Margin Read

Privacy-first translation and notes, placed where reading happens.

Margin Read is a privacy-first browser extension for bilingual reading. It keeps the original webpage text in place and adds translations below the matching source blocks, so you can compare both versions without losing the page context. Why Margin Read: - Privacy-first by design: no bundled account, cloud sync, telemetry, or official translation quota. - Bring your own provider: use OpenAI, Anthropic Claude, Google Gemini, or OpenAI-compatible local runtimes. - Local LLM support: works with LM Studio, Ollama, and llama.cpp server through OpenAI-compatible endpoints. - Reading-first layout: translations are placed where reading happens, instead of replacing the original page. - Flexible display styles: choose integrated translations or highlighted translation blocks. - Translation cache controls: use persistent cache, session-only cache, or disable caching. - Dynamic page support: observes newly inserted content on supported pages. - X optimization: optionally targets tweet text and longform article content while avoiding usernames, metrics, media, and controls. Margin Read is currently in beta. It works best on article pages, documentation pages, legacy text-heavy pages, and selected dynamic pages. Some highly interactive web apps or unusual layouts may still need tuning. Known limitations: - PDF translation is not supported yet. - OCR and image text translation are not supported yet. - Subtitle translation is not supported yet. - Input box translation is not supported yet. Margin Read is open source: https://github.com/linyiru/margin-read