Paste any English text — a benefits notice, a privacy policy, a homework page — and a model trained on 111,000 human difficulty judgments estimates the share of the US adult population and the world's adults who could read and functionally understand it. Everything runs in your browser: the model weights, the psycholinguistic lexicons, and the PIAAC literacy survey tables were loaded with this page. No text leaves your device.
Standard
Fetching the trained model + lexicons (~2 MB compressed, one-time, cached after that)…
Load a sample above to see it work — the bureaucratic benefits notice vs. its plain-language rewrite is the one to try: same content, a ~100-million-reader difference.