I see you are using Ollama from langchain.llms, you might need to try LLM instead from crewAI or completion from LiteLLM.