Frequently Asked Questions

Lilmod is a workspace built for studying Torah texts, built using Sefaria's library and enhanced with note taking and AI tools.

Most AI Torah tools are primarily chat bots. They are designed for users to expect the AI to do most of the work. This can be useful if you want AI to look something up for you, but doesn't naturally integrate into a study session. Lilmod is a workspace where users can study by themselves and use AI tools to assist them.

There's also a technical difference. Most AI Torah chatbots rely on a method called RAG (retrieval-augmented generation), which lets them answer questions using a fixed set of sources chosen in advance by the developers. If your question touches on material outside those sources, the chatbot won't be much help. Lilmod's chat feature works differently: it automatically loads whatever texts you're currently studying, so if your question can be answered from those texts themselves, Lilmod can answer it - no pre-selected source list required.

Sefaria is an extraordinary library, and Lilmod isn't trying to replace it - we offer a different user experience for interacting with texts that live in the Sefaria library. Our multi-window workspace makes it easy to switch between texts and study several in parallel, arranged in layouts you customize. That means no exiting one text to open another, and no juggling browser tabs.

On top of that, we layer in AI tools - like instant translation and word definitions - to augment your study session as you go.

You can, and plenty of people do - but Lilmod is built to make working with AI while you learn easier. Generating a translation or a definition is one click away, with no need to write a prompt or copy and paste context by hand. The same goes for Lilmod's chat windows: the AI automatically has everything it needs to answer your question, pulled directly from the text in front of you.

ChatGPT can approximate this - it will sometimes search the web and pull primarily source content from Sefaria - but it may just as easily pull from other sources across the web, including secondary sources that might mischaracterize the original text. By feeding models the direct source text and nothing else, Lilmod's chat windows often let the AI answer questions about that text more accurately.

There are three features: AI translate, AI define for selected words or phrases, and AI chat with context from the texts you have open.

For translate and define, Lilmod feeds the AI both the selected text and its broader context. For commentary texts, that includes the base text being commented on - which lets the AI produce more accurate translations and definitions than it would from the selected snippet alone. The exact prompt and model used for each feature are customizable.

Chat windows are available to paying users; translate and define are available to everyone.

AI hallucination is a real problem, even with very capable models. Lilmod addresses this in two ways. First, the AI in chat windows always has direct access to the texts being discussed, rather than working from memory alone (hallucinations are most common when the model is working from memory). Second, our default system prompt instructs the AI to flag its confidence level in each response: information drawn directly from the text it can see gets one tier, while information it's fairly confident about but can't directly verify against the text gets another.

This lets users gauge how much hallucination risk is baked into any given answer.

That said, these are mitigations, not guarantees - they reduce the risk but don't eliminate it. As with any AI tool, verifying answers yourself, or with someone more qualified, remains best practice.

Free use includes a weekly allowance for translate and define. Pro unlocks stronger models plus chat; BYOK (bring your own key) gives you the same features using your own API keys, with usage billed directly by your provider.

We tested many low-cost models and selected the ones that performed best specifically at translating commentaries on Tanach and Gemara, then made those available for free.

Pro ($20/mo): Everything just works out of the box - better translations and definitions, plus chat with the texts you have open, all included in your subscription. Pro comes with weekly quotas that are enough for most people's study routines.

BYOK, or "Bring Your Own Key" ($7/mo): The same features as Pro, but instead of using Lilmod's AI access, you connect your own. This requires getting an "API key" - a password-like code from an AI company (like OpenAI or Anthropic) that lets an app use their AI on your behalf, billed to you directly for what you use. Because you're paying the AI provider yourself instead of paying Lilmod a flat fee, BYOK costs less - and since there's no quota built in, it can be effectively unlimited if you're willing to pay for the usage. Your key stays private in your browser - Lilmod never sees or stores it.

If getting an API key sounds like more setup than you want, Pro is the simpler choice - it's ready to use immediately with no extra accounts required.

Primary texts and most translations come from Sefaria, and their licenses and attributions remain intact. If a text is on Sefaria, it's available on Lilmod; if it isn't, it isn't.

Yes, as a web app with a focused mobile layout rather than the full multi-window workspace. You can access your notes, highlights, and bookmarks, and use the AI translate and define features.

You can manage billing from your account settings. Cancel anytime - access continues through the end of your current paid period.

For support email us at support@lilmod.app.

For feedback and suggestions email us at feedback@lilmod.app.