Why do stories from ChatGPT score so low?
Ask a general assistant for a decodable story for a first grader and it writes something short and cheerful with simple sentences. Simple is not the same as decodable. It will happily reach for bright, sparkling, friends and beautiful, each of which uses a spelling pattern most first graders have not been taught yet. It has no idea which lesson your reader is on, so it cannot stay inside it. Paste one here and the red words show you exactly where it wandered off.
Does this use a different scorer from the Cody app?
No. This page sends your text to the same analyzer that scores every story Cody generates, with the same lesson rules and the same dictionary. One honest caveat: when Cody writes a story it also knows the teacher's topic words and can use a language model to spot names, so a story's own page can credit a word or two that a pasted copy cannot. Put the names in the names box and the two agree.
What does the number not tell you?
Decodability is necessary, not sufficient. A passage can be 100 percent decodable and still be dull, incoherent or wrong for a child's interests. Cody's own stories go through separate checks for coherence and content; this tool measures one thing, on purpose, so the comparison is fair.
My word is real but shows as not in the dictionary
The dictionary is large but not infinite, and it deliberately leaves out most proper nouns. Put names in the names box so they are credited. If an ordinary word is missing, it counts against the passage until we add it, which is the conservative choice.