Credits are charged for successful searches according to the configured rate for the engine, search type, country and cache status. The public plan cards show the current account allowances; the API response and usage history show actual charges.
Start from the request you plan to send
List the engine, search type and country for each workflow. Those settings can change the credit rate. Use the pricing table and current credit-rate configuration for an estimate rather than assuming every request costs the same.
Separate successful searches from retries
Successful searches consume credits. Invalid parameters are rejected before the search runs; failed searches and upstream timeouts release reserved credits. A retry without an Idempotency-Key is a new request, so design retries deliberately.
Account for cache behavior and history
Cache hits are free under the default credit rates. Usage history records request and charge details so you can compare estimates with actual consumption. If your application must avoid repeating a write-like action, send an Idempotency-Key and follow the API documentation.
Know what happens at the balance limit
A request is rejected with INSUFFICIENT_CREDITS when the available balance cannot cover it. Monitor the balance in your application and choose what to do before it runs out. Current paid checkout availability is shown above on this page.