Amazon Textract vs Docling
Pick Amazon Textract for English forms, invoices and large scanned PDF batches on AWS, where a managed API with per-page pricing and word-level confidence scores saves you running anything. Pick Docling for RAG ingestion of born-digital PDFs and Office files, where you want Markdown or lossless JSON from a self-hosted MIT-licensed library. The main split is a hosted, paid API with narrow language support against a free local library whose scan accuracy depends on the OCR engine you plug in.
Amazon Textractaws/textract | Doclingibm/docling | |
|---|---|---|
| Type | OCR API | Open source |
| Price | 25 credits / page | 2 credits / page |
| Per 1,000 pages | $25.00 | $2.00 |
| Typical speed | 1.2 s / page | 2.2 s / page |
| Best for | Forms and key-value extraction | Structured export for RAG |
| Inputs | PDF, images | PDF, images, DOCX |
| Languages | 6 languages | Latin scripts, CJK |
| Max pages per request | 3,000 | 500 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | ✓ Yes | No |
| Math | No | ✓ Yes |
| Bounding boxes | ✓ Yes | ✓ Yes |
| Confidence | ✓ Yes | No |
| Forms | ✓ Yes | No |
| Multilingual | No | No |
| On OCRRouter | Available | Available |
Prices are what you pay on OCRRouter. Speed is the provider's typical figure, not a benchmark we ran.
Which to choose
Choose Amazon Textract if
- You need English forms, invoices or tables from scanned documents, with Forms at $0.05/page and Tables at $0.015/page in US West (Oregon) for the first 1M pages a month.
- You process large scanned PDF batches: the async API takes files up to 500 MB and 3,000 pages.
- You want word and line geometry with confidence scores, and you already run on AWS.
Choose Docling if
- You are building RAG ingestion for born-digital PDFs and Office files and want Markdown, HTML, JSON or DocTags out of one API.
- You need to keep layout, reading order and table structure, which the lossless JSON export preserves for downstream tools.
- You want a free MIT-licensed library that runs locally on CPU or GPU, with pluggable OCR backends (EasyOCR, Tesseract, RapidOCR) and LangChain or LlamaIndex integrations.
Key differences
| Amazon Textract | Docling | |
|---|---|---|
| How you run it | Hosted AWS API | Python library, runs locally on macOS, Linux and Windows (x86_64 and arm64) |
| Licence | Proprietary | MIT (individual models have their own licences) |
| Price | Text detection $0.0015/page, Tables $0.015/page, Forms $0.05/page, Queries $0.015/page (first 1M pages/month, US West Oregon) | Free, self-hosted; you pay for compute |
| Output formats | JSON, plain text, bounding boxes | Markdown, HTML, JSON, DocTags, plain text |
| Input formats and limits | Async API handles PDFs up to 500 MB and 3,000 pages; sync API takes one page | PDF, DOCX, PPTX, XLSX, HTML, EPUB, images, emails and XBRL |
| Languages | Text detection: English, French, German, Italian, Portuguese and Spanish. Handwriting and Queries: English only. Vertical Japanese and Chinese not supported. | Depends on the OCR engine you plug in |
| Benchmark | RD-TableBench table accuracy 80.9% (independent) | Not published |
| Known failure | No Markdown output for RAG without post-processing | OCR regression in 2.143.1 drops words and numbers with full-page Tesseract OCR; reading-order rule can merge neighbouring columns on some layouts |
Sources
- Textract pricing
- Textract quotas
- Reducto RD-TableBench
- Docling GitHub README
- Supported formats
- Issue #5254 OCR regression
What each one is
Amazon Textract
AWS OCR with form key-value pairs, tables and per-word confidence scores.
Amazon Textract detailsTry both on your own file
Run the file once with each model and compare the output.
Use one, fall back to the other
{"model": "aws/textract","fallbacks": ["ibm/docling"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Amazon Textract alternativesDocling alternativesBest OCR models in 2026