2025-03-01 07:49:07 +08:00
2025-03-01 23:46:09 +08:00
2025-03-02 20:34:30 +08:00
2025-03-02 10:44:01 +08:00
2025-03-01 07:59:14 +08:00
2025-03-01 07:59:14 +08:00
2025-03-03 00:37:20 +08:00
2025-03-01 07:49:07 +08:00
2025-03-01 23:46:09 +08:00
2025-03-01 23:46:09 +08:00
2025-03-01 07:49:07 +08:00

MTranServer

Mini Translation Server

中文 | 日本語 | English

A high-performance offline translation server with minimal resource requirements - runs on CPU with just 1GB memory, no GPU needed. Average response time of 50ms per request.

Translation quality comparable to Google Translate.

Note: This model prioritizes performance optimization, so translation quality may not match that of large language models.

Comparison with Similar Projects (CPU, English to Chinese)

Project Name Memory Usage Concurrency Translation Quality Speed Additional Info
facebook/nllb-200-distilled-600M Very High Poor Average Slow Android's RTranslator has optimizations but still has high resource usage and slower speed
LibreTranslate Very High Average Average Medium Mid-range CPU: 3 sentences/s, high-end CPU: 15-20 sentences/s. Details
OPUS-MT High Average Below Average Fast Performance Benchmarks
MTranServer (This Project) Low High Average Ultra Fast 50ms average response time per request

*Note: Non-rigorous testing, non-quantized version comparison, for reference only.

Docker Compose Server Deployment

Currently only supports Docker deployment on amd64 architecture CPUs.

Support for ARM and RISC-V architectures is under development 😳

Usage

Currently available through browser extensions: Immersive Translation and Kiss Translator.

Client Version

Service core, Windows and Mac client versions (MTranServerCore) are under development (not yet public).

Support the Project

Support me on Afdian


WeChat: x-xnuo

X: @realxxnuo

Feel free to connect with me to discuss technology and open-source projects!

I'm currently seeking job opportunities. Please contact me to view my resume.


Description
Offline translation model server with low resource consumption, fast speed, and private deployment capability. 低资源占用速度快可私有部署的离线翻译模型服务器
Readme Apache-2.0 17 MiB
Languages
C++ 89.2%
TypeScript 5.3%
JavaScript 2.8%
Go 2.5%