Cheapest model at increasing word volume
Same average document length, only monthly word volume changes. Per-scan pricing is a flat tax per document, so it scales worse as volume climbs unless documents stay short.
| Words/month | Per-credit | Per-word | Per-scan | Cheapest |
|---|
Why the same word count produces three different bills
AI content and plagiarism detectors (GPTZero, Originality.ai, Copyleaks and similar tools) all scan text and return a verdict, but they don't charge for that scan the same way. Credit-based pricing, the shape GPTZero popularized, bills in fixed blocks — typically one credit per 100 words — so a 950-word document rounds up to 10 credits whether or not the last block is full. Per-word pricing, closer to Originality.ai's model, bills the exact word count with no rounding, which tends to be marginally cheaper at scale but offers less pricing transparency per individual scan. Per-scan pricing, closer to Copyleaks' model, charges a flat fee per document regardless of length — cheap for a batch of short social posts, expensive for a single long-form article that a per-word model would price proportionally.
The crossover between these three models is almost entirely a function of average document length. Flood the tool with short documents — tweets, product descriptions, comment moderation — and per-scan pricing's flat fee per document adds up fast relative to the tiny word count each one contains, while per-credit and per-word pricing barely notice. Feed it long documents — articles, theses, reports — and the opposite happens: per-scan pricing amortizes its flat fee over more words and starts winning, while per-credit and per-word costs climb linearly with length. There's no vendor whose sticker price is representative of your actual bill without knowing both your volume and your typical document length; run both through the table above before picking a vendor on price alone.