OCRRouter

RapidOCR vs Chandra

Pick RapidOCR for offline CPU OCR of Chinese and English screenshots and receipts, or when you need to embed OCR in a Java or C# app. Pick Chandra when you need Markdown, HTML or JSON with layout from multilingual scans, tables, forms or math, and you have a GPU or can use the hosted API. The main split is plain text and boxes on a CPU versus structured document output from a vision-language model.

RapidOCRrapidai/rapidocrChandradatalab/chandra
TypeOpen sourceOpen source
Price1 credit / page2 credits / page
Per 1,000 pages$1.00$2.00
Typical speed0.6 s / page2.0 s / page
Best forLightweight OCR on CPU or edgeComplex layouts and handwriting
InputsimagesPDF, images
LanguagesChinese, English and more40+ languages
Max pages per request100500
TablesNo✓ Yes
HandwritingNo✓ Yes
MathNo✓ Yes
Bounding boxes✓ Yes✓ Yes
Confidence✓ YesNo
FormsNo✓ Yes
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 RapidOCR if

  • It runs on CPU through ONNX Runtime with pip install rapidocr onnxruntime, and needs no GPU or Paddle framework.
  • It has bindings for Python, C++, Java and C#, so it fits desktop and server apps in many stacks.
  • It is Apache-2.0 and free to self-host with no revenue cap, though converted models keep their upstream licenses.

Choose Chandra if

  • It outputs Markdown, HTML and JSON plus layout chunks with bounding boxes and labels per block, which suits PDF to Markdown for RAG.
  • It scores 85.8 on olmOCR-bench (Chandra 2, authors' table) and 77.8% on a 43-language benchmark, so it covers far more languages than RapidOCR's default Chinese and English.
  • It handles documents with tables, forms and math, and you can self-host it or call the Datalab API from $1 per 1,000 pages.

Key differences

RapidOCRChandra
OutputPlain text and bounding boxesMarkdown, HTML, JSON and bounding boxes with layout labels
HardwareCPU by default via ONNX Runtime; optional GPU, TensorRT, OpenVINO or MNN enginesSingle NVIDIA GPU via vLLM; benchmarks run on one H100 80GB
LicenceApache-2.0; converted models keep their upstream licensesCode Apache-2.0; weights modified OpenRAIL-M, free under $2M funding or revenue, commercial license needed above that
PriceFree, self-hosted; you pay for computeSelf-hosted free within license terms; hosted API $1, $3 or $7.50 per 1,000 pages, with $20/month free tier
LanguagesChinese and English by default; other languages need extra models43-language average 77.8%, but 43.2% on Urdu and 51.1% on Sanskrit
BenchmarksNot publishedolmOCR-bench overall 85.8 (Chandra 2, by authors)
Known failure16-bit grayscale PNG or TIFF give no text or wrong text; Arabic crashes because python-bidi is undeclaredToken-loop hallucinations on some Russian and Kazakh scans; Old Scans category only 51.1 on olmOCR-bench

Sources

What each one is

RapidOCR

PaddleOCR models packaged for ONNX Runtime and other engines. Easy to deploy without Paddle.

RapidOCR details

Chandra

Datalab's OCR model that keeps layout and handles forms, tables and handwriting.

Chandra 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": "rapidai/rapidocr","fallbacks": ["datalab/chandra"],"file_url": "https://example.com/document.pdf"}

Questions

More comparisons

RapidOCR alternativesChandra alternativesBest OCR models in 2026