The detectors, one by one
Accuracy figures in this industry are almost all vendor-reported and almost never reproducible. We label which claims come from the vendor and which limits are documented.
Students have no way to check their own work before submission — the biggest source of anxiety in this niche
Score moves noticeably on lightly edited text
No free tier makes casual verification impossible
The headline accuracy figure is vendor-reported and not independently reproduced
Gives a coarse percentage with little explanation
Same company sells the paraphraser that is used to defeat detectors
Free tier is capped
No published methodology
Subscription required beyond the trial
Smaller product surface
Built as a side feature of a writing-assistant product
API-first — not aimed at individuals
Image only
Only detects content from Google models. Silence means nothing about Midjourney, DALL·E or Stable Diffusion
Stripped by most social platforms on upload — absence proves nothing