Problem Formulation
I encountered a significant delay in the response times on OpenAI’s GPT-3.5 Turbo API as compared to the ChatGPT — and I’m not alone with this (source). The latency is particularly noticeable when using certain accounts, with the delay making the API almost unusable for real-time applications.
Solution Idea: Experiment with Difference User Accounts to Prevent Throttling
When researching possible solutions to this problem, I discovered that a user experienced 30x performance differences across two different OpenAI API accounts’s playgrounds (source).
♥️ Info: Are you AI curious but you still have to create real impactful projects? Join our official AI builder club on Skool (only $5): SHIP! - One Project Per Month
Here’s what happens if you ask for a full dictionary of a certain word using two different accounts from the same user at the same time (left fast, right slow):
This suggests that the issue might be account-specific or related to allocating server resources per account. It is likely that OpenAI experiments with various throttling ideas to improve the efficiency of resource allocation — we all have experienced times when demand for OpenAI’s models was too much to handle.
Prompt Engineering with Python and OpenAI
You can check out the whole course on OpenAI Prompt Engineering using Python on the Finxter academy. We cover topics such as:
- Embeddings
- Semantic search
- Web scraping
- Query embeddings
- Movie recommendation
- Sentiment analysis
👨💻 Academy: Prompt Engineering with Python and OpenAI