Implementing exponential backoff with jitter for OpenAI API

Proper retry logic is essential for production OpenAI API usage. Here's my battle-tested implementation:

import time
import random
from openai import OpenAI, RateLimitError, APITimeoutError
def call_with_retry(func, max_retries=5):
    for attempt in range(max_retries):
        try:
            return func()
        except RateLimitError:
            wait = min(2  attempt + random.uniform(0, 1), 60)
            print(f"Rate limited. Waiting {wait:.1f}s...")
            time.sleep(wait)
        except APITimeoutError:
            wait = min(2  attempt, 30)
            time.sleep(wait)
    raise Exception("Max retries exceeded")

Key points:

Always add jitter to prevent thundering herd

Cap the max wait time

Different strategies for rate limits vs timeouts

The official SDK has built-in retries — make sure you're not double-retrying!