I cannot find custom gpt model # in my account

Subject: Assistance Required for Custom GPT Model API Access and Embedding

Dear OpenAI Support Team,

I am currently working with a custom GPT deployment named “OAD Guru,” accessible via the URL in the image.

Screenshot 2024-06-23 at 9.51.55 AM
Despite several attempts, I have been unable to successfully connect to the API using various model IDs and endpoints.

Here are the details of the issue:

  • I have an OpenAI Plus account.
  • The custom GPT models do not appear in my account’s model list.
  • I have tried using model IDs such as H7oq1QVki, babbage-002, and davinci-002.
  • I have used the following endpoints: https://api.openai.com/v1/completions and https://api.openai.com/v1/chat/completions.
  • Each attempt results in a 404 error, indicating that the model ID or endpoint might be incorrect.

Additionally, I am looking to embed this custom GPT chat bot into my Squarespace account. I have successfully embedded a generic bot, but I need the specific model number for the selected custom GPT to proceed.

Could you please provide the exact model ID and the correct API endpoint for accessing the custom GPT “OAD Guru”? Additionally, if there are any specific configurations or permissions required, please let me know.

Thank you for your assistance.

Hi, welcome to the developer forum.

Custom GPT’s can only be accessed from users of ChatGPT with a plus account. There is no API endpoint to access them externally.

You could look at the Assistants API which offers prebuilt solutions or you could create your own solution using traditional code and the API, but not for GPTs.

2 Likes

Hi Spencer. Thank you.

So, is there a way to embed a custom gpt to a page on my website?

I think the answer is no. I just wanted to be direct.

Thank you. It’s interesting that gpt 4.o does not seem to know this. I was very specific with the account situation, and custom gpt.

1 Like

This is exactly what you can do with the assistants API :laughing:

You can try building your own assistant over at the OpenAI playground by simply copy-pasting the instructions for your custom GPT into the appropriate fields :laughing:

3 Likes

I am a pretty non technical guy (for a guy who can build his own custom GPTs!)

I visited the assistant and got lost. Can you recommend an overview of assistants for me to read and or watch?

Thanks

1 Like

It can be a bit daunting if you’re not already familiar with code, but every journey starts with the first step!

Here’s the quickstart guide…

https://platform.openai.com/docs/quickstart

The OpenAI Cookbook is also useful…

There’s likely people who are offering easy-install wrappers to use Assistants on your website, but I’d be careful and do your due diligence as there’s quite a few scammers on the internet, unfortunately.

If you have any specific questions, though, feel free to stop by and let us know.

1 Like

Here’s where I really need help. I have two well working custom gpts. I literally use every day. For some reason, I don’t see them any where in mt open AI dashboard or account. So I can’t find the model numbers.

WW hat do you think?

  1. ChatGPT, a consumer chatbot end product, is in no way connected to the API. They use different billing, different resources.
    – GPTs are for end users of ChatGPT Plus, and with limited usage, now for free account users.
    – API is for developers to create their own API products, with billing by model usage.
    Your question is like “why don’t my GPTs show up in Google AI Studio” - it is just as different a product to ChatGPT as the OpenAI API.

  2. You or others cannot use a GPT except within the web or app Chatbot with subscription.

  3. You do not create or train a “model” when you make a GPT. All that a GPT is can be completely described by what’s in the instruction box, optionally with a search feature on files you’ve uploaded.
    – Regardless of what conversation you might have had with the “GPT Builder” entertaining you, a GPT is just a prompt made of instructions:
    Untitled-1

  4. You as programmer and developer must make your own implementation for the web if you want to pay for OpenAI services for others to chat with the AI you made. This can be with the chat completions endpoint and a model like gpt-3.5-turbo. You can maintain sessions of customers, their accounts, and chat history of users that have connected in a database. If you cannot already do this, then you have a high hurdle of first learning programming.

  5. Behavior of the AI comes about by system messages that tell the AI what its job is. This is similar to instructions to make a GPT, but you are in full control of the language and must treat it like another type of programming. The instructions are not placed in a container telling ChatGPT “here’s how a user wants you to act, which is called a GPT”.

So you can see, you will not find a “model” to use a GPT in the API.

2 Likes

Wow what a thoughtful response. I appreciate the time and attention.

Seems I was way off base. I could have sworn there were embed codes for the custom gpt. That is what I wanted to do in the first place.

I will mark this as closed. Thanks.