NotebookLM alternatives, now that it is called Gemini Notebook

Google renamed NotebookLM to Gemini Notebook on July 16, 2026. If you searched for one name and found the other, it is the same product. The announcement says it "remains a standalone product", and it now lives at notebook.google.

I build Understand, a voice tutor for adults, and it is one of the alternatives below. Gemini Notebook beats it at several things, and I list them near the end.

Every price and limit here comes from the vendor's own page, read on October 3, 2026, unless I name another source.

The short version

Changes that came with the name

Notebooks now appear in the Gemini app, and new sources, renames and custom instructions sync both ways. They also appear in AI Mode in Google Search, in English and outside the European Economic Area. Inside Gemini Notebook, answers are "grounded exclusively in your notebook sources". The same notebook opened in the Gemini app or in AI Mode "may also include web search and other tools".

The announcement also gave each notebook "a secure cloud computer" that runs code for data analysis, starting with AI Ultra subscribers.

The core is unchanged. You add sources, ask questions, and get answers with inline citations that open the quoted passage. A source can be a PDF, a web page, a YouTube video, an audio file, a Google Doc, an EPUB or an image, up to 500,000 words or 200 MB each.

Audio Overviews

An Audio Overview is a podcast generated from your sources. Google describes "deep-dive discussions between AI hosts that provide in-depth summaries of the key topics in your uploaded sources". Deep Dive, the default, has two hosts. The Brief is one speaker in under two minutes. The Critique and The Debate are the other formats. Output comes in more than 80 languages. You can ask for Shorter, Default or Longer in English only, and the help page gives no length in minutes for any of them.

Interactive mode

Interactive mode lets you talk to the hosts. You generate an overview, select Interactive mode, and press Join while it plays. In Google's words: "When the hosts call on you, ask your questions." They answer from your sources and then "resume the original Audio Overview".

This is a good feature. I get to ask at the moment I am confused, which a podcast has never allowed.

It is English only. It works only on newly generated overviews. People who open your shared link hear the overview and cannot join it. Google says your voice and the transcript of your questions are not stored. The mobile app page lists Interactive Mode as "Beta, English Only".

You still generate an overview first, and that overview is what resumes after your question. The hosts discuss a summary of your sources. Nothing on Google's pages describes a way to have them read a source to you word for word. The pages also do not say how many follow-up questions the hosts take before they resume, or how far an answer can go past the overview's script.

Live voice chat

On September 15, 2026 Google announced real-time voice conversation with a notebook, with no overview involved. The chat help page says "every answer is grounded in your uploaded sources, and you can interrupt at any moment". A transcript appears in the chat.

It exists only in the mobile app, for users 18 and over. The announcement says it works in nearly 100 languages and was rolling out to AI Ultra subscribers that week, with "Google AI Pro and others soon". If you pay nothing, look for the audio button in your app before you plan around it.

Video, mind maps, flashcards and the Learning Guide

Video Overviews come in Cinematic, Explainer and Short formats. Short runs about 60 seconds, and Google warns that a longer video can take more than 30 minutes to generate.

Mind Maps draw your sources as a branching diagram whose nodes you can select to ask about. The mobile app does not have them.

Flashcards and quizzes come in easy, medium and hard. A flashcard deck remembers which cards you missed and lets you practise only those.

The Learning Guide is a chat setting under Configure Chat. Google calls it "best for educational content, helping you grasp new concepts and skills efficiently".

Plans and limits

Gemini Notebook is free with a Google account. The plans page gives 100 notebooks and 50 sources per notebook without a plan. AI Plus gets 200 and 100, AI Pro 500 and 300, and the two AI Ultra tiers 500 or 600 sources. Google's AI plans cost $4.99 a month for Plus, $19.99 for Pro, and $99.99 or $199.99 for Ultra.

Limits used to refresh daily. Since September 2, usage is a single compute budget that "refreshes every 5 hours until you reach your weekly limit". AI Plus doubles it and AI Pro quadruples it. Google publishes no count of overviews or chat messages for any tier.

On models, the plans page says accounts without a plan and Plus accounts get "access to Gemini models", Pro gets "higher" access and Ultra "highest". It names no model for any tier.

Open-source and self-hosted alternatives

Open Notebook has about 39,700 GitHub stars and an MIT licence. It runs in Docker, supports more than 18 model providers including local models through Ollama, and generates podcasts with one to four speakers. You run the server and supply the model.

SurfSense is a free desktop app that indexes your documents on your own machine. Answers cite their sources, and a two-host podcast is voiced offline. Its own README says "NotebookLM wins on audio quality and it has video". A paid licence adds plugins and priority support.

Open NotebookLM turns one PDF into an MP3 dialogue and does nothing else. Its last commit is from December 2024. Podcastfy is a Python package for developers that makes audio conversations from websites, PDFs and YouTube videos.

Research tools

Elicit searches "more than 138 million papers" and lets you chat with them on its free plan. Pro, at $49 a month billed yearly, adds a systematic review workflow that screens 5,000 papers.

Perplexity Spaces are now called Projects. Each holds up to 5,000 files and 2.5 GB on every plan. Pro is $17 a month billed yearly.

ChatGPT Projects keep chats, files and instructions together: 5 files per project on Free, 25 on Go and Plus, 40 on Pro. Claude Projects are on every plan, with five projects for free accounts, and paid plans expand a project's knowledge by up to ten times through retrieval.

These are general assistants with a folder attached. They answer from their training when your files run out. Gemini Notebook's chat replies that it "can't answer this question".

Study tools

YouLearn chats with "your YouTube video, PDF, slides, websites, files and lectures". Its free plan allows three uploads and five chat messages a day, and Pro is $20 a month or $12 billed yearly. I tested it in my YouLearn review.

Mindgrasp makes notes, flashcards and quizzes from uploads. Its pricing page returned a 404 today, and a Fast.io review says it has no free tier, only a four-day trial that requires a card.

Quizlet is for drilling flashcards. Its site blocked my request. The App Store listing shows Quizlet Plus purchases at $9.99 and $44.99.

Note tools with AI

Obsidian is free, and your notes are files on your own disk. Plugins add the AI. Smart Connections shows related notes beside the one you are writing, using local embeddings and no API key.

Notion puts its Agent on the Business plan at $20 per member a month. Free and Plus get a trial.

Readwise Reader has Global Ghostreader, which searches your whole library and "answers with citations that link to the source", on web and desktop. Reader also reads documents aloud. The subscription is $12.99 a month.

Audio-first tools

ElevenReader narrates your documents, and its GenFM feature turns a PDF, article or ebook into a podcast with AI co-hosts. The free plan includes 10 hours of audio a month. Ultra is $11 a month.

Huxe, started by three former NotebookLM developers, generated podcasts from a prompt. It announced its shutdown on May 22, 2026. The other audio tools are in my post on Audible alternatives.

A paraphrase, and no record of the learner

If I had forty PDFs and an exam on Friday, I would use Gemini Notebook and nothing else on this page. It was built to synthesise many documents and cite each claim.

Learning something for good is a different job. The unit of an Audio Overview is the paraphrase. The podcast is generated, then listened to. Interactive mode lets me ask a question, and the answer is the hosts' account of the source, after which their script continues. For a proof or a dense paragraph of philosophy, I want the author's sentences. A summary keeps the conclusions and drops the steps I might have rejected.

Google documents no record of the learner that outlives a notebook. A flashcard deck remembers the cards I missed. You can opt in to share Gemini chats as context across notebooks. I found no page describing a list of the ideas I have explained correctly, in a form I could read or correct.

David Deutsch, following Popper, argues that knowledge grows by conjecture and criticism. An explanation does not pour from a podcast into a listener. I hear the words and guess what they mean. I learn whether the guess was right only when I say it back and someone pushes on it. Google's pages describe the voice chat answering my questions. They do not describe it asking me to explain.

Understand

Understand reads an article or PDF aloud word for word, one sentence at a time, with the current sentence highlighted. When you start speaking, it stops. You ask why, the tutor answers, and the reading resumes at the sentence where you interrupted. Commands such as "skip" and "go back" work by voice.

Material comes from a saved list. A bookmark button, a Chrome extension, the iOS share sheet and your X bookmarks all save to it, and you can upload a PDF of up to 25 MB. Each subject gets its own tutor with its own transcript and board.

The tutor asks you to put an idea in your own words. Your answer is written to a knowledge graph, an outline of concepts you can open, rename, merge or correct. A concept counts as something to build on after two successes in two different contexts. Hearing a document read writes nothing to it. The outline spans every subject, so a tutor on one topic knows what you showed in another.

Understand cannot query fifty sources at once and synthesise them. It has no citations panel, no video overviews and no mind maps. There is no Android app; it runs on the web and on iOS through TestFlight. A PDF must contain real text, because scans are refused and figures are ignored. It is free for now.

Reasons to stay with Gemini Notebook

Stay if your problem is volume, such as a semester of readings or a folder of contracts. Understand has no citation behind each answer and no Android app. Gemini Notebook also accepts lecture recordings and photos of handwritten notes, and Understand takes neither.

Add Understand when the goal is one text understood well enough to explain without it.

Questions people ask

Is NotebookLM the same as Gemini Notebook?

Yes. Google renamed it on July 16, 2026 and calls it "the same standalone product".

Can I talk to a NotebookLM podcast?

Yes, in English. Turn on Interactive mode for a new Audio Overview and press Join. The hosts answer from your sources and then resume.