Speech-to-text API pricing compared (per hour of audio)
Providers quote transcription prices per minute, per hour or per token, which makes them hard to compare. This table converts every verified price to a cost per hour of audio and separates batch from live services. Price your own job with the speech-to-text cost calculator; for OpenAI specifically see Whisper API pricing. Topic hub: AI voice, speech and music.
Prices last verified against official provider pages on . How we verify prices. Written by the Chelure Enterprises Editorial Team.
All prices, cheapest first
| Provider | Model | Type | Listed price | Per hour of audio |
|---|---|---|---|---|
| xAI | xAI Speech to Text (batch) | Batch | $0.10 per hour | $0.10 |
| Z.ai | GLM-ASR-2512 | Batch | $0.0024 per minute | $0.144 |
| OpenAI | OpenAI gpt-4o-mini-transcribe | Batch (deprecated) | $0.003 per minute | $0.18 |
| xAI | xAI Speech to Text (streaming) | Live | $0.20 per hour | $0.20 |
| ElevenLabs | ElevenLabs Scribe v3 / v2 | Batch | $0.20 per hour | $0.20 |
| Mistral | Voxtral Mini Transcribe 3 | Batch | $0.0035 per minute | $0.21 |
| ElevenLabs | ElevenLabs Scribe v3 Live | Live | $0.22 per hour | $0.22 |
| Mistral | Voxtral Small Transcribe 3 | Batch | $0.004 per minute | $0.24 |
| OpenAI | OpenAI gpt-transcribe | Batch | $0.0045 per minute | $0.27 |
| Gemini 3.5 Transcribe | Batch | $0.005 per minute | $0.30 | |
| OpenAI | OpenAI Whisper | Batch (deprecated) | $0.006 per minute | $0.36 |
| OpenAI | OpenAI gpt-4o-transcribe | Batch (deprecated) | $0.006 per minute | $0.36 |
| Mistral | Voxtral Mini Transcribe Realtime 3 | Live | $0.008 per minute | $0.48 |
| Gemini 3.5 Transcribe Live | Live | $0.009 per minute | $0.54 | |
| OpenAI | OpenAI gpt-live-transcribe (live) | Live | $0.017 per minute | $1.02 |
Per-hour figures are our arithmetic from the listed price (per-minute price times 60). Prices marked as estimates by the provider are blended token-based figures. Batch prices apply to finished recordings; live prices apply to streaming audio and are higher.
Reading the table
- Batch is cheapest. If you do not need results while the audio is still being spoken, use a batch model.
- Deprecated models. OpenAI's whisper-1, gpt-4o-transcribe and gpt-4o-mini-transcribe shut down on February 26, 2027; budget on gpt-transcribe instead.
- Price is not accuracy. The cheapest model may be less accurate for your language, accents or audio quality; sample your own recordings.
- Watch the add-ons. ElevenLabs charges extra for keyterm prompting and transcript editing; see ElevenLabs pricing.
Frequently asked questions
What is the cheapest speech-to-text API?
On the batch prices we have verified, xAI Speech to Text (batch) is the cheapest at $0.10 per hour. Cheapest is not best: accuracy, language coverage and features differ, so test on your own audio.
What is the difference between batch and live transcription?
Batch services transcribe a finished recording; live (streaming or realtime) services transcribe audio as it arrives and are priced higher, for example OpenAI's live models are $0.017 per minute against $0.0045 for gpt-transcribe.
Why are some prices marked estimates?
Some providers bill tokens and publish an approximate per-minute figure. OpenAI lists gpt-4o-transcribe at about $0.006 per minute and Google lists Gemini 3.5 Transcribe at about $0.005 per minute as blended estimates.
Are any of these models being retired?
OpenAI says whisper-1, gpt-4o-transcribe and gpt-4o-mini-transcribe are deprecated and will shut down on February 26, 2027.
How do I price my own job?
Multiply hours of audio by the per-hour price, or use the speech-to-text cost calculator with the model of your choice.
Sources
- OpenAI API pricing
- Google Gemini API pricing
- ElevenLabs API pricing
- Mistral pricing
- xAI models and pricing
- Z.ai pricing
Primary sources opened on November 13, 2026. Prices come from our verified price table (see the methodology).