Zero-setup local AI workstation — GUI desktop app with one-click environment install for everyone.
ModxAI Studio is a desktop application that turns a single machine into a local AI workstation. It bundles model management, AI chat, TTS, data processing, training, and Stable Diffusion–based image/video generation into one unified, GUI-driven tool — without requiring users to configure Python environments, Docker containers, or command-line scripts.
Official Site: https://www.modxai.net/en · 中文官网: https://www.modxai.net/zh
-
AI Chat (Text & Multimodal)
Local LLM chat powered byllama.cpp, with support for text Q&A, code generation, multimodal (image+text) queries, system prompts, RAG retrieval, and high-quality conversation collection for later training. -
TTS Speech Synthesis
Local text-to-speech with Chinese/English bilingual support, multiple voice presets, speed adjustment, and automatic language detection. -
Stable Diffusion Chat (SD Chat)
Conversational interface for text-to-image, image-to-image, and video generation. Per-conversation parameter profiles (resolution, steps, CFG, seeds, samplers), negative-prompt templates, and parameter recommendations. -
Model Library
Unified management for chat, multimodal, embedding, rerank and SD models. Supports one-click download from Hugging Face / ModelScope, local import of GGUF/SafeTensors, and LoRA adapters. -
Datasets & Pipelines (SFT / RAG / Audio2Text)
Pipeline-style data processing: convert, clean, split, analyze, AI-synthesize, vectorize and output standardized SFT or RAG datasets. Includes audio-to-text transcription with resumable flows. -
Model Training (LoRA & Fine-tuning)
End-to-end training workflow: preparation, real-time monitor, evaluation, interactive test, and GGUF packaging. LoRA keeps VRAM requirements low (as little as 8GB). -
One-Click Environment Setup
Built-in environment manager with sandboxed portable Python, CPU/GPU variants, mirror selection, and resumable installs. No Docker, no manual Python setup, no command line required.
| Advantage | Description |
|---|---|
| Zero Technical Barrier | GUI-first design. No Docker, no Python config, no command line. Just download, install, and use. |
| One-Click Environment | Sandboxed, portable Python with on-demand dynamic dependencies. Does not pollute your system. |
| Local-First & Privacy | Models, datasets and chat history stay on your machine. No cloud required. |
| All-in-One Workflow | Chat, TTS, SD image/video, RAG, training, model management — unified in one app. |
| Offline Ready | No account needed. Fully functional offline after initial setup. |
| Resource Efficient | Quantized models run on regular PCs. LoRA training with 8GB VRAM. |
This repository hosts the official documentation for ModxAI Studio.
- Documentation (English):
docs/en/ - Documentation (简体中文):
docs/zh/
Key sections include:
- Getting Started: Introduction and 5-minute quick start
- Features · Chat: AI chat, prompt optimization, personalized training data
- Features · TTS: Text-to-speech synthesis
- Features · SD Chat: Stable Diffusion image and video generation
- Features · Dataset: SFT data, RAG knowledge bases, audio-to-text
- Features · Models: Model library, downloads, LoRA, SD models
- Features · Training: Prepare, monitor, evaluate, test, package
- Features · Settings: Environment setup, license, global options
- FAQ: Common problems and troubleshooting
ModxAI Studio is built on top of a rich open-source ecosystem. Special thanks to: PyTorch, llama.cpp, Stable Diffusion.cpp, Transformers, TRL, PEFT, faster-whisper, Electron, Vue 3, FastAPI, and many more.
零门槛本地 AI 工作站 — 一键环境安装，GUI 桌面应用，面向所有人。
ModxAI Studio 是一款面向普通用户的本地 AI 一站式解决方案。它将大语言模型对话、TTS 语音合成、Stable Diffusion 图像/视频生成、数据处理和模型训练整合到一个简洁易用的桌面应用中 — 无需配置 Python 环境、Docker 容器或命令行操作。
官网（英文）：https://www.modxai.net/en · 官网（中文）：https://www.modxai.net/zh
-
AI 对话（文本 & 多模态）
基于llama.cpp的本地大模型对话，支持文本问答、代码生成、多模态图文理解、系统提示词、RAG 检索增强，以及高质量对话收藏用于后续训练。 -
TTS 语音合成
本地文本转语音，支持中英双语、多音色选择、语速调节、自动语言检测。 -
SD 图像/视频生成
以会话方式完成文生图、图生图和视频生成；每个会话拥有独立的参数配置，支持负面提示词模板。 -
模型库管理
统一管理聊天、多模态、Embedding、Rerank 与 SD 模型；支持从 Hugging Face / ModelScope 一键下载，本地导入 GGUF / SafeTensors，挂载 LoRA 适配器。 -
数据处理流水线（SFT / RAG / 语音转写）
流水线式数据处理：解析、清洗、切分、AI 合成、向量化，导出规范的 SFT 或 RAG 数据集；支持音频转文本、断点续跑。 -
模型训练（LoRA 微调）
覆盖从任务配置、训练监控、Checkpoint 评估、交互测试到 GGUF 打包的完整流程。LoRA 训练仅需 8GB 显存即可开始。 -
一键环境安装
内置环境管理器，提供沙盒化的便携 Python 环境，区分 CPU / GPU 方案，支持镜像源选择和断点续装。无需 Docker、无需手动配置 Python、无需命令行。
| 优势 | 说明 |
|---|---|
| 零技术门槛 | 图形界面优先。无需 Docker、无需配置 Python、无需命令行，下载安装即用。 |
| 一键环境配置 | 沙盒隔离的便携 Python，按需动态加载依赖，不污染系统环境。 |
| 本地优先 & 隐私安全 | 模型、数据集与对话记录保存在本机，无需云端。 |
| 功能全面一站式 | 对话、TTS、SD 图像/视频、RAG、训练、模型管理统一在一个应用内。 |
| 离线可用 | 无需账号注册，下载完成后可完全离线使用。 |
| 资源高效 | 支持量化模型，普通电脑可运行；LoRA 训练仅需 8GB 显存。 |
本仓库托管 ModxAI Studio 的官方文档，支持中英文双语：
- 英文文档：
docs/en/ - 中文文档：
docs/zh/
主要内容包括：
- 快速入门：产品简介与 5 分钟上手
- 功能说明：对话、TTS、SD 生成、模型库、数据处理、训练、设置
- FAQ：常见问题与故障排查
ModxAI Studio 构建在丰富的开源生态之上，特别感谢：PyTorch、llama.cpp、Stable Diffusion.cpp、Transformers、TRL、PEFT、faster-whisper、Electron、Vue 3、FastAPI 等开源项目与社区。
ModxAI Studio — Making local AI accessible to everyone.