Custom GPTs Programmatic Access

[stoufa]

Is it possible to programmatically use Custom GPTs with this library?

[kalyanamdewri]

Not directly as a Custom GPT object or GPT ID through openai-python.

Custom GPTs are configured and used inside ChatGPT. The API path is separate: build the same behavior in your app with the OpenAI API by carrying over the GPT configuration into API primitives:

• GPT instructions -> instructions / developer message in your API call
• GPT knowledge files -> file search / vector stores
• GPT actions -> function calling or your own API/tool layer
• GPT web capability -> the API web search tool when that is appropriate
• GPT code/data analysis behavior -> the API code interpreter tool when you need hosted Python execution

The relevant distinction in the OpenAI docs is that GPTs are no-code assistants built and used inside ChatGPT, while API assistants/integrations are developer-built experiences inside your own app. The GPTs FAQ also says GPTs are not the way to embed ChatGPT in an external product; for that, use the API.

Docs:

• https://help.openai.com/en/articles/8554407-gpts-in-chatgpt
• https://developers.openai.com/api/docs/guides/tools

So if you already have a Custom GPT and want programmatic access, the practical route is to recreate its instructions, files, and tool behavior in an API-backed assistant/agent rather than trying to call the ChatGPT GPT directly.

[Hosseinghorbani0]

A useful follow-up question here is whether there is any supported way to preserve the behavior of an existing Custom GPT without manually rebuilding its configuration.

For example, if a Custom GPT already has:

• a specific instruction set,
• uploaded knowledge files,
• configured Actions,
• and a particular tool configuration,

is there an API-supported mechanism to reference or export those settings into an API application, or are these configurations intentionally isolated from the API?

I understand that the API and ChatGPT Custom GPTs are separate surfaces. I'm mainly wondering whether there is an official migration/export path for an existing Custom GPT, rather than having to recreate its behavior manually.