Invoice Parser API

The Invoice Parser API supports PDF, JPEG, JPG, and PNG file formats for invoice processing. You can upload these file types directly via multipart form data or provide a publicly accessible URL to the invoice document. The API automatically detects the input format and applies the appropriate processing pipeline, using OCR for image-based invoices and text extraction for native PDF documents. This flexibility means you can integrate the API into diverse workflows whether your invoices arrive as scanned images, digital PDFs, or photographed documents from mobile devices.

The API extracts over 40 distinct data fields organized into structured categories. Vendor details include company name, address, phone, email, website, tax ID, and registration number. Customer information captures name, billing and shipping addresses, attention lines, and customer IDs. Invoice metadata covers invoice number, purchase order, issue date, due date, payment terms, and currency. Each line item is parsed with description, product code, quantity, unit price, discount, tax rate, and total amount. The API also returns comprehensive totals including subtotal, discounts, tax breakdowns by type, shipping costs, and amounts paid versus outstanding.

Our API achieves an accuracy rate of approximately 96.7% across diverse invoice formats and layouts. The system leverages advanced AI vision models that have been trained on millions of invoice documents spanning different industries, countries, and design templates. Accuracy depends on document quality factors such as resolution, contrast, and clarity of text. For scanned documents, we recommend a minimum resolution of 300 DPI with good lighting and minimal skew. The AI handles complex layouts including multi-column invoices, tables with merged cells, and invoices with logos or watermarks that might interfere with traditional OCR approaches.

Yes, the maximum file size for uploads is 1MB per request, which accommodates the vast majority of invoice documents. Standard PDF invoices typically range from 50KB to 500KB, and compressed JPEG images of invoices usually fall well under this limit. For image files, we recommend optimizing resolution to 300 DPI at the actual document size, which provides excellent OCR accuracy while keeping file sizes manageable. If you regularly process larger files such as multi-page invoice bundles, consider splitting them into individual pages before submission or contact our support team to discuss enterprise solutions with higher file size limits.

The API automatically detects and processes currency formats from around the world, including USD, EUR, GBP, JPY, and dozens of other currencies. It recognizes various number formatting conventions such as comma-separated thousands with decimal points or period-separated thousands with decimal commas used in European countries. Tax structures are parsed intelligently, identifying VAT, GST, sales tax, and other regional tax types with their respective rates and amounts. The tax breakdown section of the response provides granular detail including taxable amounts, tax rates, and calculated tax for each applicable tax type, enabling accurate financial reconciliation across international invoices.

Yes, the Invoice Parser API supports invoices written in a wide range of languages thanks to its AI-powered vision model architecture. The system can accurately extract data from invoices in English, Spanish, French, German, Portuguese, Italian, Dutch, and many other languages without requiring you to specify the language in advance. The AI model automatically identifies the document language and applies appropriate parsing rules for date formats, number conventions, and field labels specific to that locale. This makes the API ideal for international businesses processing invoices from vendors and suppliers across different countries and language regions.

The Invoice Parser API is specifically designed to handle multi-page invoices, which are common for large orders with many line items or detailed terms and conditions. When you upload a multi-page PDF, the API processes all pages sequentially, using context from earlier pages to inform parsing of later pages. The AI model understands that line items often span multiple pages, with headers repeated on each page, and intelligently combines data across pages into a single coherent output. For example, if page 1 contains vendor information and the first 20 line items, and page 2 continues with 15 more line items plus the total, the API will extract all 35 line items and associate them with the same invoice. The invoice number, dates, and vendor details are extracted from wherever they appear (typically page 1), while totals are found on the final page. Table continuation across pages is handled automatically — the parser recognizes when a table header is repeated and continues aggregating rows. For very long invoices (10+ pages), processing time increases linearly with page count, but the API maintains accuracy throughout. The response indicates which page each piece of information was extracted from, which is valuable for auditing and verification purposes. Multi-page invoices are particularly common in industries like wholesale, construction, and professional services where itemized billing is detailed.