Read between
the lines.
A second opinion for the age of synthetic prose. Paste a passage and let a research model look for patterns the eye might miss.
What are we reading?
Your signal appears here.
No text is stored. The passage is read once, in memory, then forgotten.
Reading between the lines…
Rhythm, predictability, structure, and learned textual traces.
—
Keep the result in perspective. This model can be confidently wrong. Never use one reading to accuse, grade, hire, moderate, or make a consequential decision about someone.
Built for curiosity.
Not certainty.
This reader uses mild-rgb’s ModernBERT-large detector, trained on long-form English ELI5 answers and responses from seven 2026 language models.
It performs well on aggregate research, but confidence on a single document is poorly calibrated. Human writing can look synthetic; AI writing can look human. Treat the score as a prompt to look closer—not proof of authorship.
- Best range
- 250–800 words
- Language
- English
- Model context
- 2,048 tokens used