Appearance
FeynmanLM vs NotebookLM
NotebookLM is Google's source-grounded AI assistant: upload documents into a notebook, chat with them, and generate impressive artifacts — most famously the podcast-style Audio Overviews. It's polished, generous at the free tier, and backed by Gemini. But it's built for generating study material, not for learning over time — and everything you put in lives on Google's servers, processed only by Google's models.
At a glance
| FeynmanLM | NotebookLM | |
|---|---|---|
| Price | Free to download — no purchase, no subscription | Free tier (capped); higher limits via Google AI bundles: $7.99/mo Plus, $19.99/mo Pro, $99.99+/mo Ultra |
| AI costs | Pay-as-you-go from a prepaid balance ($5 free to start): any model in Chat at provider API price +15%, or 1¢ per tool call via an AI subscription you already pay for | Included, but capped per day/tier (chats, overviews, quizzes) |
| Platforms | Native macOS app (macOS 14+) | Web + iOS + Android apps; no desktop app |
| Data storage | Local SQLite + files in your private iCloud container | Google's servers — every source must be uploaded |
| AI models | Any MCP-compatible AI (Claude, ChatGPT, Gemini, Grok), or any API provider in-app | Gemini only, no model picker, no bring-your-own-key |
| Source ingestion | Automatic: Safari reading list, Apple Podcasts, PDFs, papers, books, YouTube, X posts | Manual upload per notebook (plus Drive links and Deep Research discovery) |
| Podcasts | Follow shows, full episode transcripts as sources | Not supported (it generates podcast-style audio; it can't ingest your podcast feed) |
| Knowledge accumulation | One persistent library — sources accumulate for life | Isolated notebooks with per-notebook source caps (50–600 by tier) |
| Retention | Weekly Schedule + Feynman-technique review with saved session history | One-shot flashcards/quizzes — no spaced repetition, no tracking over time |
| Export | Files stay in your iCloud Drive; SQLite you can open | Notes to Google Docs/Sheets, audio as WAV; notebook structure and chats effectively locked in |
Pricing and limits verified July 2026 — Google reshuffled NotebookLM tiers twice in the past year; check notebooklm.google/plans for current numbers.
Pricing: free-but-capped bundles vs free-and-metered
NotebookLM itself is free, with daily caps (50 chats/day, 3 audio overviews/day, 50 sources per notebook). Lifting the caps means a Google AI subscription — $240/yr for Pro-level limits — that bundles much more than NotebookLM, and the caps still exist, just higher. You're paying a flat annual fee for permission to do more, not for what you did.
FeynmanLM is free to download and has no daily caps of its own. There's no plan to upgrade to; you top up a prepaid balance and it draws down as you use AI, starting with $5 free. In the in-app Chat you pick any model and pay that provider's own API list price plus a 15% margin, metered on the exact cost the provider reports. Or connect the Claude, ChatGPT, Gemini, or Grok subscription you already pay for over MCP and it's 1¢ per tool call — that subscription covers the inference, the cent covers reading and writing your library, and a daily spend cap keeps things bounded. A little also goes to background work like transcript cleanup and search embeddings, billed the same way. Every conversation's cost is shown to the cent.
Where NotebookLM is better
- Audio and Video Overviews. The generated podcast-style discussions are best-in-class and genuinely delightful. FeynmanLM's Read Aloud reads you the actual source; it doesn't generate a two-host show about it.
- Genuinely free AI, forever. A Google account is the only requirement, and within the daily caps the free tier costs nothing indefinitely. FeynmanLM is free to install with $5 of AI included, but sustained use draws on a balance you top up.
- Cross-platform. Web, iOS, and Android apps. FeynmanLM is Mac-only today.
- Deep Research. NotebookLM can go find sources on the web for you; FeynmanLM tracks what you're already consuming.
- Google-scale polish and Drive/Docs integration if you live in Workspace.
Where FeynmanLM is better
- Your data never leaves your control. FeynmanLM's library is a local SQLite database plus files in your private iCloud container, synced through your Apple account. NotebookLM requires uploading everything to Google, with no local copy and a weak export story.
- Any AI, not one AI. FeynmanLM works with Claude, ChatGPT, Gemini, Grok, or any API provider — use the subscription you already pay for, switch anytime. NotebookLM is Gemini-only with no model visibility at all.
- A learning system, not a generator. NotebookLM produces flashcards and quizzes as one-shot artifacts — nothing schedules review, nothing tracks what you've mastered. FeynmanLM plans your week, runs Feynman-technique sessions where you explain sources back and the AI probes your gaps, and keeps the full review history per source. See Why Quizzing?
- One library for life. NotebookLM silos knowledge into notebooks capped at 50–600 sources. FeynmanLM is a single accumulating knowledge base — five years of reading, listening, and studying in one searchable place.
- Real podcast support. Follow your shows, get full transcripts, read/listen/review. NotebookLM can't subscribe to a podcast feed at all.
- Sources flow in automatically from your Safari reading list, Apple Podcasts, iCloud folders, and reference managers — no per-notebook uploading ritual.
Which should you choose?
Choose NotebookLM for ad-hoc research sprints: dump twenty documents into a notebook, chat with them, generate an Audio Overview for the commute — especially if you're not on a Mac, or you'd rather have AI that stays free indefinitely inside its caps than AI you meter.
Choose FeynmanLM if learning is ongoing, not one-off: you want everything you read, watch, and hear accumulating in a library you own, reviewed on a schedule, with proof you actually understood it — using whichever AI you already trust.