OCRRouter

Claude vs Hunyuan OCR

Pick Claude when the job is reading long PDFs, answering questions with page citations, or extracting JSON that needs reasoning, and you accept the per-page token cost. Pick Hunyuan OCR when you need cheap self-hosted OCR with bounding boxes, scene text, translation or rare scripts, and you are outside the EU, UK and South Korea. The main split is a hosted token-priced API versus a free ~1B model you run yourself.

Claudeanthropic/claudeHunyuan OCRtencent/hunyuan-ocr
TypeVision modelOpen source
Price25 credits / page2 credits / page
Per 1,000 pages$25.00$2.00
Typical speed2.6 s / page1.5 s / page
Best forLong dense documentsChinese documents and photos
InputsPDF, imagesPDF, images
Languages50+ languages100+ languages
Max pages per request100500
Tables✓ Yes✓ Yes
Handwriting✓ YesNo
Math✓ Yes✓ Yes
Bounding boxesNo✓ Yes
ConfidenceNoNo
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 Claude if

  • You need native PDF input with page-level citations, up to 600 pages per request on 1M-context models (100 otherwise) and a 32 MB payload.
  • You want extraction to JSON or Markdown with reasoning, or document question answering over long reports.
  • You want no GPU setup: each page is read as both text and a rendered image, so charts and scans are visible to the model.

Choose Hunyuan OCR if

  • You run high-volume OCR and want to pay only for compute, not 1,500 to 3,000 tokens per page.
  • You need bounding boxes, HTML or LaTeX output, or tasks such as text spotting, image translation, and low-resource or ancient-script OCR.
  • You want a ~1B model that runs on a laptop through llama.cpp GGUF, or on a CUDA 13 GPU with vLLM.

Key differences

ClaudeHunyuan OCR
How you run itHosted APISelf-hosted via vLLM, Transformers or llama.cpp
LicenceProprietaryTencent Hunyuan Community License; excludes EU, UK and South Korea; separate license needed above 100M monthly active users
PriceToken-based, no extra PDF fee; about 1,500 to 3,000 tokens per PDF page plus image tokensFree; you pay for compute
Output formatsMarkdown, JSON, plain textMarkdown, HTML, LaTeX, JSON, plain text, bounding boxes
Input limitsUp to 600 pages per request on 1M-context models (100 otherwise), 32 MB payloadImages up to 4K and a 128K context window in v1.5
OmniDocBench / ParseBenchNo published OmniDocBench score in the leaderboards checkedParseBench-2 overall 67.11 and ParseBench mean 63.86 (v1.5)
Known limitsNo password-protected or encrypted PDFs; no bounding boxesNeeds CUDA 13 and vLLM 0.25.1+ for the unified environment; the archived 1.0 note said Transformers and vLLM outputs differed in accuracy

Sources

What each one is

Claude

General vision model with careful reading of long, dense documents.

Claude details

Hunyuan OCR

Small end-to-end OCR model from Tencent covering parsing, extraction and photo translation.

Hunyuan 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": "anthropic/claude","fallbacks": ["tencent/hunyuan-ocr"],"file_url": "https://example.com/document.pdf"}

Questions

More comparisons

Claude alternativesHunyuan OCR alternativesBest OCR models in 2026