Google Translate as a proxy: unblock without install

On a machine where you cannot install anything - a school PC, a work laptop, a borrowed phone - there are exactly two tools left: the archive and Google Translate. The translate trick works because Google's translation service fetches the page you want on Google's own servers, translates it, and shows it to you from Google's domain. The blocked site is never touched from your network, and your network only ever talks to Google [1].

It is a proxy with the limits of a proxy. This guide gives you the exact URL pattern, what survives the trip, and how to read the error states.

The URL pattern

Take the address you want, swap its dots for dashes, and append .translate.goog - then set the languages in the _x_tr_ parameters [1]:

https://example.com/page
https://example-com.translate.goog/page?_x_tr_sl=auto&_x_tr_tl=de&_x_tr_hl=de

• _x_tr_sl - source language (auto works)
• _x_tr_tl - target language, the one you will read
• _x_tr_hl - the interface language

Subdomains follow the same rule: en.wikipedia.org becomes en-wikipedia-org.translate.goog. If typing the long form feels clunky, the legacy entry point still works: open https://translate.google.com/translate?sl=auto&tl=de&u=https://example.com/page and Google redirects you into the proxy form [2].

What survives - and what does not

Survives: text-heavy pages, news articles, documentation, forum threads. The reading experience is genuinely good, and links inside the page are rewritten to keep you in the proxy [1].

Does not survive:

• Logins. You are reading a translated copy; forms and sessions do not pass through. Never type credentials into a translated page - that is a phishing pattern by construction.
• JavaScript-heavy apps. Single-page apps, infinite timelines, editors and player interfaces mostly break. Static articles are the sweet spot.
• Media and some assets. Cross-origin resources may refuse to load under Google's domain.
• The original wording. You are reading machine translation; anything where precise wording matters (law, citations) should be checked against the original through another method [5].

The error states, translated

Two failure messages cover most confusion:

1. "Can't translate this page" (the error-dog page). Nine times out of ten the URL is wrong: the parameters must be spelled _x_tr_sl (with underscores and tr), the hostname must be dashes-for-dots exactly, and there must be no leftover https:// inside the hostname part. Check character by character against the pattern above.
2. A reCAPTCHA "unusual traffic" page. Google rate-limits aggressive proxy use - shared networks and datacenter IPs hit this wall. It usually passes with the puzzle or clears after a while; if your network trips it constantly, the archive is the more reliable zero-install route.

When to reach for it

The translate proxy is the right first try when: you cannot install anything, the blocked page is mostly text, and the goal is reading. It also doubles as a poor man's cache - you get the current page, not a stored snapshot [1].

Where it loses to the alternatives:

• You need the exact original text - the archive stores the untranslated copy.
• You need to interact - a tunnel (WARP, Psiphon, Tor) does the whole site, not a reading copy.
• The site is dynamic - the archive plus a tunnel combination beats a broken translation.

As always, diagnose before you pick: OONI Probe tells you what your network is actually filtering [4], and our general unblock guide maps the symptom to the fix. The usual note applies: circumvention is legal in most places but not all, this is information and not legal advice, and a proxy this convenient is also not a privacy tool - Google sees everything you read through it [5].