My chat app is due for an AI review tomorrow, but it has been unusable for about three hours because GPT 5.6 repeatedly returns a 429 error with the message "rate_limit_exceeded" and "temporarily unavailable." I have barely used the API in the past day, and the app has no customers yet, so I don't think I'm hitting a usage throttle. I also can't find any public outage reports. Has anyone else seen this, and is there a reliable place to check?
2 Answers
You’re probably not the only one. I’ve seen elevated and inconsistent error rates across several provider models, with one endpoint sometimes working while another fails. These incidents can be intermittent and may not appear on the public status page right away, so checking the provider’s status dashboard and testing a second endpoint or backup model is worthwhile.
A 429 paired with “temporarily unavailable” can indicate provider-side capacity problems rather than your account exceeding its quota. Check the limits for that exact model, since newer models may have different limits, and add exponential-backoff retries plus a fallback model before the review. That way, a brief outage produces a short delay instead of making the whole app appear broken.
It’s a chat app, and it has been down for roughly three hours. I’ve had access to this model for about three months without any problems, so this only started today.