Skip to main content
Automatically retry failed LLM requests up to 3 times with exponential backoff. After all retries are exhausted, the request is denied. This handles transient errors like rate limits and temporary provider issues.

Policy JSON

What It Does

  • Retries failed requests up to 3 times on transient errors (429, 500, 502, 503, 504)
  • Exponential backoff — waits 1s, 2s, 4s between retries (capped at 5s)
  • Denies the request after all retries are exhausted
  • Does not retry on client errors (400, 401, 403) — those indicate configuration issues, not transient failures

Example

Request fails with a 429 (rate limit) on first attempt: Decision object (after 3 retries exhausted):

Install

Customize

Set action_on_exhausted to FALLBACK and combine with a Fallback Chain to automatically route to an alternative provider after retries are exhausted.