本地推理需自备 GGUF 模型文件,或通过应用内下载获取;语音输出需另装配套应用 Maise;构建产物在 android/app/build/outputs/apk/release 目录下。
maid
Mobile-Artificial-IntelligenceMIT活跃维护一个跨平台本地大模型聊天客户端,通过统一界面连接 llama.cpp 本地模型以及 Anthropic、OpenAI、DeepSeek、Ollama 等远程服务。
Maid is a free and open source application for interfacing with llama.cpp models locally, and with Anthropic, DeepSeek, Ollama, Mistral and OpenAI models remotely.
git clone https://github.com/Mobile-Artificial-Intelligence/maid.git
yarn install
yarn build-androidREADME 只说明应用面向 Android,本地推理依赖 llama.cpp 运行 GGUF 模型;未给出具体显存或内存要求。
在 Android 手机上本地跑 GGUF 模型,也能接入多家云端大模型 API 的聊天客户端
一个跨平台本地大模型聊天客户端,通过统一界面连接 llama.cpp 本地模型以及 Anthropic、OpenAI、DeepSeek、Ollama 等远程服务。
核心能力
- 用 llama.cpp 在设备本地离线运行 GGUF 模型
- 填写自己的 API Key 接入 Anthropic、OpenAI、DeepSeek、Ollama 等远程服务
- 在应用内浏览并一键下载 Hugging Face 上的精选模型
- 支持从本地存储加载任意 GGUF 文件
- 按会话调节 temperature、top-p、top-k 与上下文长度
容易踩的坑
- v3.0.0最新正式版2026-03-10 · 2026-03-10 11:09 UTC
Whats New
- Add vision model Support
- Add system text to speech support
- Add system speech recognition support
- Add reasoning support
- Add markdown rendering
- Add upvote / downvote report functionality (Google Requirement)
New Contributors
- @Alex-wuhu made their first contribution in https://github.com/Mobile-Artificial-Intelligence/maid/pull/743
- v2.0.72025-04-25 · 2025-04-25 01:15 UTC
查看更新说明
- Add Arabic Locale
- Add Dutch Locale
- Add Portuguese Locale
- v2.0.62025-04-18 · 2025-04-18 02:40 UTC
查看更新说明
What's Changed
- [ImgBot] Optimize images by @imgbot in https://github.com/Mobile-Artificial-Intelligence/maid/pull/683
- Supabase support by @danemadsen in https://github.com/Mobile-Artificial-Intelligence/maid/pull/685
Full Changelog: https://github.com/Mobile-Artificial-Intelligence/maid/compare/v2.0.5...v2.0.6
- v2.0.52025-04-07 · 2025-04-07 11:17 UTC
查看更新说明
- Add Polish Locale
- Add Vietnamese Locale
- Add warning for NSFW model
- v2.0.42025-04-01 · 2025-04-01 01:00 UTC
查看更新说明
What's Changed
- Gemini by @ADev-123 in https://github.com/Mobile-Artificial-Intelligence/maid/pull/674
- Huggingface by @danemadsen in https://github.com/Mobile-Artificial-Intelligence/maid/pull/678
New Contributors
- @ADev-123 made their first contribution in https://github.com/Mobile-Artificial-Intelligence/maid/pull/674
Full Changelog: https://github.com/Mobile-Artificial-Intelligence/maid/compare/v2.0.3...v2.0.4
用户点评
登录用户的真实使用记录,可按版本和硬件筛选还没有人点评。第一个说出真实体验的人最有价值。
Maid - Mobile Artificial Intelligence Distribution
Maid is a free and open source application for interfacing with llama.cpp models locally, and with API Route, Anthropic, DaoXE, DeepSeek, Mistral, Novita, Ollama, llmman, OrcaRouter, LiteLLM, Opper, Runcrate and OpenAI models remotely. Maid is built using React Native and is available for Android. The application is designed to be fast, efficient and user-friendly, making it easy for users to interact with their models on the go.
For text to speech functionality check out Maid's companion app Maise.
Features
- Local inference — run GGUF models fully on-device via llama.cpp; no internet required
- Remote providers — connect to API Route, Anthropic, DaoXE, DeepSeek, Mistral, Novita, Ollama, llmman, Opper, OrcaRouter, Runcrate, and OpenAI with your own API key, or to a self-hosted LiteLLM proxy for 100+ providers behind one endpoint
- One-tap model downloads — browse and download curated Hugging Face models (Qwen, Phi, LFM, TinyLlama, and more) directly from the app
- Bring your own model — load any GGUF file from local storage
- Conversation management — create, rename, delete, export, and import chats as JSON
- Customisable parameters — tune temperature, top-p, top-k, context length, and other generation parameters per session
- Custom system prompt — set a global system prompt and assistant persona
- Voice output — pair with Maise for text-to-speech
- Material You theming — light and dark themes that follow your system preference
- Fully open source — MIT licensed, no telemetry, no ads
Manual
The user manual is available from releases in PDF format, or can be built from source using the instructions below.
Cloning
To clone the repository, use the following command:
git clone https://github.com/Mobile-Artificial-Intelligence/maid.git
Setup
Run the following command to install dependencies:
yarn install
Testing
Run the test suite with:
yarn test
Building
To build the project, use the following command:
yarn build-android
The output APK will be located in the android/app/build/outputs/apk/release directory.
Disclaimer
Maid is distributed under the MIT licence and is provided without warranty of any kind, express or implied. Maid is not affiliated with Huggingface, Meta (Facebook), MistralAi, OpenAI, Google, Microsoft or any other company providing a model compatible with this application.
Signing
Signing Key Fingerprint
MD5:
BE:AC:29:41:F5:41:D2:26:42:DD:D1:A3:85:21:E1:16
SHA-1:
48:F6:DC:73:09:CE:19:C6:A9:70:7E:A2:9A:B7:6F:42:2D:41:32:30
SHA-256:
83:5E:D2:2E:D8:95:C4:C2:72:D6:98:AA:6E:4E:48:DB:0B:4E:36:DC:CF:70:10:D5:DE:15:03:4A:C9:E1:B9:6F
Upload Key Fingerprint
MD5:
C0:86:A0:F3:E8:E5:4D:46:60:8A:37:4E:DB:11:CC:C7
SHA-1:
77:FC:77:2B:21:5E:9F:36:31:79:09:DF:7D:F4:1F:CA:96:0C:39:17
SHA-256:
54:EE:B9:9F:14:38:D9:68:9B:C2:C6:7F:F9:DD:A3:E3:D8:28:D3:80:76:46:B7:24:46:71:9F:61:D9:63:E6:98
License
This project is licensed under the MIT License.