OCR for financial statements
Pull income statements, balance sheets and cash-flow statements out of annual reports and scanned accounts into rows a credit analyst or auditor can model.
PDF, PNG, JPG, WebP, HEIC. Up to 10 files at onceNo file?
What makes financial statements hard to read
- Two or three years side by side in narrow columns
- Negative numbers in brackets and figures in thousands
- Notes references and subtotals that must not be double counted
Fields coming with document_type
{"company": "Harlow & Pike Manufacturing Ltd","statement": "balance_sheet","period_end": "2025-12-31","currency": "GBP","units": "thousands","line_items[]": "label, current_year, prior_year","total_assets": "48,215","total_liabilities": "29,760"}Best models for financial statements
1Reducto ParseHigh-accuracy parsing for complex financial and legal documents, with citations.20 credits / page
reducto/parse20 credits / page2Mistral OCRFast document OCR that returns clean Markdown with tables and embedded images.25 credits / pagemistral/mistral-ocr25 credits / page3Azure Document IntelligenceAzure layout model with tables, selection marks and prebuilt invoice and ID readers.60 credits / pageazure/document-intelligence60 credits / pageCall it from your code
curl https://api.ocrrouter.ai/v1/ocr \
-H "Authorization: Bearer $OCRROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "reducto/parse",
"file_url": "https://example.com/invoice.pdf",
"format": "json"
}'