Skip to main content

BYOK with OpenRouter

Nlarj uses Bring Your Own Key (BYOK) for all AI features. Your church pays the AI provider directly — we never bill you for AI usage.

Why BYOK?​

  • You control costs — set spend limits, pick any model, switch any time
  • You control privacy — use whichever provider you trust
  • No markup — you pay provider cost, not our rate
  • Scale freely — no per-seat AI fees from us

What providers are supported?​

In Settings → AI Settings, you can connect:

  • OpenRouter — one key for 100+ models from Anthropic, OpenAI, Google, Mistral, Meta, DeepSeek, Qwen, and more
  • Google AI Studio — direct access to Gemini (text) and Imagen (image generation)

You can connect one or both. If both are connected, you can pick which provider handles image generation in the Image Generation section of AI Settings.

Why OpenRouter?​

OpenRouter is a single API that routes to 100+ AI models. Benefits:

  • One key for all providers
  • Switch models without changing credentials
  • Transparent pricing — see exact cost per model
  • No minimum spend — pay only for what you use
  • Free-tier models available for small churches

Setup — OpenRouter​

1. Create an OpenRouter account​

  1. Go to openrouter.ai
  2. Sign up (GitHub, Google, or email)
  3. Add a payment method and some credits

2. Add credits​

  • Minimum: $5 — enough for dozens of sermons on free-tier or cheap models
  • Recommended: $20/month
  • You can also use 100% free models (look for the green FREE badge in the model list) and pay nothing

3. Connect in Church Admin​

Two ways to connect — both in Settings → AI Settings → OpenRouter:

  • Sign In (recommended) — opens your browser, you authorize Nlarj, and the key flows back automatically
  • Enter API Key — paste an existing key from openrouter.ai/settings/keys. The key must start with sk-or-.
Keep your key secret

Treat it like a password. Anyone with this key can spend your credits.

4. Pick a model​

After you connect, expand the AI Model section and browse the model list. You can:

  • Search — type to filter by name, ID, or provider
  • Spot free models — green FREE badge
  • Spot top picks — gold ★ TOP PICK badge marks models we've tested and recommend for sermon work (currently Gemma 3, Gemini 2.0 Flash, DeepSeek, Llama 4, Qwen 2.5, and Mistral Small)

Tap any model to select it. Your choice is saved instantly and used for every generation going forward.

Setup — Google AI Studio​

If you want to use Gemini directly or generate images with Imagen:

  1. Get an API key at aistudio.google.com/apikey
  2. In Church Admin → Settings → AI Settings → Google AI Studio
  3. Tap Enter API Key, paste your key (starts with AIzaSy...), tap Connect

Once connected, Google AI Studio powers image generation by default (Imagen) and can also serve Gemini models for text.

Image provider preference​

Under Image Generation in AI Settings you can choose:

  • Imagen (Google) — high-quality, Google's flagship image model
  • DALL-E (via OpenRouter) — OpenAI's image model

There's also an Allow Fallback toggle — if your preferred provider fails, try the other one automatically.

Spend controls​

OpenRouter lets you cap spending at the key level — the cleanest way to enforce a budget:

  1. OpenRouter dashboard → Keys → Edit
  2. Set Credit limit or monthly cap
  3. OpenRouter blocks requests once you hit the limit

Full usage history is available in the OpenRouter dashboard.

Security​

  • Your API keys are stored securely on your device
  • Keys never leave the device except to call OpenRouter or Google directly
  • Logs never contain the full key

See Phishing & Security for key hygiene.

Disconnecting​

Any time you want to revoke access on a device:

  • Settings → AI Settings → Disconnect next to either provider
  • The key is wiped from the device immediately
  • AI features stop working on that device until you reconnect

Troubleshooting​

"Invalid API key" — Copy the key fresh from the provider. Watch for trailing spaces. OpenRouter keys start with sk-or-; Google AI keys start with AIzaSy.

"Insufficient credits" — Check your OpenRouter balance. Add credits or pick a free-tier model.

"Model not available" — The model you selected may have been removed. Open the model list and pick a current one.

Image generation fails — Check both the preferred provider and (if enabled) the fallback provider are set up. See the error dialog for specifics.