OCRRouter

Marker vs GLM-OCR

Pick Marker if you want to self-host PDF to Markdown for books and papers with equations, or need RAG chunks and HTML or LaTeX output, and your company is under $5M or you will buy a licence. Pick GLM-OCR for high-volume, low-cost conversion under a plain MIT licence, or for pulling fields into a JSON schema. Check GLM-OCR carefully on financial tables, since it can drop table rows and lose decimal commas.

Markerdatalab/markerGLM-OCRzai/glm-ocr
TypeOCR APIOpen source
Price14 credits / page6 credits / page
Per 1,000 pages$14.00$6.00
Typical speed2.5 s / page1.3 s / page
Best forBooks and research papersLow-cost general documents
InputsPDF, images, DOCXPDF, images
Languages90+ languages100+ languages
Max pages per request1,000500
Tables✓ Yes✓ Yes
HandwritingNoNo
Math✓ Yes✓ Yes
Bounding boxes✓ Yes✓ Yes
ConfidenceNo✓ Yes
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 Marker if

  • Its model weights are free for research, personal use and startups under $5M, and the hosted API includes about 5,000 free pages per month.
  • It outputs Markdown, HTML, JSON block trees, RAG chunks, LaTeX and images, while GLM-OCR gives Markdown, JSON and plain text.
  • Its fast mode runs on CPU with about 18 MB of ONNX weights at 23.6 pages per second, and it also runs on Apple Silicon.

Choose GLM-OCR if

  • Its weights are MIT licensed, so there is no revenue cap on commercial use.
  • The Z.ai API costs $0.03 per million tokens for input and output, and self-hosting is free apart from compute.
  • It can extract fields into a user-defined JSON schema, which suits forms.

Key differences

MarkerGLM-OCR
LicenceCode Apache-2.0; weights under a modified AI Pubs Open Rail-M licence, free for research, personal use and startups under $5MWeights MIT; SDK repo and bundled PP-DocLayoutV3 Apache-2.0
How you run itLocal on GPU, CPU or Apple Silicon (MPS), or the hosted Datalab API; vllm mode needs an NVIDIA GPUZ.ai API, or self-host with vLLM, SGLang, Ollama or transformers; not supported on aarch64 Linux
PriceFree self-hosted within the weight licence; hosted API has about 5,000 free pages per month, then paid plans$0.03 per million tokens via API; free self-hosted
Output formatsMarkdown, HTML, JSON, chunks, LaTeX, imagesMarkdown, JSON, plain text, plus schema-based field extraction
BenchmarksolmOCR-bench 83.7 in vllm mode and 66.6 in CPU fast mode; OmniParseBench 85.9 in vllm mode; all author-reportedOmniDocBench v1.5 94.62 and olmOCR-Bench 75.2 (excluding headers and footers), both author-reported; ParseBench mean 29.6, independent
API limitsNot published100 pages, 50 MB per PDF and 10 MB per image per call
Known failureCPU fast mode scores lower on olmOCR-bench (66.6 versus 83.7 in vllm mode)Repeats OCR output unless a stop sequence is set, drops table footer and summary rows, and loses comma decimal separators in cropped table regions

Sources

What each one is

Marker

Hosted version of the open-source Marker and Surya models. Strong on books and papers.

Marker details

GLM-OCR

Small open-source OCR model from Z.ai with layout regions. MIT licensed.

GLM-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": "datalab/marker","fallbacks": ["zai/glm-ocr"],"file_url": "https://example.com/document.pdf"}

Questions

More comparisons

Marker alternativesGLM-OCR alternativesBest OCR models in 2026