How to recognise when llm is from ChatGPT or from Huggingface?

[lfoppiano]

Is there a smart way to discriminate the llm object to be from huggingface or ollama or openai?

for example something smarter than

if type(llm) is HuggingFaceHub:
...
else
...

this is useful for example to choose whether to use a ChatPromptTemplate or to get the results by calling .content or not (when they are already string, for example from ollama / zephyr)

It seems that mixing open and close LLMs ends up mixing a lot of if -else for just some changes in the replied objects

Thank you in advance!