OCRRouter

olmOCR vs Unstructured

Pick olmOCR when you need clean Markdown from PDFs at scale and have an NVIDIA GPU with 12 GB or more, or can call a hosted endpoint. Pick Unstructured when your input is a mix of file types such as Word, PowerPoint, HTML and email, or you want a hosted API with 10,000 free pages. The main split is focus: olmOCR is a single PDF-to-Markdown model, while Unstructured is an ingestion pipeline that outputs typed elements.

olmOCRallenai/olmocrUnstructuredunstructured/unstructured
TypeOpen sourceOCR API
Price2 credits / page20 credits / page
Per 1,000 pages$2.00$20.00
Typical speed1.5 s / page3.0 s / page
Best forBulk PDF to text for datasetsMixed file types into RAG
InputsPDF, imagesPDF, images, DOCX
LanguagesMainly EnglishMultilingual
Max pages per request1,0001,000
Tables✓ Yes✓ Yes
Handwriting✓ YesNo
Math✓ YesNo
Bounding boxesNo✓ Yes
ConfidenceNoNo
FormsNoNo
MultilingualNo✓ 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 olmOCR if

  • You are converting large volumes of PDFs to Markdown for LLM training or RAG, and Ai2 puts self-hosted cost under $200 per million pages.
  • Your documents include equations, tables or handwriting, and olmOCR scores 82.4 on olmOCR-Bench, above Marker (76.1), MinerU 2.5.4 (75.2) and Mistral OCR (72.0).
  • You want open weights, training data and code, all released.

Choose Unstructured if

  • You ingest PDFs, HTML, Word, PowerPoint, emails and images through one library with many connectors.
  • You want to try a hosted API first: the first 10,000 pages are free, then $0.015 per page.
  • You need bounding boxes in the output, or want to choose between the Auto, Fast, High-Res and VLM strategies to trade speed for quality.

Key differences

olmOCRUnstructured
LicenceApache-2.0 (toolkit)Apache-2.0 (open-source library); proprietary (hosted platform)
How you run itSelf-host with vLLM on an NVIDIA GPU with at least 12 GB VRAM and 30 GB free disk, or use third-party hosted endpointsSelf-host the Python library or use the hosted API
PriceSelf-hosted: under $200 per million pages (Ai2). DeepInfra hosted: $0.09 input / $0.19 output per million tokensFirst 10,000 hosted pages free, then $0.015 per page. Open-source library is free. Business is custom
Output formatsMarkdown, JSONJSON, HTML, plain text, bounding boxes
Input typesPDFs and imagesPDFs, HTML, Word, PowerPoint, emails and images
Benchmark82.4 on olmOCR-Bench (v0.4.0)Table cell content: 0.426 open-source vs 0.866 hosted (vendor's own benchmark)
Known failuresOut of memory on a 12 GB GPU (issue #424), stuck on the last few pages (issue #433), unreliable image bounding boxes (issue #437)hi_res tables cut off at edges and drop first letters (issue #3262), table OCR ignores language settings (issue #1694), sends usage telemetry by default

Sources

What each one is

olmOCR

Open-source vision model from Ai2 (often searched as olm ocr) that turns PDFs into clean, reading-order text. Latest release: olmOCR 2.

olmOCR 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": "allenai/olmocr","fallbacks": ["unstructured/unstructured"],"file_url": "https://example.com/document.pdf"}

Questions

More comparisons

olmOCR alternativesUnstructured alternativesBest OCR models in 2026