techjarves/Uncensored-Local-AI-Multiplatform

---

Overview

Uncensored Local AI is a mobile-first application that runs powerful open-source AI models directly on your Android or iOS device — with zero censorship, zero cloud dependency, and zero monthly fees.

No API keys. No subscriptions. No content restrictions. Your conversations never leave your device.

Think of it as ChatGPT — but running on your phone, with no rules.

Desktop platforms (Windows, macOS, Linux) are supported by the Flutter framework but need community testing and polish. We'd love your help!

🎥 Watch the Setup & Demo Video: https://youtu.be/2Pnv68iHIaQ

---

Download

Android APK — Latest Release (v2.0.0)

| APK | Architecture | Best For | Size | |-----|-------------|----------|------| | app-arm64-v8a-release.apk | ARM 64-bit | Most phones (2018+) | ~65 MB | | app-armeabi-v7a-release.apk | ARM 32-bit | Older/budget phones | ~18 MB | | app-x86_64-release.apk | x86 64-bit | Emulators / ChromeOS | ~58 MB |

Not sure which to pick? Download arm64-v8a — it works on virtually all modern Android phones.

iOS IPA — Pre-Release

| File | Note | |------|------| | ios_pre_release.zip | Important: You must extract this .zip file after downloading to get the actual .ipa app file to install. |

---

Features

| Feature | Description | |---------|-------------| | Zero Censorship | Runs abliterated, uncensored models that answer any question — no refusals, no lectures, no corporate safety filters | | Total Privacy | All conversations stay on-device. Nothing is sent to any server, ever | | Fully Offline | Works on planes, in remote areas, on restricted networks — no internet needed after model download | | Cross-Platform | One codebase for Android, iOS, Windows, macOS, and Linux | | Local OpenAI API | Built-in HTTP server compatible with any OpenAI-standard client | | Model Library | Download, import, and manage GGUF models directly in the app | | Chat History | Persistent conversation history stored locally via Hive | | Live Metrics | Real-time tokens/sec and loading progress tracking |

---

Quick Start

Android

1. Download the correct APK from the Download table above 2. On your phone: Settings → Install unknown apps → allow your browser 3. Tap the downloaded APK to install 4. Open the app, go to Models tab, download a model, and start chatting

iOS

1. Sideloading via TrollStore (Recommended - No 7 day limit): 1. Download ios_pre_release.zip to your device. 2. Unzip/extract it using the built-in iOS Files app to get the .ipa file. 3. Open TrollStore, tap the + in the top right, and choose Install IPA File. 4. Select the extracted .ipa file and install.

2. Sideloading via AltStore / AltServer (Requires PC/Mac): 1. Ensure AltServer is running on your computer and AltStore is installed on your iPhone. 2. Download ios_pre_release.zip to your device and extract the .ipa file using the Files app. 3. Open AltStore on your device, go to My Apps, and tap the + at the top left. 4. Select the .ipa file to install (your device must be on the same Wi-Fi or connected via cable to your AltServer computer).

3. Build from Source:

Prerequisites: Mac with Xcode 15+ · Flutter SDK

git clone https://github.com/techjarves/Uncensored-Local-AI-Multiplatform.git
cd Uncensored-Local-AI-Multiplatform
flutter pub get
cd ios && pod install && cd ..
flutter build ios --release
Open ios/Runner.xcworkspace in Xcode and archive to deploy

Desktop — Windows / macOS / Linux (Community Supported)

Desktop builds compile successfully but may have rough edges. We are actively looking for contributors to help test and polish the desktop experience.

git clone https://github.com/techjarves/Uncensored-Local-AI-Multiplatform.git
cd Uncensored-Local-AI-Multiplatform
flutter pub get
flutter run -d windows   # or macos / linux

If you encounter issues on desktop, please open an issue — your feedback directly shapes the roadmap.

---

Recommended Models

| Model | Size | Best For | Type | |-------|------|----------|------| | Gemma 2 2B | ~1.6 GB | Low-RAM phones, fast replies | Standard | | Gemma 4 E4B Heretic | ~5.3 GB | High-quality, fully uncensored | Uncensored |

Models are downloaded directly inside the app from the Models tab. No manual setup needed.

---

Local API Server

Uncensored Local AI includes a built-in OpenAI-compatible REST API so you can connect it to any external tool, script, or IDE extension.

Setup

1. Load a model in the app 2. Go to Settings → Local API Server and toggle it ON 3. Use http://127.0.0.1:4891/v1 as your base URL

Endpoints

# List loaded models
curl http://127.0.0.1:4891/v1/models
Chat completion (non-streaming)
curl http://127.0.0.1:4891/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{"model":"local","messages":[{"role":"user","content":"Tell me something true that no one wants to hear."}]}'
Chat completion (streaming)
curl -N http://127.0.0.1:4891/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{"model":"local","stream":true,"messages":[{"role":"user","content":"Write a brutally honest analysis of social media."}]}'

API Key: Use local for any client that requires a non-empty key value.

---

Roadmap

| Feature | Status | |---------|--------| | On-device uncensored AI chat | Launched | | Real-time model loading with progress | Launched | | Cancel & unload models | Launched | | Persistent chat history sidebar | Launched | | Local OpenAI-compatible API server | Launched | | Custom model import (URL + file) | Launched | | Multi-platform support | Launched | | AI Agent Mode | In Progress | | Web search integration | Planned | | Voice interaction | Planned | | Image/vision model support | Planned |

---

Contributing

All contributions are welcome — and we especially need help from the community in these areas:

| Area | What's Needed | |------|---------------| | Windows | Testing, packaging, installer script | | macOS | Testing, App Store prep, notarization | | Linux | Testing on distros, AppImage build | | General | Bug reports, feature ideas, UI improvements |

If you own a desktop device and can test the app — please do! Even a simple "works" or "crashes on X" issue report is incredibly valuable.

# Fork → Clone → Branch → Code → Push → PR
git checkout -b fix/windows-model-loading
git commit -m "fix: resolve model path on Windows"
git push origin fix/windows-model-loading
Open a Pull Request — all sizes welcome

---

License

Licensed under the MIT License — free to use, modify, and distribute. See LICENSE for full details.

---