Using GPT-4o for llm

[JFFD10]

Can someone tell me how to switch to GPT-4o with my API, the standard LLM of Vanna?
And for training the vector database, is it possible to use another embeddings model?
I'm not saying to use ChromaDB.

[AsaelCerros]

You change the model property in your config from the base class. eg: vn = MyVanna(config={'api_key': 'sk-...', 'model': 'gpt-4-...'})

[JFFD10]

Yeah it works thanks, but if i want to do embeddings with a better model in the vector database of Vanna its possible?

And for ChromaDB it's possible to change it too?