Configuration

Settings overview

Access Settings from the gear icon in the sidebar or the system tray menu. Settings are organized into categories.

AI models

Default model

Gearbox Relay ships with Qwen 3.5 0.8B (Q4_K_M quantization) as the default enrichment model. This ~500 MB model runs via llama-cpp-2 on desktop and provides summaries and tags for your captures.

Model manifest

The model manifest at src-tauri/models/manifest.json defines available models:

{
  "version": 1,
  "default": "qwen-3.5-0.8b",
  "models": {
    "qwen-3.5-0.8b": {
      "display_name": "Qwen 3.5 0.8B Q4_K_M",
      "family": "qwen",
      "url": "https://huggingface.co/Ares-Realm-Studios/...",
      "filename": "qwen3.5-0.8b-q4_k_m.gguf",
      "size_bytes": 524288000,
      "min_ram_mb": 500,
      "context_length": 4096
    }
  }
}

Adding a custom model

1. Download a GGUF model that fits your hardware
2. Place the .gguf file in your models directory
3. Add an entry to the manifest with the model's metadata
4. Restart Gearbox Relay

Swapping models

If you have multiple models configured, you can switch between them in Settings → AI Models. The active model handles all new captures. Existing captures retain their existing enrichment data.

Mobile AI runtimes

On mobile, the AI runtime is platform-specific:

iOS — Uses Apple's MLX framework for efficient on-device inference. The iOS app downloads the Qwen 3.5 0.8B model in a 4-bit quantized format optimized for Apple Silicon.

Android — Uses Google's MediaPipe for on-device AI. The MediaPipe LLM Inference API handles model loading and inference on Android devices.

Sync settings

Enabling sync

Sync is disabled by default. To enable:

1. Go to Settings → Sync
2. Enter your sync server URL
3. Sign in with your account credentials
4. Choose sync options

Sync behavior

Automatic sync — Sync runs automatically in the background when changes occur. You can also trigger a manual sync from the sync status indicator.

Conflict resolution — Gearbox Relay uses LWW (last-writer-wins) conflict resolution. When the same capture is modified on two devices, the version with the higher last_modified timestamp wins.

Pause sync — You can pause sync from the system tray. Paused sync means changes are stored locally but not transmitted to other devices.

Device management

View connected devices in Settings → Sync → Devices. Each device shows:

• Device name and type
• Last sync time
• Sync status (active, paused, offline)

Remove a device to stop syncing from that device. Removing a device does not delete local data.

Telemetry

Crash reporting

Opt-in crash reporting via Sentry is disabled by default. When enabled, crash reports include:

• Stack traces from Gearbox Relay
• Performance metrics
• Device information (OS, memory)

No capture content, text, or personal data is included. All PII is scrubbed from enabled reports.

Analytics

Usage analytics are never collected. We do not track how you use Gearbox Relay.

Preferences

Clipboard watcher

Enable/disable — Toggle the clipboard watcher on or off. When disabled, copied text is not automatically captured.

Pause duration — Option to pause clipboard monitoring for a set period (1 hour, 4 hours, until restarted).

Notifications

Configure desktop notifications for:

• New capture confirmations
• Sync status changes
• AI enrichment completion

Storage

View storage usage:

• Total captures stored
• Sync storage used (vs. your plan limit)
• AI model size
• Database size

Clear local cache to free storage space. This removes cached AI responses but not your captures.

Keyboard shortcuts