# Pricing — Spoken (spoken.md)

> Machine-readable pricing for AI agents and procurement tools.
> Human-readable version: https://spoken.md/#pricing
> Last updated: 2026-05-26

## What you're buying

A credit. One credit = one successful transcript fetch.
- Errors (404, 502) are never charged.
- Repeat fetches of the same episode are free (already cached against your key).
- Credits never expire.
- No subscription, no overage, no annual commitment.

## New-customer packs

Endpoint: `POST https://spoken.md/buy[?pack=100|500|2000]` (default pack=100)

| Pack | Price (USD) | Per transcript | Discount |
|------|-------------|----------------|----------|
| 100  | $15         | $0.15          | —        |
| 500  | $50         | $0.10          | 33% off  |
| 2000 | $160        | $0.08          | 47% off  |

## Top-up packs (existing keys)

Endpoint: `POST https://spoken.md/top-up?key={your_api_key}[&pack=100|500|2000]`

| Pack | Price (USD) | Per transcript |
|------|-------------|----------------|
| 100  | $10         | $0.10          |
| 500  | $40         | $0.08          |
| 2000 | $120        | $0.06          |

## Free trial

Demo API key: `pt_demo`
- No signup, no payment.
- Works on a single demo episode.
- Returns a full transcript so the format and quality can be evaluated before purchase.

## Currency, tax, and billing

- Currency: USD.
- VAT/sales tax: not currently collected (Dutch BTW registration pending).
- Payment processor: Stripe Checkout. Cards accepted in regions Stripe supports.
- Invoices: emailed automatically after each purchase by Stripe.

## API-side billing signals

The API returns billing-relevant headers on every successful transcript fetch:
- `X-Credits-Remaining: {n}` — credits left on the key.
- `X-Credits-Charged: {0|1}` — whether this fetch was charged (0 = repeat fetch or demo).

When credits hit zero, the next request returns HTTP 402 with a JSON body containing `top_up_url`. Agents can POST to that URL to redirect the user to Stripe Checkout for a top-up.

## Comparison to speech-to-text APIs

For published podcasts, Spoken is typically 5–10x cheaper than transcribing raw audio with Deepgram, AssemblyAI, or Whisper — once audio handling, diarization, and speaker-naming work are included. Spoken is not a substitute for speech-to-text on your own audio (meetings, calls, recordings).

See: https://spoken.md/why-spoken