Speed and Conversational Large Language Models: Not All Is About Tokens per Second | IEEE Journals & Magazine | IEEE Xplore