SwatowAPI

English

简体中文

English

简体中文

Appearance

Billing ​

You pay per call. Each request is charged against the official pricing of the model you call, with no markup and no hidden fees. Failed and timed-out requests are not billed. Input and output tokens are metered per call, every call carries a trace id, and top-ups and spend reconcile in one ledger.

How billing works ​

  • Pay per call: you only pay for successful calls.
  • Charged against official pricing, no markup: cost tracks the model vendor's official price, with no markup in between.
  • Failed / timed-out requests are not billed: requests that error upstream or time out cost nothing.
  • Input and output tokens are metered separately: every call meters input and output token usage on its own.
  • Every call carries a trace id: used for reconciliation and troubleshooting in the console.
  • Top-ups and spend reconcile in one ledger: top-ups and spend settle in a single ledger, and channel status is public.

TIP

This page does not list per-token prices. For the exact price of each model, see the console pricing page, which also states the billing unit (typically per million tokens).

How charges are calculated ​

The cost of a single call is the input and output token usage for that call, each multiplied by the model's official unit price, then summed:

text
call cost = input tokens  x model input price
          + output tokens x model output price

Notes:

  • Input and output usually have different unit prices, so they are calculated separately.
  • Unit prices track the model vendor's official pricing; this gateway adds no markup. See the console pricing page for exact prices.
  • Tokens from failed / timed-out requests are not counted toward cost.

Check your usage ​

Endpoint: GET /v1/usage ​

Call GET /v1/usage to get your account's usage and spend summary. Like GET /v1/models, it is shared across protocol surfaces.

bash
curl https://swato.ai/v1/usage \
  -H "Authorization: Bearer $SWATOWAPI_KEY"

Domestic (汕拓智算) users swap the Base URL for the domestic console's own origin; the path structure is identical.

Console usage ledger ​

The usage ledger in the console shows per-call detail: each record carries the model, input / output tokens, cost, and a trace id. Top-ups and spend reconcile in the same ledger, so you can verify every charge.

Where to find prices ​

WARNING

Do not hardcode any fixed unit price into your application. For the exact price of each model, rely on the console pricing page; prices may change.

  • Exact unit price per model: the pricing page in the console.
  • Which models your key can actually call: rely on what GET /v1/models returns.
  • Limits — rate limits and quotas.
  • FAQ — common questions about top-ups, billing, and logs.