Pay-as-you-go speech-to-text pricing, with no subscription

Speechall is pay-as-you-go. You buy prepaid credits, and transcription usage is charged against your credit balance. There is no subscription and no recurring monthly charge: checkout is a one-time payment for the credits you choose to buy.

There is a minimum top-up amount. It is set in the active billing configuration, so this page does not repeat it; the billing page shows the current minimum, the service fee and the exact total before you pay.

How it works

  1. Buy credits. In billing, enter the amount of credit you want to add. The quote shows that base amount plus a service fee, and the number of credits you will receive. You pay the total through Stripe Checkout, where applicable tax is added.
  2. Use the API. Each transcription request uses the model you specify. Usage is charged against your balance at that model’s rate.
  3. Watch your balance. The billing page shows your remaining credits and recent purchases. The usage page shows requests and audio duration by API key and by model.

When your balance runs low, you top up again. Nothing renews automatically.

Currency and credits

Top-ups are priced in EUR: you enter a base amount in EUR, and checkout is charged in EUR. The base amount is converted to USD at the exchange rate shown in the quote, and credits are denominated in USD. One credit is $0.01 USD equivalent. The service fee is charged on top of the base amount and is not converted into credits.

Do not compare a EUR checkout total directly with a USD per-hour model rate. Work out how many credits you receive, then estimate usage from the model’s rate.

Fees

The service fee and the minimum top-up can change, so they are not listed here. Open billing, enter an amount, and the quote shows the fee breakdown before you go to checkout. For a longer walkthrough of the quote, see the speech-to-text API cost guide.

One balance for every provider and model

Credits are not tied to a single provider. The same balance pays for any model in the catalog, whichever provider runs it. Trying several providers does not require several accounts, API keys or invoices, and you can compare their transcripts in the playground using one balance. Switching between them means changing the model ID in your request. See Getting Started for a first request.

Find the rate for each model

Every model has its own rate, and rates differ between models and between file (batch) and streaming use. The table shows the current rate per audio hour for each model, read from the live catalog. Each model page repeats it with the model’s features. Rates can change, so check here or in the catalog rather than copying a figure.

ModelProviderFile, per audio hour Streaming, per audio hour
AssemblyAI Universal-3.5 Pro
assemblyai.universal-3-5-pro
AssemblyAI$0.21 $0.21
AssemblyAI Universal 2
assemblyai.universal-2
AssemblyAI$0.37 $0.37
Microsoft MAI-Transcribe-2
microsoft.mai-transcribe-2
Azure$0.10 No streaming
Azure AI Speech-to-Text
azure.standard
Azure$0.18 No streaming
Cloudflare Whisper
cloudflare.whisper
Cloudflare$0.027 No streaming
Cloudflare Whisper Tiny (EN)
cloudflare.whisper-tiny-en
Cloudflare$0.027 No streaming
Cloudflare Whisper Large v3 Turbo
cloudflare.whisper-large-v3-turbo
Cloudflare$0.0306 No streaming
Deepgram Nova
deepgram.nova
Deepgram$0.258 $0.348
Deepgram Nova 2
deepgram.nova-2
Deepgram$0.258 $0.348
Deepgram Nova 3
deepgram.nova-3
Deepgram$0.312 $0.552
Deepgram Base
deepgram.base
Deepgram$0.75 $0.87
Deepgram Enhanced
deepgram.enhanced
Deepgram$0.87 $0.99
ElevenLabs Scribe
elevenlabs.scribe-v1
ElevenLabs$0.40 No streaming
Gemini 2.5 Flash-Lite
gemini.gemini-2.5-flash-lite
Gemini$0.0121 No streaming
Gemini 2.5 Flash
gemini.gemini-2.5-flash
Gemini$0.1222 No streaming
Gemini 2.5 Pro
gemini.gemini-2.5-pro
Gemini$0.261 No streaming
Gemini 3.5 Transcribe
gemini.gemini-3.5-transcribe
Gemini$0.306 No streaming
Gladia Solaria
gladia.standard
Gladia$0.612 No streaming
Groq Whisper Large v3 Turbo
groq.whisper-large-v3-turbo
Groq$0.04 No streaming
Groq Whisper Large v3
groq.whisper-large-v3
Groq$0.111 No streaming
IBM Watson Speech to Text
ibm.standard
IBM$1.20 No streaming
Mistral Voxtral Mini Transcribe
mistral.voxtral-mini
Mistral$0.06 No streaming
Mistral Voxtral Mini Transcribe v2
mistral.voxtral-mini-v2
Mistral$0.18 No streaming
OpenAI GPT-Transcribe
openai.gpt-transcribe
OpenAI$0.27 No streaming
OpenAI Whisper
openai.whisper-1
OpenAI$0.36 No streaming
Rev AI Enhanced
revai.fusion
Rev AI$0.30 No streaming
Rev AI Reverb ASR
revai.machine
Rev AI$0.30 No streaming
Smallest AI Pulse
smallestai.pulse
Smallest AI$0.18 No streaming
Smallest AI Pulse Pro
smallestai.pulse-pro
Smallest AI$0.24 No streaming
Soniox Async v5
soniox.stt-async-v5
Soniox$0.10 No streaming
Grok Speech to Text
xai.grok-stt
SpaceXAI$0.10 No streaming
Speechmatics Standard
speechmatics.standard
Speechmatics$0.24 No streaming
Speechmatics Enhanced
speechmatics.enhanced
Speechmatics$0.40 No streaming
NVIDIA Parakeet TDT 0.6B V3 Realtime
togetherai.nvidia-parakeet-tdt-0.6b-v3
Together AI$0.21 No streaming
Thinking Machines Inkling
togetherai.thinkingmachines-inkling
Together AI$0.612 No streaming

35 models from the Speechall catalog, last checked November 7, 2026. Rates are what Speechall charges per hour of audio, in USD. See the model catalog for every feature and the full list of variants.

Credits are deducted at these rates. The service fee is paid once, when you buy credits, and is not added to each request.

Estimate cost from a per-hour rate

The catalog’s hourly figure is the per-second rate multiplied by 3,600. To estimate a month of transcription, multiply audio hours by the hourly rate.

This example uses a placeholder rate, not a real Speechall or provider price. Look up the real rate for your model on /models.

InputExample value
Placeholder rate$0.36 per audio hour
Audio per month50 hours
Estimated usage50 × $0.36 = $18.00

To cover that month, you would buy roughly $18.00 of credits. The service fee is charged on top of the base amount at checkout, so check the quote on billing for the amount you actually pay. Allow extra for test runs and reprocessed files, since each request is usage. If a model’s rate differs for streaming, use the streaming rate for live audio.

When going direct to a provider can be better

A gateway is not always the cheapest or best choice. Going direct to a single provider may suit you if:

  • You process large, steady volumes with one provider and can get a volume discount or a negotiated contract.
  • You need an account-level feature, agreement or support arrangement that only the provider offers.
  • You have settled on one provider and do not expect to switch, so a single integration is simple enough.

Speechall fits better when you want to try or combine several providers with one balance and one API, or when your volume is too small for a negotiated rate to matter. Because credits include a service fee on top of the base amount, compare the all-in cost of both routes for your own workload. The gateway vs direct integration post covers the trade-offs in more detail.

Related pages