Fast scan vs Deep scan. Pick deliberately.
Same scanner, two very different reads. Fast answers instantly and is built for quick looks and bulk. Deep runs every check we have and is the one to trust when the answer matters. Both are free.
Deep for anything important. Fast for speed and volume.
If a result could touch a grade, a job, a dispute or a decision about a person — use Deep. If you need an instant first impression, or you are working through a stack of documents — Fast is what it is for, and its score tells you which ones deserve a Deep look.
| Fast scan | Deep scan | |
|---|---|---|
| What runs | The core pattern checks — sentence rhythm, phrasing, predictability, typography — combined by a trained scoring profile. | Everything. The full pattern set plus our trained deep-analysis model, run across every section of the document, with the strictest calibration we have. |
| Speed | Instant. Milliseconds of computing — which is also why it is unlimited. | A few seconds. Roughly a thousand times more computing per document. |
| Accuracy | A deliberate trade: tuned first to avoid flagging human writing, at the cost of catching less AI. A clean fast result is not a guarantee. | Substantially more accurate in both directions — catches far more AI and is far less likely to flag a person by mistake. |
| Built for | Quick looks. First passes. Checking a batch of entries fast. Deciding what deserves a closer read. | Grades, submissions, disputes, editorial calls, anything you might act on or show to someone else. |
| How to read it | As a screening pass: "clear", "worth a deep check", or "leaning AI". The middle answer is the honest one surprisingly often. | As our best read of the document — score, per-sentence marking, and the evidence behind every signal that fired. |
| Price | Free, no account, unlimited. | Free with a free account — the account is the abuse guard, not a paywall. |
Fast is tuned to be safely wrong, not confidently wrong.
Every detector makes mistakes. What you can choose is the direction. Our fast tier is tuned so that when it errs, it errs by saying "worth a deep check" — not by branding a person's writing as AI, and not by handing out clean bills of health it cannot back. That costs it raw catch-rate, and we would rather say so here than let a quick check be mistaken for a verdict. The Deep scan exists precisely so that being careful costs you nothing but a few seconds.