One credit covers a document up to 5 pages. No overage invoices, no surprise metering — just a quota you can plan against.
If you process documents longer than 50 pages, need volume beyond Growth, or want schemas built and tuned against your own files, we price it directly. Send us a few representative documents and we'll come back with a quote and an accuracy benchmark.
Typical reply within one business day.
Documents are billed whole, not by the page. Longer files simply use more credits.
Every plan accepts documents up to 50 pages. Need longer documents?
| Free | Starter | Growth | Enterprise | |
|---|---|---|---|---|
| Extraction credits | 25/mo | 500/mo | 3,000/mo | Custom |
| Pages per document | 50 | 50 | 50 | Negotiated |
| Schemas | 2 | 10 | 50 | Unlimited |
| White-label sites | 1 | 5 | 25 | Unlimited |
| REST API + upload-site webhooks | Yes | Yes | Yes | Yes |
| Custom confidence thresholds | Yes | Yes | Yes | Yes |
| AI schema inference | Yes | Yes | Yes | Yes |
| Custom-built schemas | — | — | — | Yes |
| Support | Priority | Dedicated | Contract-specific |
One credit covers one document of up to 5 pages — which is most invoices, W-2s, receipts, dec pages, and applications. Longer documents use one additional credit per 5 pages, so an 8-page bank statement uses 2 credits and a 22-page loan package uses 5. API responses report the charge in meta.credits_used.
Extraction cost scales with page count, and a hard ceiling is how we keep pricing predictable instead of metering you by the page. 50 pages covers the overwhelming majority of business documents. If your workflow routinely involves longer files — deposition transcripts, full loan files, policy booklets — that is an Enterprise conversation, and we will price it against your actual documents.
Extractions return a clear quota error rather than silently billing you overage. Your quota resets at the start of each billing period, and you can upgrade at any time from your dashboard — the new quota applies immediately.
Send a supported document and schema slug to the shared extraction endpoint. Dokyumi handles document understanding, field extraction, and validation, then returns consistent JSON with confidence metadata. The underlying system can improve without changing your integration.
Yes. Upload 1–3 sample documents and our AI detects the fields to extract, or define your schema manually with the builder. Enterprise plans include schemas we build and tune for you.
Documents are processed in real time and stored in encrypted Cloudflare R2 storage. OCR results are cached by file hash for performance. Self-service deletion is not currently available; contact hello@dokyumi.com to request deletion and confirm the applicable scope and timing.
Yes. Upgrade, downgrade, or switch billing intervals at any time. Annual plans save 17% and are billed once per year.
Not sure which plan fits your document volume?
Ask us