docTR vs Unstructured
Pick docTR when you need word-level OCR with coordinates that you run yourself in Python, for example on invoices and receipts that feed your own extraction. Pick Unstructured when you ingest mixed file types such as PDFs, Word, PowerPoint, HTML and emails for RAG. The main split is scope: docTR returns words and lines only, while Unstructured partitions whole files into typed elements.
docTRmindee/doctr | Unstructuredunstructured/unstructured | |
|---|---|---|
| Type | Open source | OCR API |
| Price | 1 credit / page | 20 credits / page |
| Per 1,000 pages | $1.00 | $20.00 |
| Typical speed | 0.9 s / page | 3.0 s / page |
| Best for | Word-level OCR with positions | Mixed file types into RAG |
| Inputs | PDF, images | PDF, images, DOCX |
| Languages | Mainly Latin scripts | Multilingual |
| Max pages per request | 500 | 1,000 |
| Tables | No | ✓ Yes |
| Handwriting | No | No |
| Math | No | No |
| Bounding boxes | ✓ Yes | ✓ Yes |
| Confidence | ✓ Yes | No |
| Forms | No | No |
| Multilingual | No | ✓ Yes |
| 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 docTR if
- You want a nested page, block, line and word structure with coordinates as JSON, from a library that runs locally on PyTorch.
- You plan to train or swap your own models, with a choice of detection and recognition architectures and straight or rotated boxes.
- You want one Apache-2.0 licence for everything, with ready Docker images, a Hugging Face Spaces demo and a key information extraction predictor.
Choose Unstructured if
- You need PDFs, HTML, Word, PowerPoint, emails and images handled in one library, with many connectors.
- You want a hosted API with the first 10,000 pages free, then $0.015 per page.
- You want to choose a strategy (Auto, Fast, High-Res or VLM) to trade speed for quality, such as Fast on born-digital PDFs.
Key differences
| docTR | Unstructured | |
|---|---|---|
| Licence | Apache-2.0 | Apache-2.0 for the open-source library; proprietary for the hosted platform |
| How you run it | Library in Python on PyTorch, plus Docker images. Self-hosted only. | Self-host the open-source library or call the hosted API |
| Price | Free, self-hosted (you pay for compute) | Hosted: first 10,000 pages free, then $0.015 per page. Business is custom. Open-source library is free. |
| Output formats | JSON, plain text, bounding boxes | JSON, HTML, plain text, bounding boxes |
| Tables | No table, formula or Markdown output; words and lines only | Table output exists, but the vendor's own benchmark shows table cell content at 0.426 for the open-source library versus 0.866 hosted |
| Hardware | CPU or NVIDIA GPU; GPU Docker images need CUDA 12.2 or newer on the host | Not published |
| Known failures | Split-crop merging misjoins readings across repeated characters (#2229); cropping images gives poorer results (#2007); ONNX export of MASTER truncates decoding steps (#2224) | hi_res tables cut off at edges and drop first letters (#3262); table OCR ignores language settings (#1694); extra Tesseract pass on tables adds time (#2580); library sends usage telemetry by default |
Sources
- docTR GitHub README
- docTR documentation
- Unstructured pricing
- Unstructured GitHub README
- Unstructured issue #3262
- docTR issue #2229
What each one is
docTR
Open-source OCR library from Mindee with separate text detection and recognition models.
docTR detailsUnstructured
Document ETL for RAG. Partitions many file types into typed elements; open-source library too.
Unstructured 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": "mindee/doctr","fallbacks": ["unstructured/unstructured"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
docTR alternativesUnstructured alternativesBest OCR models in 2026