Start free
Upgrade to scale

Pay as you go

Free

1,000 free pages, then pay as you go.

Everything included:

1,000 free pages to get started
OCR for scans and faxes, tables, checkboxes, handwriting OCR, and source bounding boxes
Schema extraction with grounded, verified citations
Automatic schema generation from the document
Playground and self-serve dashboard
Unlimited team seats

Afterwards, $3 parse / $12 extract per 1,000 pages.

Enterprise

Custom

Production processing at scale.

Everything in Pay as you go, plus:

Parsing
Compare Plans	Pay as you go Get started for free	Enterprise Talk to us
OCR on scans, faxes, and rotated pages	Included	Included
Full table structure, with row and column spans	Included	Included
Checkbox and selection state	Included	Included
Handwriting OCR	Included	Included
Figures and images	Included	Included
Bounding boxes on everything	Included	Included
PDF, DOCX, and PPTX input	Included	Included
Extraction
Structured schema extraction	Included	Included
Automatic schema generation from the document	Included	Included
Hallucination guard: values that fail grounding are nulled and flagged, not guessed	Included	Included
Verbatim quote verification: every citation is a real substring of the source	Included	Included
Extraction citations	Included	Included
Bounding boxes on extracted values	Included	Included
Tooling
Playground: benchmark on your own documents	Included	Included
Self-serve dashboard and API keys	Included	Included
Usage visibility and auto-recharge	Included	Included
Unlimited team seats	Included	Included
Commercial terms
Rate	$3 / $12 per 1,000 pages	Volume rates
Rate limits	Standard	Raised on request
Data retention	Standard	Custom terms
BAA	Fast-path, talk to us	Included
Support	Email	Priority

Common questions

If something isn't here, email hello@hanji.dev.

One page of the source document. Plain text, dense tables, scanned forms, OCR, handwriting, and bounding boxes are included in the same page-based rate. Parse is $3 per 1,000 pages. Extract is $12 per 1,000 pages, with the required parsing included.
Extract includes parsing plus additional passes to return structured fields with grounding for each value. The price is fixed per page, so it does not change based on document type, OCR needs, tables, or configuration.
Gemini is model infrastructure priced by tokens. Hanji is a document-processing pipeline priced by pages. It includes OCR fallbacks, table structure, retries, bounding boxes, and grounded structured output. You can review our benchmarks for accuracy and hallucination rates on real documents.
Yes. For Enterprise customers, we can benchmark Hanji on your real documents and downstream success metrics, then tune the workflow around the fields and edge cases that matter most.
Yes. If you need HIPAA support or a BAA to evaluate real documents, contact us and we can prioritize the process.
Yes. Higher-volume customers can receive negotiated volume pricing based on the same dollars-per-1,000-pages unit. Contact us at hello@hanji.dev.
Each operation will have its own page-based rate. The billing unit stays the same: processed pages, not customer-facing credits.

Bring us the docs your pipeline struggles with. We’ll work through the edge cases and tune Hanji for your workflow.