explaingit

xxnuo/mtranserver

★ 4,513C++Audience · developerComplexity · 2/5Setup · easy

TLDR

Self-hosted translation server that runs small offline AI models on your own machine for fast, private text translation across major languages in about 50ms, with no cloud service or GPU required.

Mindmap

mindmap
  root((MTranServer))
    What it does
      Offline translation
      Private local server
      50ms fast responses
    Deployment options
      Desktop app
      Docker
      Node.js CLI
    Integrations
      Immersive Translate
      DeepLX
      VS Code plugin
    Audience
      Privacy-focused users
      Developers
Click or tap to explore — scroll the page freely

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

Things people build with this

USE CASE 1

Replace cloud translation APIs in browser extensions like Immersive Translate or DeepLX with a free local server.

USE CASE 2

Run private offline text translation without sending any data to external services.

USE CASE 3

Deploy a local translation service with Docker and connect it to any compatible browser extension or app.

USE CASE 4

Use the MTranCode VS Code plugin to translate code comments directly inside your editor via a local MTranServer.

Tech stack

C++Node.jsDocker

Getting it running

Difficulty · easy Time to first run · 30min

The first translation for each language pair triggers an automatic model download, all subsequent requests for that pair work offline.

In plain English

MTranServer is a translation server you run on your own computer or server. It uses small offline AI models to translate text between the world's major languages without sending anything to a cloud service. The project emphasizes low memory usage and fast response times: according to the README, individual translation requests return results in around 50 milliseconds once the models are loaded. It does not require a graphics card. You can run it on Windows, Mac, or Linux. There is a desktop application that installs with a single download and adds a system tray icon for managing the server. Developers can also start the server from the command line using a Node.js package runner, or deploy it using Docker with a short configuration file. The first time you request a translation for a particular language pair, the server downloads the required model automatically. After that, the same pair works offline. A key feature is compatibility with the APIs of well-known translation tools. If you have browser extensions or apps that work with Immersive Translate, Kiss Translator, DeepL, DeepLX, or Google Translate, you can often point them at your local MTranServer instead, replacing the cloud service with your own private instance. The README includes a table of which URL paths correspond to each supported tool. The README notes that translation quality from these small offline models is lower than what you would get from large cloud-based translation services. The project is designed for speed, privacy, and cost-free operation rather than maximum accuracy. For high-quality translation needs, the author recommends using a large language model API instead. A companion VS Code plugin called MTranCode lets you translate code comments directly inside supported editors by connecting to a running MTranServer instance.

Copy-paste prompts

Prompt 1
Show me how to set up MTranServer using Docker on my Mac so I can use it with the Immersive Translate browser extension.
Prompt 2
How does MTranServer download language models, and how do I pre-load models for the language pairs I use most often?
Prompt 3
How do I configure the Immersive Translate or DeepLX browser extension to point to my local MTranServer instead of the cloud API?
Prompt 4
What are the real tradeoffs between MTranServer offline models and a cloud translation API for everyday document translation?
Prompt 5
How do I install and use the MTranCode VS Code plugin with a running MTranServer instance to translate code comments?
Open on GitHub → Explain another repo

← xxnuo on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.