OpenAI API — September, 2026
Significant enhancements to API error reporting provide clearer distinctions between user-induced rate limiting and temporary server issues. Applications will now receive a 429 error with the 'slow_down' code for exceeding traffic limits, separate from a 503 error with 'server_is_overloaded' for model overload. This precision enables developers to implement more intelligent and efficient retry mechanisms, often guided by an explicit Retry-After header, improving application stability and user experience.
Source: official ChatGPT & Codex release notes. Summaries are written from the official notes; see how we work.
Changed
- Updated API errors so applications can distinguish traffic that increases too quickly from temporary model overload.
- Traffic that increases too quickly can return a 429 error with the slow_down code. Temporary model overload returns a 503 error with the server_is_overloaded code. Both responses may include Retry-After. When the header is present, wait at least as long as it specifies before retrying. If it's missing, use exponential backoff. See the error codes guide and rate limits guide.
Get notified the minute the next ChatGPT & Codex release ships
AI Drops sends a push with a plain-English summary of every release.
Get AI Drops on the App Store