Turn PDFs into structured data at scale. Powered by frontier open-weight OCR models with a type-safe TypeScript SDK.

• Best-in-class OCR - PaddleOCR-VL-0.9B for accurate text extraction
• Structured extraction - Define schemas, get JSON back
• Built for scale - Queue-based processing for thousands of documents
• Type-safe SDK - Full TypeScript support with React hooks
• Real-time updates - WebSocket notifications for job progress
• Self-hostable - Run on your own infrastructure

npm install ocrbase

import { createClient } from "ocrbase";

const { parse, extract } = createClient({
  baseUrl: "https://your-instance.com",
  apiKey: "ak_xxx",
});

// Parse document to markdown
const job = await parse({ file: document });
console.log(job.markdownResult);

// Extract structured data
const job = await extract({
  file: invoice,
  hints: "invoice number, date, total, line items",
});
console.log(job.jsonResult);

See SDK documentation for React hooks and advanced usage.

See Self-Hosting Guide for deployment instructions.

Requirements: Docker, Bun

MIT - See LICENSE for details.

For API access, on-premise deployment, or questions: adammajcher20@gmail.com