OCR Image/PDF to Text API

When to use this API

OCR API is designed for document ingestion pipelines where raw files must be converted into usable text. It supports operational ingestion flows with request tracking, usage metering and controlled output handling. OCR quality depends on scan quality, language, resolution, layout and readability, so results should be reviewed before downstream automation.

• You need searchable text from scanned or image-based documents.
• You want OCR in a backend workflow without maintaining your own OCR stack.
• You process large document volumes and need async execution support.

Typical use cases

• Document ingestion for finance and operations.
• Search indexing of uploaded files.
• Pre-processing for downstream AI and extraction workflows.

Billing and APIRAPTOR token model

OCR usage is metered per processed OCR page through the `ocr_page` service billing event.

• Tokens are the APIRAPTOR communication unit: 1 token = 100 subtokens.
• Trial availability, token amount and duration come from Billing configuration.
• All services are available from day one; OCR is not premium-gated per plan.
• This is service billing event metering, not ordinary API-call billing.

OCR quality boundary

OCR results are provider-backed text extraction signals and artifacts, not a guarantee of perfect transcription.

• Quality depends on scan resolution, contrast, language, handwriting, layout complexity, skew/noise and source readability.
• Low-quality scans, dense tables, handwriting or mixed-language layouts can reduce confidence and completeness.
• Review OCR output before using it for irreversible operational decisions.

Output boundary

OCR returns text-oriented outputs and OCR artifacts; it is not a domain-specific field extraction engine by itself.

• CSV/XLSX outputs are table-aware OCR artifacts when tables are detected and configured.
• JSON output describes OCR results and artifacts, not guaranteed invoice, receipt or legal-document field extraction.
• Use specialized extraction services when you need domain-specific structured fields.

Sync and async usage

Sync and async requests use the same canonical OCR POST endpoint.

• Sync uses POST /v1/ocr with mode=sync for fast-path inputs.
• Async create uses POST /v1/ocr with mode=async.
• OCR status uses GET /v1/ocr/jobs/{jobId}; OCR result uses GET /v1/ocr/jobs/{jobId}/result.