Simply set api_base to any OpenAI compatible server:

interpreter --api_base <custom_endpoint>