Appending to the answer message the time needed to generate it
Just append inference time and LLM used to the generated message