Built to keep your document intact.
Most translators flatten a PDF into a wall of text. PDFPal rebuilds the translated file to mirror the original — page for page, element for element.
Every column, table and margin exactly where it was
PDFPal separates the visual layout from the text, then re-anchors the translation onto the original layout and rescales each paragraph to fit its box.
Equations are masked, never mangled
Before translating, PDFPal detects math, sub/superscripts and inline notation and masks each as a placeholder. The text translates around them, and the formulas drop back in verbatim.
Scanned & image PDFs
Scanned-PDF mode keeps image-heavy pages clean once a text layer exists. There is no built-in text recognition (OCR) yet, so run a scan through an OCR tool first.
Consistent terminology
Automatic term extraction keeps key vocabulary consistent across a 200-page document. Bring your own glossary too.
Free & premium engines
Start on the free Qwen2.5-7B or Gemini Flash-Lite, then switch to Gemini Flash, GPT or Claude per document for the highest quality.
Private by design
Processed on isolated workers and delivered through short-lived signed links. PDFPal never uses your documents to train models; the text to translate is sent to the provider that runs the engine you choose, which may include a model-routing service (see the Subprocessors in our Privacy Policy).