, no end date given; Deepgram Flux English (streaming) at $0.0065 a minute instead of $0.0077, no end date given; Deepgram Nova-3 Multilingual (streaming) at $0.0058 a minute instead of $0.0092, no end date given. The tables use regular prices.
Not counted here: OpenAI's gpt-transcribe (batch, $0.27 an hour), OpenAI's gpt-4o-mini-transcribe (batch, $0.18 an hour, an estimate), OpenAI's gpt-4o-transcribe (batch, $0.36 an hour, an estimate), OpenAI's gpt-4o-transcribe-diarize (batch, $0.36 an hour, an estimate) and OpenAI's gpt-live-transcribe (streaming, $1.02 an hour).
Every model on each side
Deepgram
| Model | Audio | Per minute | Per hour |
|---|---|---|---|
| Nova-3 Monolingual | batch | $0.0043 | $0.258 |
| Whisper Large | batch | $0.0048 | $0.288 |
| Nova-3 Multilingual | batch | $0.0052 | $0.312 |
| Nova-3 Monolingual | streaming | $0.0077 | $0.462 |
| Flux English | streaming | $0.0077 | $0.462 |
| Flux Multilingual | streaming | $0.0078 | $0.468 |
| Nova-3 Multilingual | streaming | $0.0092 | $0.552 |
Whisper API
| Model | Audio | Per minute | Per hour |
|---|---|---|---|
| Whisper (whisper-1) | batch | $0.006 | $0.36 |
| gpt-realtime-whisper | streaming | $0.017 | $1.02 |
Full details: Deepgram pricing and Whisper API pricing. Compare every vendor in the speech-to-text cost calculator, or see the cheapest speech-to-text API.
Prices on this page were checked against the official pricing pages on .