
Doc2X
Translate PDFs with AI and keep the original layout.
About Doc2X
Doc2X is an AI PDF translator built around one job: giving you a translated version of a PDF that still looks like the original. You upload a document, choose a target language and a translation model, and Doc2X parses the file, translates it and returns a layout-preserved translated PDF in which, according to the homepage, fonts, colors, tables and images stay where they belong. The problem it addresses is familiar to anyone who has pasted a research paper, manual or contract into a general translation tool: the text comes back, but headings, columns, tables, formulas and figures are lost, and the reader has to rebuild the document by hand or flip between two badly matched versions. Doc2X keeps pages, headings and visual structure close to the source so the translation can be read, compared and shared as a document rather than as loose text.
The homepage says translation is "powered by GPT-4o, DeepSeek, Claude, and dozens of top-tier models". The help center is more specific about what a user can pick today: DeepSeek V4 Flash is the default and the only model on the free plan, while DeepSeek V4 Pro and Gemini 3.1 Flash Lite Preview are listed for paid subscriptions, and the pricing page describes paid model access as "Gemini, GPT, and more". On privacy, Doc2X states that documents are encrypted in transit and never used to train models, and its help center says translation history is retained for 30 days, so results you want to keep should be exported in time. Sign-in with email or Google is required before uploading. The site does not name the company that operates it; it directs holders of older accounts to a separate legacy service, Doc2X CN, at doc2x.noedgeai.com. Pricing is credit based, with a free plan that refills monthly and two paid subscriptions billed monthly or yearly.
You start on the Workbench. After signing in, you drag a PDF into the upload area, and Doc2X shows the file name, size, page count and the size limit for your plan before anything is spent: 20 MB on Free, 200 MB on Starter and 500 MB on Pro. Scanned PDFs are supported because Doc2X parses each page before translating, though the FAQ notes that very low-quality scans can reduce recognition accuracy.
Next you choose the target language and a translation model. The help center recommends keeping the default model for most documents and switching to a higher-quality model only when wording accuracy matters. Paid users can also open Advanced settings to skip tables, code, figures or references that should stay unchanged, and to turn on Enhanced translation, which uses more document context for long or specialized documents.
Doc2X then uploads, parses and translates the file. Before the task starts it shows a credit estimate based on page count, model and options, and it settles the final charge after processing, returning the difference if the real cost was lower. The credits guide lists per-page rates of 0.5 credits for DeepSeek V4 Flash, 1.0 for DeepSeek V4 Pro and 1.5 for Gemini 3.1 Flash Lite Preview, plus a small per-page surcharge for PDFs longer than 50 pages.
Finally you open the completed task in the reader. The layout-preserved translated PDF is the main view, a comparison view shows the source and the translation side by side, and when the task includes it a text-first Markdown view opens in the same reader. Completed tasks stay under Tasks so you can reopen or export them later within the retention window.
- •Layout-Preserving PDF Translation - The core output is a fixed-layout translated PDF that keeps fonts, colors, tables and images in place rather than flattening the document into plain text. The homepage demonstrates it on a real arXiv paper shown next to its source PDF, and the result can be exported from the detail view.
- •Side-by-Side Comparison and Text View - A comparison view puts the source and translated PDF next to each other so you can check an important paragraph against the original, and a text-first Markdown detail view is available inside the reader when the task includes it, without a separate upload or mode.
- •Choice of Translation Models - You choose the model for each task. DeepSeek V4 Flash is the default and the free-plan model; paid plans add stronger models metered by credits, and the help center notes that models from different providers can have different translation preferences.
- •Advanced Translation Controls - On paid plans the pricing page lists enhanced context translation, a custom glossary and selected-section translation, along with more export templates. The help center describes ignore options for tables, code, figures and references, and an Enhanced translation mode for terminology-heavy academic, technical, legal or financial documents.
- •Scanned and Large Documents - Because files are parsed before translation, scanned and layout-heavy PDFs can be processed, and paid plans raise the per-file limit to 200 MB or 500 MB, allow five signed-in devices instead of three, and remove the free limit of one export every 30 seconds.
- •Privacy and Quality Support - Files are encrypted in transit and, according to the site, never used to train models. Paid plans include credit compensation when a translation is unsatisfactory: you email support with your account email, the task ID and a description of the problem, and credits are returned once the issue is confirmed.
Doc2X fits students and researchers who need to read papers, theses and technical reports in another language while keeping equations, figures and references where the author put them; the Pro plan is pitched at students, researchers and professionals. The reference-skipping option and the arXiv example make it clear that academic PDFs are a central use case.
Professionals who work with manuals, reports and contracts can use the comparison view to check critical passages, Enhanced translation for consistency across long documents, and the glossary to keep terminology fixed. Freelancers handling light work and occasional longer documents are the stated audience for the Starter plan.
Casual users can try the free monthly credits on small PDFs up to 20 MB with the default model before deciding whether they need larger files, stronger models or the advanced options. Teams with strict data requirements should note that the operator is not named on the site and that task history is kept for only 30 days.
Pricing
$0 - $24.90/mo
- Free$0/mo
- Starter$9.90/mo
- Pro$24.90/mo
From the vendor pricing page, 2026-10-15
















