Yup! We rely on the AI SDK for model routing, and they have an Ollama provider, which will handle pretty much any local model.