Skip to main content
fetchbean runs on prepaid credits. You top up a balance from the dashboard, and each API call is charged against it. There are no subscriptions, invoices, or surprise bills — when your balance runs out, calls are rejected rather than billed on credit. This model gives you precise, predictable control over your spending.

​
Pricing

Credits are priced at a flat rate:
  • 1 credit = 0.0001∗∗,so∗∗0.0001**, so **1 buys 10,000 credits.
  • Each tool has either a fixed credit cost per call or a metered cost based on response size.
  • Top up any amount from the Balance section of the dashboard. Payments are processed securely by Dodo Payments.
Browse per-tool credit costs for every provider and endpoint at fetchbean.com/catalog.

​
You only pay for successful calls

fetchbean uses a reserve-then-settle model to ensure you never pay for calls that don’t deliver results:
1

Reserve

When you make a call, fetchbean reserves an estimated maximum credit amount from your balance upfront.
2

Execute

fetchbean forwards your request to the upstream provider and waits for a response.
3

Settle

Once the call completes, fetchbean settles the actual cost — which is never more than the reserved maximum. Any unused reservation is immediately returned to your balance.
Provider failures and timeouts are billed at zero. If an upstream provider errors or doesn’t respond in time, you are not charged anything for that call.
Your spend can never exceed your balance. If your balance can’t cover the estimated maximum for a call, the call is rejected upfront rather than overdrawing you.

​
Metered tools

Some tools — such as read — are metered by response size rather than charged a flat fee per call. You pay for the data you actually receive, up to a fixed ceiling. The reserve-then-settle model still applies: fetchbean reserves the ceiling upfront, then settles the real usage when the response arrives.

Fixed-cost tools

Charged a flat credit amount per successful call, regardless of response size.

Metered tools

Charged based on the size of the response returned, with a maximum ceiling per call.