Word timestamps
Start and end time for every word, for subtitles and clip cutting. ×1.5; segment timestamps are free.
Whisper API Pricing From $0.052 per Audio Hour
Whisper CT2 returns the transcript with word timestamps, speaker labels and source metadata in one call. Upload a file or send a link from YouTube, X, Twitch, Kick or TikTok. Files and YouTube cost $0.052 per hour, Twitch, Kick and X links $0.15 to $0.26 per hour, and TikTok is priced per video. The /price endpoint gives the amount before the job runs.
$5 in free credits to start. No credit card required.
- per hour for audio files and YouTube links
- $0.052
- per hour for Twitch and Kick VODs
- $0.15
- per hour for X videos
- $0.26
- per TikTok video up to 1 minute
- $0.015
Speech-to-Text API Pricing for 9 Models from 6 Providers
List prices and documented options from each vendor's pricing page and docs, checked 27 October 2026. The table covers price and options only. Run your own audio through the models you consider before you choose.
| Provider and model | Per audio hour | Word timestamps | Speaker labels | Input |
|---|---|---|---|---|
| deAPI · Whisper CT2 | $0.052 for files and YouTube | Yes, ×1.5 | Yes, ×1.5 | File, or link from YouTube, X, Twitch, Kick, TikTok, X Spaces. Twitch and Kick $0.15 per hour, X $0.26, TikTok per video |
| Groq · whisper-large-v3-turbo | $0.04 | Yes | Not in docs | File (25 MB free tier, 100 MB dev tier) or URL; minimum billed 10 s |
| Groq · whisper-large-v3 | $0.111 | Yes | Not in docs | File or URL; minimum billed 10 s |
| AssemblyAI · Universal-2 | $0.15 | Yes | +$0.02/h | File up to 2.2 GB or public URL up to 5 GB |
| OpenAI · gpt-4o-mini-transcribe | $0.18 | No | Separate deprecated model | File up to 25 MB |
| ElevenLabs · Scribe v2 | $0.22 | Yes | Included, up to 32 speakers | File or URL, including YouTube and TikTok links |
| Deepgram · Nova-3 (monolingual) | $0.258 | Yes | Included | File or URL |
| OpenAI · gpt-transcribe | $0.27 | No (whisper-1 only) | Not documented | File up to 25 MB |
| OpenAI · Whisper (whisper-1) | $0.36 | Yes | No | File up to 25 MB |
deAPI price from the /price endpoint on 27 October 2026 for a one-hour file. Groq describes its turbo model as a pruned, fine-tuned version of whisper-large-v3. AssemblyAI's default model is Universal-3.5 Pro at $0.21 per hour. OpenAI's docs mark whisper-1, gpt-4o-transcribe, gpt-4o-mini-transcribe and gpt-4o-transcribe-diarize as deprecated, shutting down on 26 February 2027.
What the Transcription API Price Includes
-
-
Speaker labels
Who said each segment, for interviews, podcasts and calls. ×1.5, applied once even with word timestamps on.
-
Links from five platforms
YouTube, X, Twitch, Kick and TikTok, plus X Spaces. We fetch the media, so you don't download and re-upload it.
-
Source metadata
Title, channel, description, duration and, where the platform gives them, view counts, for $0.005.
-
Volume and control
Limits go up after the first payment, and [email protected] handles higher volumes. Spend caps per key, prepaid credit only.
-
Data handling
Generated files are deleted no later than 24 hours after job creation.
- $5 in free credits
- No credit card required
- Prepaid, no subscription
Price Your Own Audio
Create a key, call /price with the length of your files and compare. $5 in credits after email verification, no card.
See also YouTube Transcript API, TTS API Pricing and Transcription MCP Server.
Frequently Asked Questions
Transcription API prices and options
It depends on where the audio comes from. Audio files and YouTube links cost $0.052 per hour, Twitch and Kick VODs $0.15, X videos $0.26, and a TikTok video up to one minute $0.015. The /price endpoint returns the exact amount before each job runs.
Word-level timestamps or speaker labels multiply the price by 1.5, once, even with both on. Source metadata for links adds $0.005. Segment timestamps are free.
Yes. OpenAI's docs say whisper-1, gpt-4o-transcribe, gpt-4o-mini-transcribe and gpt-4o-transcribe-diarize shut down on 26 February 2027.
Yes. deAPI's OpenAI-compatible gateway serves /v1/audio/transcriptions. Change the base URL, the API key and the model name in your OpenAI client. The gateway takes audio files up to 20 MB.
Whisper CT2 (model ID WhisperLargeV3Ct2), Whisper on the CTranslate2 runtime. It supports segment and word timestamps, speaker labels, source metadata and links from five platforms.