Production AI applications must handle errors gracefully. OpenAI has several error types you need to handle.

Common Error Types

• RateLimitError: Too many requests — retry with exponential backoff
• APIConnectionError: Network issue — retry
• APIStatusError: 4xx/5xx HTTP errors
• AuthenticationError: Invalid API key
• BadRequestError: Invalid request (content policy, too many tokens)

Pattern

from openai import OpenAI, RateLimitError, APIConnectionError

try:
    response = client.chat.completions.create(...)
except RateLimitError:
    # retry logic
except APIConnectionError:
    # retry logic
except Exception as e:
    # log and fail gracefully

Reference:

Error codes documentation

https://en.wikipedia.org/wiki/Special:Search?search=Error%20Handling