Simply set api_base to any OpenAI compatible server:
api_base
interpreter --api_base <custom_endpoint>