Reasons for Slow AI Responses Using OpenAI API | Clay

Reasons for Slow AI Responses Using OpenAI API

So just for testing I have the most simple AI prompt, yet it takes MULTIPLE minutes. So question is: why are these AI columns so slow? I use my own API token of OpenAI for the 4o model. We’ve been using the API for many months so I don’t think we’re rate limited for just doing a few testing prompts to get random words. Any ideas? 😄 Thanks!!

5 comments

Channeled
APP
·
·
The slow performance of AI columns using your own OpenAI API token for the 4o model might be due to your OpenAI API usage tier. For reliable use of OpenAI in Clay, you need 30,000 TPM (tokens per minute) for ChatGPT columns. If you're on a lower tier, it can cause slow performance. To resolve this: 1. Check your API usage tier at https://platform.openai.com/settings/organization/limits 2. Ensure your plan meets the required TPM 3. If needed, upgrade your OpenAI API key to a higher tier You can find more information about usage tiers here: https://platform.openai.com/docs/guides/rate-limits/usage-tiers If you need further assistance, please request human help by clicking on "Talk to a person".
Rutger B.
·
·
No, my limits are way higher than that if I am correct:
Rutger B.
·
·
uh I switched to the Clay API key, ran another row and that was very fast. Then switched back to my own key, ran another row and now that’s fast too 😄
Bo (.
·
·
Hey, Totally get how frustrating that delay can be. While we don’t control any API limits on your end, there have been some recent performance slowdowns that might explain the hiccups you experienced. It’s also worth noting that running multiple AI rows—even with a simple prompt—can still take a bit of time, depending on model load and traffic. That said, switching between API keys (like you did) sometimes helps reset thing as it'll stop the previous rows Let me know if it slows down again and we’ll dig deeper.
Channeled
APP
·
·