PDFMathTranslate

What is PDFMathTranslate?

PDFMathTranslate (pdf2zh) translates PDFs page by page and writes both a translated and a side-by-side bilingual PDF, preserving formulas, charts, the table of contents and annotations. It uses a layout-detection model and can call Google, Bing, DeepL, OpenAI, Azure OpenAI, Gemini, Ollama or any OpenAI-compatible model as the translator. It runs as a web UI in Docker, a command-line tool or a Python API.

Where it fits in a 10–200 person business

• Translate supplier datasheets, manuals or contracts into your team's language without retyping the layout
• Produce bilingual PDFs of technical papers or tenders for review
• Keep translation fully in-house by pointing it at a local Ollama model

These use cases are our editorial suggestions, not claims made by the project.

Can a business use PDFMathTranslate? (license)

This is a plain-English summary of the license text, not legal advice. Third-party models you run with it have their own licenses.

Data & compliance: does PDFMathTranslate phone home?

Checked against the project's own docs and repository on Oct 22, 2026. "Not documented" means we could not find the answer in official sources; it does not mean the feature is missing or that no data is sent. Third-party services you connect (model APIs, vector databases, search engines) have their own data policies. Compare all apps.

What hardware does PDFMathTranslate need?

The project does not publish a minimum CPU/RAM figure, so we leave the official value blank rather than guess.

No official RAM or CPU figures. Python installs need Python 3.11 or 3.12.

Source: github.com/PDFMathTranslate/PDFMathTranslate#install

How much does it cost to host PDFMathTranslate per month?

How to deploy PDFMathTranslate

Method: Docker

docker pull byaidu/pdf2zh
docker run -d -p 7860:7860 byaidu/pdf2zh
# open http://localhost:7860/  (mirror: ghcr.io/byaidu/pdfmathtranslate)

From the official instructions: github.com/PDFMathTranslate/PDFMathTranslate#install. Always check the source for the current version before you run it.

Known gotchas when self-hosting PDFMathTranslate

• The default translation service is Google, so document text is sent to Google unless you choose another service. For in-house translation select Ollama or an OpenAI-compatible endpoint you host. source
• The app downloads its layout model (DocLayout-YOLO ONNX) from Hugging Face on first use; where Hugging Face is blocked, set HF_ENDPOINT to a mirror. source
• The --ultrafast mode skips layout detection and OCR, so pages without a native text layer (scans) are left untranslated. source

Alternatives in Speech & translation

• whisper.cpp Fast C/C++ port of OpenAI Whisper speech recognition with a small HTTP transcription server.
• LibreTranslate Self-hosted machine translation API with no dependency on Google or DeepL.
• Speaches OpenAI-compatible speech-to-text and text-to-speech server (faster-whisper, Kokoro, Piper).

Sources

• GitHub repository: github.com/PDFMathTranslate/PDFMathTranslate
• Official documentation, Hardware requirements, Deploy instructions, Gotcha 2: github.com/PDFMathTranslate/PDFMathTranslate#install
• License text: github.com/PDFMathTranslate/PDFMathTranslate/blob/main/LICENSE
• Gotcha 1: github.com/PDFMathTranslate/PDFMathTranslate/blob/main/docs/ADVANCED.md#services
• Gotcha 3: github.com/PDFMathTranslate/PDFMathTranslate#readme
• Hetzner Cloud pricing (cost estimate): www.hetzner.com/cloud/regular-performance
• GitHub stars, commits and releases: GitHub REST API, fetched Oct 22, 2026