AI API error 429: rate limits, credits and spend caps
A 429 is a request limit response, not proof of a provider outage. Read the provider error code, message and retry headers before retrying.
Quick answer
A 429 is a request limit response, not proof of a provider outage. Read the provider error code, message and retry headers before retrying.
What each provider documents
OpenAI distinguishes request rate limits, exhausted credits and organization or project limits. Inspect error.code; retrying a billing or usage limit does not restore access.
Claude uses rate_limit_error for rate limits and some spend caps. A usage-tier spend-cap response has no retry-after header. Fast traffic growth can also trigger rate limits.
Gemini describes 429 RESOURCE_EXHAUSTED among errors that can justify exponential backoff. Inspect the error details before treating a limit as transient.
Recommended application response
Reduce request concurrency for a transient rate limit, honor a supplied retry delay, and set a maximum attempt count and total deadline. Account for retries already performed by your SDK. A missing retry header does not identify the cause on its own.
Preserve a sanitized error body and request ID. Check your account configuration privately when the error reports a credit or spend limit; do not publish credentials. Compare the relevant API status scope with your request evidence.
Related service pages
Each entry shows its own collection coverage. A linked service is not necessarily automatically collected.
Editorial review date applies to this guide. Live report collection times appear on the service pages.