OCRRouter

GOT-OCR 2.0 vs Chunkr

Pick GOT-OCR 2.0 if you run your own GPU and need formulas, sheet music or chemical structures, or OCR on a marked region of a page. Pick Chunkr if you need PDFs, slides or Word files turned into HTML, Markdown or JSON chunks for RAG, with a hosted API as an option. The main split is scope: GOT-OCR 2.0 is a 580M-parameter model that reads content, while Chunkr is a parsing service that does layout analysis, OCR and chunking.

GOT-OCR 2.0stepfun/got-ocr-2Chunkrchunkr/chunkr
TypeOpen sourceOCR API
Price1 credit / page14 credits / page
Per 1,000 pages$1.00$14.00
Typical speed1.5 s / page3.0 s / page
Best forFormatted text, formulas and chartsLayout-aware chunks for RAG
InputsimagesPDF, images, DOCX
LanguagesChinese, EnglishMultilingual
Max pages per request1001,000
Tables✓ Yes✓ Yes
HandwritingNoNo
Math✓ Yes✓ Yes
Bounding boxes✓ Yes✓ 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 GOT-OCR 2.0 if

  • You need LaTeX, TikZ, SMILES or kern output, so formulas, chemical structures and sheet music come out in a usable format.
  • You want OCR on one region of a page, selected by box coordinates or color.
  • You want a plain model with Apache-2.0 code that loads in Hugging Face Transformers, with batched inference and multi-page OCR, and you pay only for compute.

Choose Chunkr if

  • You are building RAG over PDFs, PPT, Word or image files and need semantic chunks with bounding boxes.
  • You want the choice of a hosted cloud API or a Docker self-host with GPU, CPU-only and Mac ARM options.
  • You already have an OpenAI-compatible LLM endpoint and want layout parsing that outputs HTML, Markdown or JSON.

Key differences

GOT-OCR 2.0Chunkr
What it is580M-parameter OCR model for text, formulas, tables, charts and sheet musicDocument parsing service with layout analysis, OCR with bounding boxes and semantic chunking
How you run itSelf-hosted only, via Hugging Face Transformers; CUDA 11.8 and PyTorch 2.0.1 per the README, and a llama.cpp port for CPUCloud API or self-host via Docker; the self-host version needs an external LLM endpoint
LicenceCode Apache-2.0; training data CC BY-NC 4.0AGPL-3.0 or a commercial licence; closed-source products need the commercial licence
PriceFree, you pay for computeHosted credit-based plans, prices not published; open-source version is free, you pay for compute and an LLM endpoint
Output formatsPlain text, Markdown, LaTeX, TikZ, SMILES, kernHTML, Markdown, JSON, bounding boxes
Input typesImages; region OCR and multi-page OCR with dynamic resolutionPDF, PPT, Word and images; no Excel in the open-source version
Known failureReported loops of repeated characters mid-recognition, wrong text on blank images, and worse results in format mode than in plain modeOpen-source version uses community models that are less accurate than the cloud's in-house models
BenchmarksNot publishedNot published

Sources

What each one is

GOT-OCR 2.0

Unified OCR model that reads plain text, tables, formulas, sheet music and charts.

GOT-OCR 2.0 details

Chunkr

Document parsing API that segments and chunks pages for RAG. Self-hostable version available.

Chunkr 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": "stepfun/got-ocr-2","fallbacks": ["chunkr/chunkr"],"file_url": "https://example.com/document.pdf"}

Questions

More comparisons

GOT-OCR 2.0 alternativesChunkr alternativesBest OCR models in 2026