token_reel

https://github.com/stringsn88keys/token_reel

token_reel renders a "CLI demo"-style animated GIF of a prompt being answered: the prompt appears, there's a configurable pause (time to first token), and the response streams in word-by-word or character-by-character at a configurable tokens/sec rate -- handy for READMEs, blog posts, and talks that want to show off an LLM CLI without shelling out to a real model (or a real screen recorder).