Transcription / Cost calculator
Transcription cost calculator
Enter how many hours of audio you have. The table multiplies each vendor's listed price per audio hour for batch or pre-recorded audio, read from the vendor's own page on October 14, 2026. We run MyAi, and several vendors below list a lower price than we do; the table shows them first.
A weekly one-hour podcast with a 5-year back catalog is about 260 hours.
| Service | Per hour | List cost | After credit |
|---|---|---|---|
Groq whisper-large-v3-turbo (Batch API) 50% batch discount; results in 24 hours to 7 days | $0.02 | $2.00 | — |
Groq whisper-large-v3-turbo 25 MB / 100 MB per file | $0.04 | $4.00 | — |
Salad Transcription API Lite 0–8,333 h/mo tier | $0.08 | $8.00 | — |
Together AI Whisper Large v3 | $0.09 | $9.00 | — |
MyAi whisper-large-v3-turbo (us) speaker labels and captions included | $0.10 | $10.00 | $9.00 |
Groq whisper-large-v3 | $0.111 | $11.10 | — |
AssemblyAI Universal-2 +$0.02/hr for speaker labels | $0.15 | $15.00 | $0.00 |
OpenAI gpt-4o-mini-transcribe | $0.18 | $18.00 | — |
Salad Transcription API 0–8,333 h/mo tier; $0.18 and $0.16 above that | $0.20 | $20.00 | — |
AssemblyAI Universal-3.5 Pro +$0.02/hr for speaker labels | $0.21 | $21.00 | $0.00 |
Deepgram Nova-3 (Growth) Growth plan rate; plan terms not checked | $0.22 | $22.00 | $0.00 |
Deepgram Nova-3 (pay as you go) | $0.26 | $26.00 | $0.00 |
Deepgram Whisper Large | $0.29 | $29.00 | $0.00 |
OpenAI gpt-4o-transcribe | $0.36 | $36.00 | — |
OpenAI gpt-4o-transcribe-diarize the OpenAI model with speaker labels | $0.36 | $36.00 | — |
How to read this. "After credit" subtracts the free credit each vendor lists for new accounts where we confirmed it: MyAi's 10 free hours, AssemblyAI's $50 and Deepgram's $200. A dash means no credit is counted. List prices only; volume discounts, minimum billing per file (Groq bills at least 10 seconds per request, Salad at least 36 seconds) and taxes are not included.
Price is not the whole job. Groq and OpenAI cap files at 25 MB (Groq 100 MB on its dev tier), so long recordings must be compressed or split first. MyAi takes a list of URLs in one call, splits files up to 4 hours for you, and returns speaker labels and SRT/VTT; it is not for confidential audio, because it runs on independent operators' GPUs. The long-audio comparison covers limits, webhooks and caption formats for each option.
Not included. Replicate (openai/whisper): No per-audio-hour price: billed for GPU time (T4 at $0.000225/sec); the model page estimates about $0.0012 per run. Fireworks AI: Could not verify: no speech-to-text price on the pricing page.
Prices checked October 14, 2026. Prices change; check each source before you commit.
Try it on your own audio
10 free audio hours, no card. Then $0.10 per audio hour.
Start free with 10 hours