Translate PDF
Translate a PDF into another language entirely on your device with Chrome's built-in AI — nothing is uploaded. Get a new PDF plus a .txt.
Machine translation normally means sending your document to somebody's server. This tool does not: it uses the translation model built into Chrome itself, running on your own machine. The text is pulled out of the PDF page by page, the source language is detected for you, and the translation is typeset into a new PDF alongside a plain .txt of the translated text.
That design has a hard requirement attached, and it is fair to know it before you start: the browser must actually provide those on-device models. In practice that means a recent desktop Chrome with the built-in AI features available, and the first use of a language pair downloads a model, with a progress bar while it does. On a browser without them the tool will tell you the pair is not available rather than quietly falling back to an upload.
How it works
- Drop in a PDF with selectable text.
- Let the source language be detected, and choose the target language.
- Wait for the language pack if it is the first time you use that pair.
- Download the translated PDF, and the .txt if you want the raw text.
What people use it for
- Reading a contract or an official letter in a language you do not speak, without sending it to a translation service.
- Translating internal documentation that policy does not allow you to upload anywhere.
- Getting a rough working translation of a paper while offline.
Frequently asked questions
- Is my file uploaded to a server?
- No. The file is opened by JavaScript and WebAssembly running inside your own browser tab. It is never uploaded, there is no server-side processing and nothing is stored — closing the tab discards everything.
- It says the language pair is not available.
- Either your browser does not provide the on-device translation models — they are a recent Chrome desktop feature — or that particular pair has no model to download. There is no server fallback by design: translating elsewhere would mean uploading your document, which is the thing this tool exists to avoid.
- Does the translated PDF look like the original?
- No. The text is re-typeset into a clean new document, not painted back into the original layout. Translated text also changes length — German runs long, and a line that fitted a column often will not — so keeping the original design would push text out of its boxes. Use the .txt if you only need the words.
- How good is the translation?
- Good enough to understand a document, and it runs privately, which is the trade. It is a compact on-device model, not a large cloud one. For anything legally binding, have a human translate it.
- Can I translate a scanned PDF?
- Not directly — there is no text to translate. Run OCR first to add a text layer, then translate the result.
Related tools
- 043043PDF to WordGet an editable .docx — flowing paragraphs with headings, plus the images carried over.
- 044044Word to PDFConvert .docx documents to a typeset PDF — headings, lists, tables and images. Entirely in your browser.
- 045045Excel to PDFTurn .xlsx, .xls, .ods or .csv sheets into a clean tabular PDF — auto orientation, repeated headers, wide tables split by columns. Formulas use saved values.
- 046046PDF to ExcelExtract tables and text from a PDF into an .xlsx workbook — one sheet per page, European number formats recognized. Scanned PDFs need OCR first.