OCRRouter

PaddleOCR (PP-OCR) vs Typhoon OCR

Pick PaddleOCR (PP-OCR) for fast text-line OCR on Chinese and English printed text, receipts, signs and screenshots, on CPU, edge devices or in the browser. Pick Typhoon OCR for Thai and Thai-English documents such as government forms, financial reports and books, where you need Markdown, HTML tables or LaTeX out. The main split is output: PP-OCR returns lines and boxes, Typhoon returns structured documents.

PaddleOCR (PP-OCR)paddle/paddleocrTyphoon OCRtyphoon/typhoon-ocr
TypeOpen sourceOpen source
Price1 credit / page2 credits / page
Per 1,000 pages$1.00$2.00
Typical speed0.7 s / page1.8 s / page
Best forFast printed text, Chinese and EnglishThai documents
InputsPDF, imagesPDF, images
Languages80+ languagesThai, English
Max pages per request500500
TablesNo✓ Yes
HandwritingNoNo
MathNoNo
Bounding boxes✓ YesNo
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 (PP-OCR) if

  • It is a 34.5M-parameter PP-OCRv6 medium model that runs on CPU, NVIDIA GPU and other accelerators, with a reported 5.2x CPU speedup using OpenVINO.
  • One PP-OCRv6 model covers 50 languages, including Chinese, English, Japanese and 46 Latin-script languages.
  • It runs in the browser through the PaddleOCR.js SDK, and in Python or C++ for edge and mobile use.

Choose Typhoon OCR if

  • It reads Thai and Thai-English documents, and V1.5 beat GPT-5 and Gemini 2.5 Pro on Thai government forms (BLEU 0.870).
  • It outputs Markdown, HTML tables with merged cells, LaTeX equations and marked checkboxes, so one pass gives a structured document.
  • It scores 84.7 on olmOCR-bench and 90.97 on OmniDocBench v1.6 at 2B parameters, and is also available on the Typhoon API as typhoon-ocr-v2.

Key differences

PaddleOCR (PP-OCR)Typhoon OCR
OutputPlain text, bounding boxes and JSON. Tables and formulas need the separate PP-Structure or VL pipelines.Markdown, HTML and LaTeX, including tables with merged cells.
Languages50 languages in one PP-OCRv6 model, including Chinese, English and Japanese.Thai and English only. No other language was evaluated.
How you run itLibrary only: paddleocr Python package, C++ or the PaddleOCR.js browser SDK.Self-hosted weights or the Typhoon API (typhoon-ocr-v2).
Licence and priceApache-2.0. Free, self-hosted; you pay for compute.Apache-2.0, subject to OpenTyphoon Terms and Conditions. Free self-hosted; API pricing not published.
Model sizePP-OCRv6 medium: 34.5M parameters.2B parameters (V2 and V1.5), plus a 0.8B V2 variant.
BenchmarksNot published. The authors report PP-OCRv6 medium beats PP-OCRv5_server by 4.6% on detection and 5.1% on recognition.ThaiOCRBench 59.9, olmOCR-bench 84.7, OmniDocBench v1.6 90.97 (V2 2B, by the authors).
Known failure modesNotImplementedError on PaddleOCR 3.7.0 with PaddlePaddle 3.3.1 (issue #18162). PP-OCRv6 weights contain denormals that make CPU inference 4 to 11x slower (RapidOCR issue #743).V2 can loop until the token limit on dense multi-column pages. Key-information extraction scores only 21.0 on ThaiOCRBench. It can output text not in the image.

Sources

What each one is

PaddleOCR (PP-OCR)

Baidu's classic open-source text detection and recognition pipeline. Light enough to run on CPU.

PaddleOCR (PP-OCR) details

Typhoon OCR

Open vision model tuned for Thai and English documents, with Markdown output.

Typhoon OCR 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","fallbacks": ["typhoon/typhoon-ocr"],"file_url": "https://example.com/document.pdf"}

Questions

More comparisons

PaddleOCR (PP-OCR) alternativesTyphoon OCR alternativesBest OCR models in 2026