ScanRead.ai

ScanRead.ai is a free OCR tool that extracts editable text from images, PDFs, screenshots, and handwritten documents. Supports 100+ languages with 22 specialized tools (handwriting recognition, math → LaTeX, receipt OCR, table detection, screenshots, Arabic/Japanese/Chinese OCR). What problem it solves: Generic OCR tools handle Latin text well but break on handwriting, CJK languages, math equations, and structured documents like tables. ScanRead is purpose-built for these edge cases. How we built it: Powered by PP-OCRv5 + PaddleOCR-VL (2M params). Benchmarks competitively against Google Vision and AWS Textract — especially in CJK languages. Stack: Next.js, PaddleOCR (PP-OCRv5 + PaddleOCR-VL), Cloudflare Workers, Postgres. Free tier: 20 pages/day, no signup needed. Pro from $10/mo for 3,000 pages, batch processing, watermark-free .txt/.md/.docx export.