Skip to main content

AI Keys

Nlarj never pays for AI — your church brings its own key (BYOK). API keys are stored on-device only and used to call the provider directly from the app.

Open Settings → AI Settings.

Supported providers​

The app supports exactly two providers:

ProviderWhere to get the keyWhat it unlocks
OpenRouteropenrouter.ai/settings/keysAccess to a large catalog of chat models (Gemini, DeepSeek, Llama, Qwen, Mistral, and more) via a single key
Google AI Studioaistudio.google.com/apikeyGemini chat models and Imagen image generation

You can connect one, both, or neither.

Connect OpenRouter​

Two ways to connect:

Option A — Paste your API key​

  1. AI Settings → Enter API Key
  2. Paste a key that starts with sk-or-
  3. Tap Connect

If the key is valid, the card switches to the connected state and the model list loads below.

Option B — Browser sign-in​

  1. AI Settings → Sign In
  2. The app opens OpenRouter in your browser
  3. Approve, and it returns a key automatically

Pick a chat model​

After connecting OpenRouter:

  1. Expand AI Model → OpenRouter Model
  2. Use the search box to filter by model name, provider, or ID
  3. Tap a model to select it

Models marked FREE have a free tier at OpenRouter. Models marked ★ TOP PICK are curated as good defaults for church / sermon work. The top-pick list is fetched from OpenRouter's remote config when you open the screen; if that fetch fails, the app falls back to a built-in curated list of Gemma, Gemini 2.0 Flash, DeepSeek Chat, Llama 4, Qwen 2.5, and Mistral Small.

Each row shows the provider name and the model's context window size. Selecting a model writes it immediately; there's no separate save button.

Connect Google AI Studio​

  1. AI Settings → Google AI Studio → Enter API Key
  2. Paste a key starting with AIzaSy
  3. Tap Connect

The app validates the key against Google's API before storing it. Once connected, Gemini and Imagen models become available for the features that use them.

Image generation preference​

AI Settings → Image Generation → Preferred Provider lets you pick which service generates images:

  • Imagen (Google) — requires Google AI Studio connected
  • DALL-E (OpenRouter) — requires OpenRouter connected

Allow Fallback — when on, if your preferred provider fails or isn't configured, the app tries the other one. When off, a failure in the preferred provider surfaces as an error.

Refresh model list​

Tap the refresh icon in the top bar of AI Settings to re-fetch the OpenRouter catalog. Use this after new models are released or if the list looks stale.

Disconnect​

Each provider card has a Disconnect button in its connected state.

  • OpenRouter Disconnect — removes the stored API key from this device and clears the selected model. AI features that require OpenRouter stop working until you reconnect.
  • Google AI Studio Disconnect — removes the stored Google key. Imagen image generation and Gemini chat features stop until you reconnect.

You can reconnect any time; disconnecting doesn't affect anything at the provider side.

Security notes​

  • Keys live in secure device storage only. They do not travel to Nlarj servers.
  • If you suspect a key is compromised, rotate it at the provider first, then paste the new key into the app.
  • Never share your key — not in email, not in screenshots, not in support tickets.

What isn't in the app​

Some things often associated with BYOK workflows are not built into the AI Settings screen today:

  • Per-feature model overrides (one model for sermons, another for chat) — the selected OpenRouter model is used globally
  • In-app usage logs or spend dashboards — check usage at your provider's console
  • In-app spend caps — set those at OpenRouter / Google AI Studio instead
  • Automatic key-rotation reminders
  • Providers other than OpenRouter and Google AI Studio

If you need any of those, handle them at the provider side, or contact support to discuss.