I changed my code to allow for parallel API requests, and now I get this all the time.
If I do them one by one, it works, so it’s not a budget thing because I still have plenty of money left in the budget.
OpenAI API returned TooManyRequests. Please try again
You also have rate limits on the OpenAI API, which is based on a “tier” system of account age plus amount paid.
Free testing use, which some may have when creating new accounts and using nano models, is extremely limited, at 3 requests-per-minute.
After purchasing prepaid credits, that RPM will increase, but per-model token pools are still quite low per model, to where you can have a single call fail merely on a large input.
Review the entry documentation and your own organization’s limits in the platform site to understand.
You will likely want a better diagnosis - the API doesn’t return that particular message, so likely something else in your software stack or the service you are employing to make API calls is re-writing the 429 error or its http body.
2 Likes