OCRRouter

PaddleOCR-VL vs Unstructured

Pick PaddleOCR-VL if you want to self-host a parser for scanned, multilingual or formula-heavy documents and keep full control of the model. Pick Unstructured if you ingest many file types (PDF, HTML, Word, PowerPoint, email) for RAG and want a hosted API with 10,000 free pages. The main split is a single document-reading model versus a multi-format ingestion pipeline.

PaddleOCR-VLpaddle/paddleocr-vlUnstructuredunstructured/unstructured
TypeOpen sourceOCR API
Price6 credits / page20 credits / page
Per 1,000 pages$6.00$20.00
Typical speed1.3 s / page3.0 s / page
Best forCJK documentsMixed file types into RAG
InputsPDF, imagesPDF, images, DOCX
Languages109 languagesMultilingual
Max pages per request5001,000
Tables✓ Yes✓ Yes
HandwritingNoNo
Math✓ YesNo
Bounding boxes✓ Yes✓ Yes
Confidence✓ YesNo
FormsNoNo
Multilingual✓ Yes✓ Yes
On OCRRouterAvailableAvailable

Prices are what you pay on OCRRouter. Speed is the provider's typical figure, not a benchmark we ran.

Which to choose

Choose PaddleOCR-VL if

  • It reads text, tables, formulas and charts in 109 languages, including Cyrillic, Arabic, Devanagari and Thai, in one model.
  • At under 1B parameters it runs on much smaller GPUs than 7B document VLMs like olmOCR.
  • Version 1.6 scores 96.33 on OmniDocBench v1.6, the top figure its authors report for that benchmark.

Choose Unstructured if

  • It takes PDFs, HTML, Word, PowerPoint, emails and images through one library, which suits mixed-file RAG ingestion.
  • The hosted API gives you the first 10,000 pages free, then $0.015 per page, so you can start without running any GPU.
  • The Auto, Fast, High-Res and VLM strategies let you trade speed for quality, and Fast suits born-digital PDFs.

Key differences

PaddleOCR-VLUnstructured
How you run itSelf-host only, via PaddleOCR, vLLM or transformers. NVIDIA GPU recommended; CPU is possible but slow.Self-host the open-source library or call the hosted API.
LicenceApache-2.0Apache-2.0 for the open-source library; proprietary for the hosted platform.
PriceFree; you pay for compute.Hosted: first 10,000 pages free, then $0.015 per page. Business plan is custom. Open-source library is free.
Output formatsMarkdown, JSONJSON, HTML, plain text, bounding boxes
Input scopeDocument pages and images: text, tables, formulas and charts.PDFs, HTML, Word, PowerPoint, emails and images.
TablesRecognizes tables, but users report missing table content on version 1.6 (issue #18314).Vendor benchmark of table cell content: 0.426 open-source vs 0.866 hosted. Table OCR ignores language settings (issue #1694).
Known failureVersion 1.6 regresses Russian and Kazakh Cyrillic on born-digital PDFs, runs about 11x slower with use_cache false, and hangs on blurry images.hi_res tables can be cut off at the edges, dropping first letters. The library sends usage telemetry by default.

Sources

What each one is

PaddleOCR-VL

Open-source document model from Baidu. Excellent on Chinese, Japanese and Korean.

PaddleOCR-VL details

Unstructured

Document ETL for RAG. Partitions many file types into typed elements; open-source library too.

Unstructured details

Try both on your own file

Run the file once with each model and compare the output.

PDF, PNG, JPG, WebP, HEIC. Up to 20 MB each, 10 files.No file?

Use one, fall back to the other

{"model": "paddle/paddleocr-vl","fallbacks": ["unstructured/unstructured"],"file_url": "https://example.com/document.pdf"}

Questions

More comparisons

PaddleOCR-VL alternativesUnstructured alternativesBest OCR models in 2026