How to make an API call to a custom GPT model?

+1 for the capability to call CustomGPT via API integration from a custom app.

If they want us to use assistants for this purpose they should make them the same features! They are not the same. What’s the point of CustomGPT if we cannot use outside the playground?

PLEASE FIX ASAP

3 Likes

i would recommend enhanceai’s api and stuff, i mean they r totally on a whole other level man!

Funny…here’s ChatGPT’s input.
"To connect to your custom GPT model created through GPT Builder using the OpenAI Chat API in Python, you can follow these steps:
…Use the openai.ChatCompletion.create() method to generate responses from your custom GPT model. Make sure to pass the model parameter with the ID of your custom GPT model. Here is an example code snippet…
:stuck_out_tongue_winking_eye:

It was virtually impossible to get the correct answer, which is “GPTs are only for within ChatGPT Plus” from ChatGPT itself, employing all variety of techniques short of making a “say no” GPT.

If browsing was explicitly “disabled”, ChatGPT would continue through example hypothetical code.

This seems to be an effect of static instructions possibly missing “knowledge cutoff” (besides the quality of inference on search results):

ChatGPT's system prompt Feb 21, variations

API, programmed: no problem denying

GPT-4
I’m sorry for any confusion, but as of my last training cut-off in April 2023 [ed: this was system prompt], OpenAI has not introduced a feature called GPTs within ChatGPT that allows users to create, interact with, or share a GPT. The information you’re asking about seems to be beyond my current knowledge…

GPT-4-Turbo
As of my last update in April 2023, OpenAI had not announced a feature specifically called “GPTs” within ChatGPT that matches the description you’ve provided. However, based on your description, it sounds like a hypothetical feature that allows users to create, interact with, and share custom instances or versions of GPT models. Given this context, I’ll provide a general approach on how you might interface with such a feature using a Python script, assuming it works similarly to interacting with an API. This approach will be speculative and based on general principles of interacting with web APIs using Python…

So if that browser tool ever makes it here to find the right answer:



Using ChatGPT GPTs (aka Gizmos, agents) outside of ChatGPT Plus


Question: Can I access or use a GPT I created via an API?
Answer: No, a GPT can only be used within, created by, and shared with current subscribers to ChatGPT Plus (or Teams, Enterprise).


Question: Can I put a GPT in my website for visitors?
Answer: No, a GPT can only be used within, created by, and shared with current subscribers to ChatGPT Plus.


Question: Can I link to a GPT so anybody can use it, like to ask about my services?
Answer: No, a GPT can only be used within, created by, and shared with current subscribers to ChatGPT Plus.


Question: How can I repurpose another user’s GPT, programming expertise, backend action resources that power it, documents, for my own profit?
Answer: Really?