What Is Google NotebookLM? (2026 Guide)

What Is Google NotebookLM?

If you went looking for NotebookLM recently and found something called Gemini Notebook instead, nothing has been discontinued. Google renamed the product on 16 July 2026, and the thing you were looking for is still there, still free to start, and still built around the same idea.

That idea is worth understanding, because it is genuinely different from how ChatGPT, Claude, or Gemini's own chat interface work. Those tools answer from everything they know. NotebookLM answers only from the documents you give it — and tells you which one each sentence came from.

This guide covers what it does, how the source grounding actually works, what the limits are on each plan, and the cases where a general assistant is the better choice.

What Is Google NotebookLM?

Google NotebookLM — now officially Gemini Notebook — is a research tool that answers questions about documents you upload. You create a notebook, add sources to it, and then ask questions. The model reads only those sources and cites them in every answer.

It launched in 2023 as an experiment called Project Tailwind and has grown into one of Google's most widely used AI products, with more than 30 million users and over 600,000 organisations on it at the time of the rename.

The short version of why people like it: it does not make things up about your material. Because it is confined to the sources in the notebook, it cannot pull in a half-remembered fact from its training data and present it as something your document said. When it does not know, it says the sources do not cover it.

Is your content being cited by AI tools, or ignored by them? Get a free SEO audit at https://www.seo-stuff.com/free-audit and find out where you stand.

Why the Name Changed and What It Means for You

The rename caused more confusion than it needed to, so it is worth being precise about what did and did not change.

Same product, same notebooks

Google described Gemini Notebook as "the same standalone product". Existing notebooks, sources, and saved notes carried over. Nothing needed migrating and nothing was deleted. If you had a notebook in NotebookLM, it is in Gemini Notebook now, under the new name.

The real change was code execution

Alongside the name, Google added something substantive: each notebook now gets access to a secure cloud computer, which lets the model write and run code against your sources rather than only reading them. In practice that means you can hand it a spreadsheet of figures and ask for an actual calculation, and it will compute the answer instead of estimating it from the text.

That capability arrived first for Google AI Ultra subscribers and Workspace business customers on the higher access tiers, with a rollout to Pro users on the web following.

Notebooks now reach into the Gemini app and Search

The other part of the change is where notebooks live. You can create and open them inside the Gemini app, with syncing between the app and the standalone site, and Google has said notebooks are coming to AI Mode in Search. The direction is clear enough: the notebook is becoming a feature of Gemini rather than a separate destination.

How Does Gemini Notebook Actually Work?

The mechanics are simple, but the constraint at the centre of it is what makes the output trustworthy.

You choose the sources

A notebook starts empty. You add sources: PDFs, Google Docs and Slides, websites, pasted text, YouTube videos, and audio files. These are the entire world the model can see for that notebook.

It answers only from those sources

Ask a question and the model searches your sources, assembles an answer from what it finds, and stops there. If the answer is not in your material, it tells you so rather than filling the gap. This is the opposite of the usual failure mode with general chatbots, where a plausible-sounding answer arrives with no way to tell whether it came from your document or from somewhere else entirely.

Every claim carries a citation

Each answer comes with inline citations that point at the specific passage behind it. Clicking one opens the source at that point. For anything you intend to rely on — a contract, a compliance document, a set of research papers — this is the feature that matters most, because it turns "the AI said so" into something you can check in two seconds.

It reformats your sources into other things

Beyond answering questions, a notebook can turn its sources into study guides, timelines, briefing documents, and the audio and video summaries the product became known for. These are generated from your sources too, so they inherit the same grounding.

How Do You Get Started? Step by Step

Step 1 — Sign in

Go to the Gemini Notebook site and sign in with a Google account. A standard consumer account is enough to start; you do not need a paid subscription for the core features.

Step 2 — Create a notebook and name it for one topic

Click to create a new notebook. Give it a narrow name — one project, one subject, one case. The temptation is to make one giant notebook, and it is the single most common mistake, because relevance drops as the source set gets broader.

Step 3 — Add your sources

Upload files, paste links, or pull in Google Drive documents. Add the material that genuinely bears on the question. Padding a notebook with loosely related documents makes answers worse, not better.

Step 4 — Ask a real question

Start with something specific rather than "summarise this". Specific questions produce specific, citable answers: "what notice period does clause 14 require", not "tell me about the contract".

Step 5 — Check the citations

Click through on at least a couple of citations before you trust a notebook on anything that matters. The grounding is good, but verifying it is cheap and it tells you quickly whether your sources actually contain what you assumed they did.

Want to know which of your pages AI tools actually read? Request a free SEO audit at https://www.seo-stuff.com/free-audit.

What Are the Limits on Each Plan?

Google publishes the caps on its own support pages and labels them subject to change, so treat the figures below as the shape of the tiers rather than a permanent contract — check the current numbers before planning around them.

The free tier is more generous than people expect

A standard free account allows 100 notebooks and 50 sources in each one. For most individual research projects that is not a constraint you will hit. The free tier is a genuine working tier, not a trial.

Paid tiers mainly raise the source cap

Moving up the plans increases both limits, but the source cap per notebook is the one that changes most:

• Standard (free): 100 notebooks, 50 sources per notebook
• Plus: 200 notebooks, 100 sources per notebook
• Pro: 500 notebooks, 300 sources per notebook
• Ultra, 20 TB plan: 500 notebooks, 500 sources per notebook
• Ultra, 30 TB plan: 500 notebooks, 600 sources per notebook

There are also usage quotas that reset on a daily or monthly cycle depending on the plan, which govern how much you can do rather than how much you can store.

The source cap matters more than the notebook cap

If you are deciding whether to pay, the question to ask is how many documents belong in a single notebook. A hundred notebooks of 50 sources is plenty for most people. One literature review with 200 papers in it is not possible on the free tier, and that is the case where upgrading actually buys you something.

When Should You Use It?

Gemini Notebook is a specialist. It is excellent at a narrow set of jobs and mediocre outside them.

Working through a fixed body of documents

Research papers, case files, regulatory documents, a competitor's published material, transcripts from a round of customer interviews. Anything where the set of sources is known, bounded, and the answer must come from inside it.

Anything where provenance matters

If you will have to defend the answer — to a client, a regulator, a supervisor — the citations are the whole point. No other mainstream AI tool makes checking the source this fast.

Learning material you did not write

Lecture notes, a dense manual, a technical specification. The study guide and audio summary outputs are genuinely good at making unfamiliar material approachable, and because they are grounded, they do not introduce errors the original did not contain.

Reconciling documents that disagree

Put two versions of a policy in one notebook and ask where they differ. This is a task general assistants handle badly and a grounded notebook handles well.

Where It Falls Down

It cannot help with what you did not upload

This is the design, not a bug, but it catches people out. Ask a notebook a general question and it will decline. If your question needs outside knowledge, you are in the wrong tool.

Source quality is your problem

Grounding guarantees the answer reflects your sources. It guarantees nothing about whether your sources are right. A notebook full of outdated documents will confidently give you outdated answers, correctly cited.

Long documents still get summarised

A 400-page PDF inside a notebook is not read exhaustively for every question. Retrieval picks the relevant passages, which is usually what you want, but it means a detail buried in an unlikely place can be missed. For questions that hinge on one clause, say which document and roughly where.

The feature set moves quickly

Between the rename, code execution, and the move into the Gemini app and Search, this product has changed substantially in a short time. Anything you read about it — including this — is worth checking against Google's own pages if you are making a decision on it.

Not sure whether your site is visible in AI search at all? A free SEO audit at https://www.seo-stuff.com/free-audit will show you.

NotebookLM vs the Alternatives

The honest comparison is that these tools are not competing for the same job, and picking by brand rather than by task is how people end up frustrated.

• vs ChatGPT: ChatGPT wins on general reasoning, writing, and anything needing knowledge beyond your files. Gemini Notebook wins decisively when the answer must come from your documents and be traceable. Use ChatGPT to draft; use a notebook to check what your sources actually say.
• vs Claude: Claude handles long documents well in a single conversation and is strong on careful analysis of a text you paste in. Gemini Notebook is better when you have many documents rather than one, and when you need the citation trail across all of them.
• vs Perplexity: Perplexity cites too, but it cites the open web. Gemini Notebook cites your private corpus. Perplexity answers "what does the internet say"; a notebook answers "what do my files say".
• vs Gemini chat: Same model family, different constraint. Gemini chat will answer from its own knowledge; the notebook will not. If you want it confined to your sources, use the notebook.

The practical workflow for most people is both: a general assistant for thinking and drafting, and a notebook for anything that has to be right and attributable.

Final Thoughts

Google NotebookLM is now Gemini Notebook, and the rename is the least interesting thing about it. What makes it worth using is the constraint — it answers from your sources and nothing else, and it shows you where each answer came from. That is a narrower promise than a general assistant makes, and it is a far more reliable one.

Start on the free tier. A hundred notebooks with 50 sources each is enough to find out whether the grounded approach suits how you work, and the only real reason to pay is a single project with hundreds of documents in it.

If you are weighing up the wider Google AI stack while you are here, our guide to what you actually get from Google Gemini's free tier covers where the free limits sit across the rest of the product family and when upgrading is worth it.

Ready to see how your site performs in AI and traditional search? Claim your free SEO audit at https://www.seo-stuff.com/free-audit and get a prioritised action plan.