Data & Retrieval
OCR & document AI
Extract text, tables and fields from PDFs and scans.
What it is
Document AI reads PDFs, scans and photos and returns text, tables, layout and structured fields (invoice totals, IDs), increasingly as clean Markdown ready for LLMs.
How pricing works
Billed per page; table/form extraction or premium models cost more per page.
We compare everything in one unit: USD per 1K pages.
Typical uses
- Invoice and receipt processing
- Ingesting PDFs into RAG
- Form digitisation
What to compare
- Table accuracy
- Handwriting support
- Markdown/JSON output
- Languages
Watch out for
- Multi-column layouts and tables are where cheap OCR fails; test on your hardest documents.
All 7 APIs, cheapest first
Free to browseTick 2 to 4 APIs to compare them side by side.
| API | $ / 1K pages | Table extraction | Handwriting | Markdown output | Quality | Price data | Compare |
|---|---|---|---|---|---|---|---|
| Google Document AI (Enterprise OCR) Google (Gemini / Vertex) | $1.50 | No | Yes | No | – | Unverified sample | |
| LlamaParse LlamaParse | $3.00 | Yes | No | Yes | 84 | Unverified sample | |
| Mistral OCR 4.1 Mistral OCR | $4.00 | Yes | Yes | Yes | 86 | Verified Oct 4, 2026 | |
| Reducto Parse Reducto | $10.00 | Yes | — | Yes | – | Verified Oct 6, 2026 | |
| Azure Document Intelligence (Layout) Microsoft Azure AI | $10.00 | Yes | Yes | Yes | – | Unverified sample | |
| Unstructured (pay-as-you-go) Unstructured | $15.00 | Yes | — | Yes | – | Verified Oct 6, 2026 | |
| Textract (tables + forms) Amazon Textract | $65.00 | Yes | Yes | No | 82 | Unverified sample |