Botoi Readability API vs Readability Analysis Tool (RAT)

Botoi Readability API analyzes English text and returns Flesch-Kincaid Grade, Flesch Reading Ease, Gunning Fog, and underlying counts. A request sends a required text string to POST /v1/text/readability and receives JSON. Flesch Reading Ease scores run from 0 to 100, with higher values indicating easier reading; Gunning Fog estimates the years of formal education needed. The formulas and syllable heuristics are calibrated for English, so grade mapping is not meaningful for other languages. Botoi recommends sending at least 100 words because short samples can produce unstable scores. Suggested uses include enforcing plain-language standards, reviewing legal or medical drafts, refining blog posts for SEO, and comparing writing across authors. Examples are provided in cURL, JavaScript, TypeScript, Python, and Go. The Free plan is free and lists 1,000 requests per day and 10 per minute; the endpoint page separately states five requests per minute for its free tier. Starter costs $9 a month. Requests above a per-minute or daily cap receive HTTP 429, and daily limits reset at midnight UTC.

Readability Analysis Tool (RAT) checks German text for readability anomalies using configurable rules. It accepts .docx files and marks detected issues as comments in a separate copy with a -rat.docx suffix, leaving the original intact. RAT also creates an HTML report with statistics, readability formulas, anomalies, and a quality gate indicating whether text may be too trivial or difficult. Its documented formulas are Flesch and Wiener Sachtextformel. Report details can include reading and speaking time, sentence and word counts, average word and sentence measures, common parts of speech, and configured keyword percentages. Users can adjust anomaly rules and thresholds; the report records anomalies they reject as false positives or mark as incorporated. RAT runs from the command line using a Java JAR, with configuration and output-directory options. The project lists performance of 40 seconds for 10,000 words, or 50 pages. It runs on Linux and Windows, uses Java, UIMA, and DKPro Core, and is licensed under GPLv3.