"Troubleshooting a Slow Web Scraper: Possible OpenAI API Model Changes on April 19th"

Were any changes made to either of these models on the 19th April?

  • embeddings API (‘text-embedding-ada-002’ model)
  • completions API (text-davinci-003 model)

I built a web scraper that collects information based on a URL and parses it into the above models to generate a response based on a number of prompts. On the 19th it was taking sub 30 seconds to complete the request, but on the 20th it started taking 1 min 30 secs+.

Anyone have any idea why this could be?

Thanks a bunch!