[pixl97]

LLMs use tokens. Tokens don't have to be text, hence multimodal AI. Fee free to call them different senses if you want.