i was making aiagent in n8n and i was using on-prem LLM mistral on my 32GB MAC mini.for a simple query like “hi” it is taking 20-30 sec to response .whereas when i am using the same LLM in terminal it responds in miliseconds . what could be the reasons and how to fix these things.
Hi @Rohit_llm, in my experience, using AI Agents is going to be much slower than a direct message chat with an LLM. Reason being, the AI Agents have longer reasoning times to make sure they use all tools available to them. Even if there are no tools attached and you only have it hooked up to the chat model, it may take longer to finish executing just due to the fact that its trying to figure out if it needs to do more before finishing. I was building an AI powered workflow recently and ended up switching from an Agent to a simple chat messaging model because of the added complexity
@NCione thankyou for your respose, But the same AiAgent is giving faster response when i am using OpenAi model gpt-4.1-mini as llm.I think their may be the issue with some configuration or with N8N.
i am making aiagent which will have knowledge about my company data and respond according to that data , and i want to use locally installed LLM model as llm for my agent.