Whisper API alternative: speech to text prices compared
What one hour of pre-recorded audio costs at 4 speech to text APIs, read from each pricing page on 14 November 2026, next to Glyphcall Voice at $0.05 for up to 60 minutes.
| Service | Model | Price as listed | One hour of audio | To start |
|---|---|---|---|---|
| Glyphcall Voice | Whisper large-v3-turbo | $0.05 per file up to 60 minutes; $0.01 per file up to 5 MB | $0.05 | No account: card credits from $1, or an agent pays per call in USDC |
| AssemblyAI | Universal-2 | $0.15 per hour | $0.15 | Account; $50 free credit, no card |
| OpenAI | gpt-4o-mini-transcribe | $0.003 per minute | $0.18 | Account and API key; No free API credit stated |
| ElevenLabs | Scribe v2 / v3 | $0.20 per hour | $0.2 | Account; Sign-up; no free allowance stated on the API price page |
| AssemblyAI | Universal-3.5 Pro | $0.21 per hour | $0.21 | Account; $50 free credit, no card |
| Deepgram | Nova-3 (one language) | $0.0043 per minute | $0.258 | Account; $200 free credit, no card |
| OpenAI | Whisper | $0.006 per minute | $0.36 | Account and API key; No free API credit stated |
Sources: ElevenLabs · OpenAI · Deepgram · AssemblyAI. Groq's pricing page showed no figure that day and is not listed. The cheapest listed rival per hour is AssemblyAI Universal-2 at $0.15.
What you get
- $0.05 for up to 60 minutes of MP3 or WAV (25 MB) by URL or upload; $0.01 for any short file up to 5 MB.
- SRT or VTT subtitles and word timestamps in the same answer.
- $0.05 for a summary with key points and action items on top of the transcript.
- 90+ languages. No account: card credits from $1, or an agent pays per call in USDC.
- Failed calls are never charged. Audio is held in memory for the request only, never stored or logged.
- Free and private: transcribe in your browser, the file never leaves your device.
One request
curl "https://agent-data-tools.revmesh2074.workers.dev/v1/audio/transcriptions" \
-H "Authorization: Bearer $API_KEY" \
-F file=@podcast.mp3 -F model=whisper-1 -F response_format=srtQuestions
Is it the same Whisper model?
It runs Whisper large-v3-turbo, the open model OpenAI released, on Cloudflare's network. OpenAI's own API lists Whisper and its newer gpt-4o transcribe models.
Can I keep my OpenAI code?
Yes. POST /v1/audio/transcriptions takes OpenAI's request (file, language, response_format json, text, srt, vtt or verbose_json). Set the client's base URL here and use your key as the API key.
How is a long file priced?
One call reads up to 60 minutes of MP3 or WAV (25 MB) for $0.05. A short clip in any format up to 5 MB is $0.01. Per-minute services cost less than us only for clips of a few minutes.
Where are the others better?
Speaker labels (diarization), live streaming transcription, and models tuned for accents and noise. Deepgram, AssemblyAI and ElevenLabs offer those; we do not yet.
Text to speech prices
| Service | Model | Per 1,000 characters |
|---|---|---|
| Glyphcall Voice | Aura-1 / Aura-2 | $0.02 / $0.04 |
| OpenAI | tts-1 | $0.015 |
| OpenAI | tts-1-hd | $0.03 |
| Deepgram | Aura-2 | $0.03 |
| ElevenLabs | Flash / Turbo | $0.04 |
| ElevenLabs | Multilingual v2, v3 | $0.08 |
OpenAI's tts-1 costs less per character than our voices. Sources: ElevenLabs · OpenAI · Deepgram.