An Android chat client for connecting to your local LM Studio server.
| Feature | Details |
|---|---|
| Chat interface | Clean chat-bubble UI with user (right) and assistant (left) messages |
| Model selector | Lists all models loaded in your LM Studio instance via the /v1/models endpoint |
| WiFi-only mode | Optionally block requests over mobile data, preventing accidental external exposure |
| HTTPS / TLS | HTTPS is the default; HTTP cleartext requires an explicit opt-in per host |
| Self-signed cert support | Optional trust-all mode for servers with self-signed or custom TLS certificates |
| API key auth | Store your LM Studio API key; it is persisted with Android Keystore-backed EncryptedSharedPreferences |
| System prompt | Configurable system prompt sent at the start of every conversation |
| Temperature | Adjustable sampling temperature (0.0 – 2.0) via a SeekBar |
| Connection test | One-tap test to verify connectivity before saving settings |
| Verbose logging | Optional HTTP body logging for debugging (ProGuard-stripped in release builds) |
- Android 7.0+ (API 24+)
- LM Studio >= 0.2 running on a machine reachable from the phone (e.g. same WiFi network)
- LM Studio server started with Enable CORS enabled (or appropriate network settings)
- JDK 17
- Android SDK (API 34 build tools included)
./gradlew assembleDebugThe resulting APK is at app/build/outputs/apk/debug/app-debug.apk.
./gradlew testapp/src/main/java/com/xrmatic/lmstudio/
├── MainActivity.kt # Launcher: routes to Settings or Chat
├── ChatActivity.kt # Main chat screen
├── SettingsActivity.kt # Connection & security settings
├── api/
│ ├── LMStudioApiService.kt # Retrofit interface (OpenAI-compatible API)
│ └── LMStudioClient.kt # OkHttp / Retrofit builder with security options
├── adapter/
│ └── MessageAdapter.kt # RecyclerView adapter for chat bubbles
├── model/
│ └── ApiModels.kt # Data classes for API request/response and UI
├── prefs/
│ └── AppPreferences.kt # Encrypted + plain SharedPreferences wrapper
├── util/
│ └── NetworkUtils.kt # WiFi / network state helpers
└── viewmodel/
└── ChatViewModel.kt # MVVM ViewModel (Coroutines + LiveData)
- On first launch you are taken to the Settings screen.
- Enter the server URL, e.g.
http://192.168.1.100:1234(LM Studio default port is 1234). - If you have API authentication enabled in LM Studio, enter your API key.
- Toggle WiFi only to prevent requests over mobile data.
- If your server uses a self-signed TLS certificate, toggle Accept self-signed certificates.
- Tap Test connection to verify, then Save & Connect.
Security note: HTTP (cleartext) is shown with a warning banner. Use HTTPS wherever possible. The "Accept self-signed certificates" mode disables hostname verification -- use it only on trusted local networks.
LM Studio exposes an OpenAI-compatible REST API. The app uses:
| Endpoint | Purpose |
|---|---|
GET /v1/models | List loaded models |
POST /v1/chat/completions | Send a chat message and receive the assistant reply |
MIT