Text this: No Train, No Pain? Assessing the Ability of LLMs for Text Classification with no Finetuning