Getting API error for long prompts but no errors with short prompts

API
1

I get below error when I use prompt with 3000 tokens with expectation of 500 tokes on completions. It works perfectly fine when i ask simple promots like"how are you". I am using DaVinci model. Earlier it was working perfectly fine for the same prompt.

RateLimitError: Rate limit reached for default-text-davinci-003 in organization org-xxxxxxxx on tokens per min. Limit: 150000.000000 / min. Current: 315000.000000 / min. Contact support@openai.com if you continue to have issues. Please add a payment method to your account to increase your rate limit. Visit OpenAI API to add a payment method.

2

Do you have other members in your org consuming API?

Also 3000 token for prompt is very long.Consider prompt engineering. Read: Best practices for prompt engineering with OpenAI API | OpenAI Help Center

3

Sometimes if you send repeated requests to the GPT call, it shows this error. Try and put some sleep or delay between your requests and see whether that helps

4

I am not sending repeated requests. if i send tokens with less than 20 lines then it works.

5

Hi @lalituor

Please post your entire JSON API completion response using / wrapped with Markdown triple backticks.

Thanks

:slight_smile: