We have a GPT that we have trained in the GPT store. We haven’t made it public yet as we are still fine-tuning. Is it possible to use the API to access it? I attempted to create a project-specific key, but didn’t see a way to connect it to a GPT model in the store. Any pointer to some docs would be appreciated. Thanks!
1 Like
Hi and welcome!
No there is no way to access GPT’s made in ChatGPT with the API.
You could use the Assistants API to recreate your GPT since it has similar functionality.
1 Like
Ah - very nice… so I train an assistant in the same way i trained the GPT, and then I can get a model ID that I can use to access it from the API? is that correct? If so, I see there is a way to create project keys. Can they be used to point to a specific assistant?
1 Like
I don’t think GPTs are technically trained but are given a system prompt and data according to your conversion with the GPT builder.
With the Assistant API you would write your assistants instructions and also provide files for its knowledge.
You would typically use the model id such as gpt-4o and you’ll need to create an API key from your project that developers can use to integrate your assistant into a front end.
Does anybody know a workaround for this? Since I am facing same issue, problem is that I also wanted to make the most of the gpt actions. So I would like to use the gpt as an intelligent chatbot capable of some actions towards my API