Make sure your AI output actually reads at the right level
Language models frequently default to long, formal, multi-syllable prose that scores several grades above what a general audience comfortably reads. This tool runs your LLM output through four established readability formulas — Flesch Reading Ease, Flesch-Kincaid Grade Level, Gunning Fog, and SMOG — and tells you whether it matches your target audience, then flags the sentences dragging the score up.
How it works
Paste the text and the tool tokenizes it into sentences and words, estimates syllables with a vowel-group heuristic, and counts complex words (three or more syllables). From those counts it computes all four indices. Flesch Reading Ease returns a 0-100 score where higher is easier; the other three return a US grade level. You pick a target grade, and the tool shows how far each index is from it. Finally it ranks individual sentences by length and difficulty so you know exactly which ones to simplify.
Tips and example
For consumer-facing copy, aim for a Flesch Reading Ease of 60+ and a grade level around 7-9. If your model output lands at grade 13, the fastest fix is to add a constraint to the prompt — “write at a 7th-grade reading level, short sentences, plain words” — and re-score. Watch the flagged sentences: a single 40-word sentence stuffed with jargon can pull the whole document up two grades, so breaking it into two or three plain sentences often does most of the work. The syllable estimate is heuristic but the indices are robust across a paragraph.