LibreTranslate

What is LibreTranslate?

LibreTranslate is a free and open source machine translation API that you run yourself. It translates text between languages using the open source Argos Translate engine, so every request stays on your own hardware — no Google, no Azure, and no data leaving your network. It ships with a simple web UI and a REST API.

What is LibreTranslate best for?

Teams that need translation but can’t send text to a third-party cloud — healthcare, legal, government, and defense, or anyone handling private data. It’s also a fit for developers who want an unmetered translation API for bulk or batch jobs without paying per character, and for offline or air-gapped deployments.

What can LibreTranslate do?

• Translate text through a clean REST API (/translate, /detect, /languages) with an included web interface
• Auto-detect the source language before translating
• Run fully offline once the language models are downloaded
• Translate files (.txt, .docx, .odt, and more) in addition to raw text
• Deploy with pip, Docker, or Kubernetes, with optional CUDA/GPU acceleration
• Add or remove language models to control download size and memory use
• Self-host behind an API-key and rate-limiting layer for multi-user access

Is LibreTranslate free?

Yes — the software is free and open source under AGPL-3.0, and self-hosting costs only your own server. The project also runs a managed API at libretranslate.com for people who don’t want to host it: the public endpoint now requires a (free, rate-limited) API key, with paid Pro (around $29/mo) and Business (around $58/mo) tiers for higher limits. You’re paying for hosting and throughput, not the software.

Where does LibreTranslate fall short?

• Translation quality trails the commercial engines. For technical, legal, or low-resource language pairs, output is less consistent than Google Translate or DeepL, which train on far larger datasets. For high-traffic pairs like English–Spanish it’s serviceable, but it rarely matches DeepL’s fluency.
• Fewer languages. It covers roughly 40–50 languages depending on your deployment, versus 200+ on Google Translate — so exotic or rare pairs may simply be unavailable.
• Language models are heavy. Each pair is a downloaded model that consumes disk and RAM; loading many languages at once meaningfully raises the memory footprint of the container.

What does LibreTranslate replace?

LibreTranslate is a self-hosted alternative to Google Translate, DeepL, and Amazon Translate. It does the same translate-this-text job through an API, but you run it on your own infrastructure — trading some translation quality for privacy, offline operation, and no per-character billing.

FAQ

Is LibreTranslate open source? Yes. It’s licensed under AGPL-3.0, a strict copyleft license, and the code is fully public on GitHub. The AGPL means that if you offer it as a network service, you must make your source modifications available.

Can I self-host LibreTranslate for free? Yes. Self-hosting is completely free — you only pay for the server it runs on. The hosted API at libretranslate.com is the paid, managed option for those who don’t want to run it.

Is LibreTranslate a good Google Translate alternative? For privacy and cost, yes — it keeps text on your servers and has no per-character fees. If you need the highest translation quality or the widest language coverage, Google Translate and DeepL are still ahead.

What do I need to run LibreTranslate? A server with Docker (or Python and pip), enough RAM and disk for the language models you load, and optionally a GPU for faster throughput. A small deployment with a few language pairs runs comfortably on a modest VPS.