MTranServer

MTranServer is an ultra-fast offline translation model server designed for minimal resource consumption. It does not require a GPU and achieves an average response time of 50ms per request. It supports major world languages and focuses on offline translation, response speed, multi-platform deployment, and local execution to provide unlimited free translation.

Note: Due to model optimization for speed and size, translation quality may not match large-scale online translation models. For high-quality requirements, online LLM APIs are recommended.

For containerized deployment, create a compose.yml file in an empty directory and use the following configuration. This setup maps the local ./models directory to the container to persist downloaded models and exposes port 8989.

services:
  mtranserver:
    image: xxnuo/mtranserver:latest
    container_name: mtranserver
    restart: unless-stopped
    ports:
      - "8989:8989"
    environment:
      - MT_HOST=0.0.0.0
      - MT_PORT=8989
      - MT_OFFLINE=false
      # - MT_API_TOKEN=ihr_geheimer_token_hier
    volumes:
      - ./models:/app/models

Then run:

docker pull xxnuo/mtranserver:latest
docker compose up -d

To use MTranServer with the Immersive Translate plugin:

1. In Immersive Translate settings, enable Beta features in Developer Mode.
2. Under Translation Services, find Custom API Settings.
3. Set the API URL based on your security requirements:
   • Without password: http://localhost:8989/imme
   • With password: http://localhost:8989/imme?token=your_token (replace your_token with your MT_API_TOKEN).
4. To maximize performance, increase Max requests per second (e.g., to 512) and set Max paragraphs per request to 1 in the Custom API settings.

To install the server globally on your system, use npm (or bun, pnpm, etc.). After installation, you can start the server by running the mtranserver command.

npm i -g mtranserver@latest

mtranserver

For a quick start without permanent installation, you can run the MTranServer directly using a package manager like npx, bunx, or pnpx.

npx mtranserver@latest

To use mtranserver in your Node.js or Bun applications, install the latest version using the following command:

bun install mtranserver@latest

MTranCode is a plugin for VS Code, Cursor, Augment, and other VS Code-based IDEs that translates code comments.

1. Search for MTranCode in your IDE's plugin marketplace and install it.
2. By default, it attempts to connect to the server at http://localhost:8989. You can change this address in the plugin settings if your server is hosted elsewhere.

MTranCode is a comment translation plugin that supports VS Code-based IDEs including VS Code, Cursor, and Augment.

1. Search for MTranCode in your IDE's extension marketplace to install it.
2. By default, the plugin calls the server at http://localhost:8989 to translate comments and code. You can change this address in the plugin settings.

To fully utilize MTranServer performance in the Kiss Translator plugin:

1. Scroll down to the Custom interface in the Settings page.
2. Set Maximum Concurrent Requests (e.g., 100).
3. Set Request Interval Time (e.g., 1).
4. Adjust these values based on your specific server hardware configuration.