Configuration
Sythoria works out of the box, but offers several customization options to tailor the experience to your workflow.
Model selection & health checks
Choose your active model from the dropdown in the chat input bar. Available models are the ones you've configured in Settings. To add or remove models:
- Open Settings from the sidebar
- Add models with the Add Model button, or delete existing ones with the trash icon
- Each model shows a live connection indicator.
Connection indicators
- Green Dot: Connected (the model endpoint is active and responsive).
- Pulsing Yellow Dot: Connecting (a connection check is in progress).
- Red Dot: Error (connection check failed or request was refused).
- Gray Dot: Disconnected or disabled.
Health check mechanism
Sythoria automatically performs connection health checks in the background to ensure your models are ready to respond:
- Interval: Checks are run every 5 minutes.
- Throttling: Consecutively triggered manual or automatic checks are rate-limited to once every 30 seconds to avoid provider API abuse.
- Background activity: You can toggle background checks under Settings. If Disable Background Activity is enabled, all background health checks stop immediately, and indicators return to gray.
See Multi-Provider Chat for details on managing multiple providers.
Temperature
Control the randomness of responses via the Temperature slider in Settings:
| Value | Behavior | Best for |
|---|---|---|
| 0.0 | Deterministic, focused outputs | Factual tasks, data extraction, coding |
| 0.7 | Balanced creativity (default) | General conversation, writing, analysis |
| 2.0 | Maximum randomness | Brainstorming, creative writing, variety |
Lower temperatures produce more consistent, predictable responses. Higher temperatures introduce more variation and creativity.
Web search
Sythoria can augment chat responses with live web search results. When enabled, the AI can search the web and read web pages to provide up-to-date information.
Supported search providers
| Provider | Requires | Best for |
|---|---|---|
| Google Custom Search | API key + CX ID | High-quality results, widely available |
| SearXNG | Base URL only | Self-hosted, privacy-preserving |
| Firecrawl | API key | Web scraping and search combined |
| Custom | Varies | Any OpenAI-compatible search API |
Setting up web search
- Open Settings and scroll to Web Search APIs
- Click Add Search API
- Choose a provider and enter your credentials
- Enable the search API with the toggle switch
- Use the search toggle in the chat interface to enable/disable search per conversation
Keyboard shortcuts
| Shortcut | Action |
|---|---|
Enter | Send message |
Shift+Enter | Insert new line |
Escape | Stop streaming |
⌘K / Ctrl+K | Search (in docs) |
Data management
All data is stored locally within the desktop application. You can:
| Action | How |
|---|---|
| Export a conversation | Click the export button in the sidebar conversation menu — saves as Markdown |
| Clear all conversations | Use Clear All Chats in the sidebar, or clear application data |
| Delete a single conversation | Click the trash icon on the conversation in the sidebar |
Resetting the application data resets everything — conversations, API keys, and settings.
Appearance
Toggle between light and dark mode in Settings under Appearance. Your preference is saved automatically and persists across sessions.
Provider configuration
Each model in Settings has these fields:
| Field | Description |
|---|---|
| Name | A display name for your reference (e.g., "Work GPT-4o") |
| Provider Preset | Auto-fills the API base URL and default model |
| API Base URL | The endpoint for API calls (auto-filled by presets) |
| Model ID | The model identifier to use (e.g., gpt-4o, claude-sonnet-4-20250514) |
| API Key | Your authentication key (stored securely in your OS keychain) |
See the Providers section for provider-specific setup guides.
