Privatranslate

PrivaTranslate is a privacy-first web page translator for people who want translation without sending page content to a cloud translation service.

Instead of relying on remote translation APIs, PrivaTranslate connects to a local LLM provider that you configure, such as Ollama, LM Studio, or another OpenAI-compatible local endpoint. Page text is sent only to your configured localhost provider.

Key features:

- Local LLM translation through Ollama, LM Studio, or OpenAI-compatible local servers
- Three translation modes: Overlay, Replace, and Side-by-Side
- Structure-preserving replacement for readable translated pages
- Restore action for reverting replaced page text
- Right-click context menu for translating pages or selected text
- Configurable source language, target language, model, timeout, and concurrency
- Optional batching for short phrases to reduce local request overhead
- No analytics, no tracking, no developer-operated backend

Translation modes:

- Overlay: keep the original page intact and hover translated text to inspect the translation.
- Replace: replace page text in place while preserving the surrounding page structure.
- Side-by-Side: open a comparison view with original text and translated text in separate columns.

Requirements:

PrivaTranslate requires a local model provider running on your machine. For example, you can run Ollama locally and install a translation-capable model such as qwen2.5:7b, or use LM Studio with its local OpenAI-compatible server.

Privacy:

PrivaTranslate does not collect personal information, browsing history, page content, translation requests, translation results, analytics, telemetry, or tracking data. Settings are stored in Chrome storage. Optional provider API keys are stored locally in your browser profile.

PrivaTranslate is not a cloud translation service. Translation quality and speed depend on the local model and provider you choose.