What systems are supported for deployment?

[Chinaniu]

"What operating systems and architectures are supported? I'm particularly interested in Linux (ARM64, amd64) and Windows environments."

[dosubot[bot]]

MinerU supports Linux, Windows, and macOS with the following specifics:

Linux:

• Both amd64 (x86_64) and ARM64 (AArch64) architectures are supported (Docker support)
• Requires distributions from 2019 or later
• Python 3.10-3.13 supported
• Uses vLLM as the inference framework

Windows:

• Fully supported
• Python 3.10-3.12 only (Python 3.13 not supported due to the ray dependency) (README)
• Uses LMDeploy as the inference framework

macOS:

• Requires version 14.0 or later
• Python 3.10-3.13 supported
• GPU acceleration available on Apple Silicon

For GPU acceleration across platforms, MinerU supports Volta and later architecture NVIDIA GPUs or Apple Silicon devices.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

   

[Chinaniu]

Does the open-source deployment have all these features, such as the ability to parse PDF files?

[dosubot[bot]]

Yes, the open-source version of MinerU has full PDF parsing capabilities and all features are available—there's no paid tier or feature restrictions. MinerU is entirely open-source under AGPL v3.

Key features available in the open-source deployment include:

• PDF to Markdown/JSON conversion with layout analysis and reading order detection
• Formula parsing with LaTeX output (inline and display)
• Table extraction with HTML output
• Image extraction with captions
• Code block and algorithm recognition
• OCR support for 109 languages (README)

You can also choose from three processing backends depending on your hardware:

• Pipeline: CPU-only, ~82% accuracy, no hallucinations
• VLM: GPU required, ~90% accuracy, Chinese/English
• Hybrid: GPU required, ~90% accuracy, multi-language

Input formats supported: PDF, PNG, JPEG, WebP, GIF, BMP, TIFF, and JP2.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}