AI research over the actual text
An assistant that reasons over scripture and the original-language lexicons — not a generic chatbot guessing.
Why bring your own key?
- Your choice of model. OpenRouter lets you pick the model — Claude, GPT-4o, and more — for the same grounded study assistant.
- Private. Your questions go straight from the app to OpenRouter. There's no Scribe server in the middle, and no account.
- No markup. You pay OpenRouter directly for usage — typically cents per study session — with no subscription on top.
- Optional. Reading, word study, and search are fully offline and never depend on the AI or a network.
What it's grounded in
The assistant is given the passage you're studying, the relevant Strong's entries and LSJ/BDB definitions, and the cross-references — so answers stay anchored to the text rather than free-floating.
Which models can I plug in?
Paste any of these IDs into Settings → AI Research. All seven run the full multi-step agentic research:
| Model | Plug-in ID | |
|---|---|---|
| Claude Sonnet 4.6 | anthropic/claude-sonnet-4.6 | Default · balanced |
| Claude Opus 4.1 | anthropic/claude-opus-4.1 | Most capable |
| GPT-4o | openai/gpt-4o | |
| GPT-4o mini | openai/gpt-4o-mini | Cheapest |
| Gemini 2.5 Pro | google/gemini-2.5-pro | |
| Gemini 2.5 Flash | google/gemini-2.5-flash | Fast |
| DeepSeek V3.1 | deepseek/deepseek-chat-v3.1 | Budget |
Any other model from openrouter.ai/models works too — it just runs as a single grounded request rather than the multi-step research.
Setting it up
- Unlock Scholar (the one-time $29.99 purchase).
- Create an OpenRouter API key at openrouter.ai.
- Paste the key into Scribe → Settings → AI Research, and choose a model.
- Ask questions — the assistant answers grounded in the text, lexicons, and cross-references.