Identify the resilience pattern

A downstream payments API your service depends on occasionally returns errors or times out under load. When that happens, your service keeps hammering it with the same requests, which slows the API down even more and turns a brief hiccup into a prolonged outage. You want a pattern that stops sending requests to the failing dependency for a short period, gives it time to recover, and then probes to see if it is healthy again.

Which resilience pattern best fits this situation?