Gemini Notebook (formerly NotebookLM) Usage Limits (2026)

Seen “You’re almost at your AI usage limit” under the chat box lately? That line is the visible end of a quiet change to Gemini Notebook (formerly NotebookLM) usage limits. On September 2, 2026, Gemini Notebook stopped counting fixed daily allowances, like 3 Audio Overviews a day, and started measuring how much work each request takes. Here’s how the new system works, what each plan gets, and how to make it last.

Key Takeaways

• Since September 2, 2026, limits refresh every five hours until a weekly cap (Gemini Notebook Help, 2026).
• Prompt complexity, chat length, source count and features all count.
• Plus gets 2x the standard allowance, Pro 4x.
• Notebook and source limits didn’t change.

Gemini Notebook limits at a glance

Here’s every number in one place. Usage refreshes every five hours on every plan, until you reach that plan’s weekly cap.

Plan	Notebooks	Sources each	Per source	Chat and Audio Overview usage	Usage refresh
Free	100	50	500K words or 200 MB	Standard	Every 5 hours.
Google AI Plus	200	100	500K words or 200 MB	2x standard	Every 5 hours.
Google AI Pro	500	300	500K words or 200 MB	4x standard	Every 5 hours.
Google AI Ultra (20 TB)	500	500	500K words or 200 MB	5x Pro	Every 5 hours.
Google AI Ultra (30 TB)	500	600	500K words or 200 MB	20x Pro	Every 5 hours.

Checked October 2, 2026, against Gemini Notebook Help: plan limits, usage limits and source size. Google doesn’t publish a fixed number of chats or Audio Overviews under the new system, only these multipliers.

What are the Gemini Notebook usage limits now?

Gemini Notebook now uses compute-based limits. Compute-based limits are allowances that scale with how much work each request takes. Instead of a fixed number of chats or overviews per day, your allowance refreshes every five hours until you reach a weekly limit (Gemini Notebook Help, 2026). Google announced the change on August 28, 2026, and it began rolling out to consumer accounts on the web and mobile on September 2 (Google, 2026). It mirrors the system the Gemini app adopted in May (9to5Google, 2026). The trade-off is simple. Light users, for instance, get more freedom, because five quick questions no longer burn five of a fixed fifty. Heavy users, on the other hand, lose predictability, since one long session in a big notebook can use more than they expect. In other words, the limit now follows the weight of your work, not the number of clicks.

When does the limit reset?

Every five hours, until you hit the weekly cap. The line under the chat box gives the exact time, and Settings, then Usage, shows your remaining usage and when it refreshes (Gemini Notebook Help, 2026).

Does it apply to work and school accounts?

Google’s announcement covers consumer accounts. Work and school accounts, by contrast, get Gemini Notebook through their Workspace edition, with limits set by that license (Gemini Notebook Help, 2026). If that’s you, trust the usage line under the chat box, and ask your admin which license you’re on.

What uses up your limit fastest?

Google lists four factors: prompt complexity, chat length, the number of sources, and the features you use (Google, 2026). The help center adds the models involved. There’s no published price per action; however, the first three share a root cause. Longer inputs cost more to process in Transformer models: researchers describe standard self-attention as an operation “which scales quadratically with the sequence length” (Beltagy et al., arXiv, 2020). Google’s own report says Gemini models “build on top of Transformer decoders” (Gemini Team, arXiv, 2023). As a result, a long thread, where every reply carries the history, and a question asked across 300 sources both hand the model far more text than a short question in a fresh chat. That said, Google doesn’t publish how Gemini Notebook prices each request, so treat this as the general reason rather than an exact rate.

We analyzed those four factors against everyday habits, and the table below maps them. Treat it as direction, not exact cost. Specifically, the Video Overview and Slide Deck row reflects Google’s own advice to schedule those outputs for later.

Usually lighter	Usually heavier
A short question in a fresh chat	A long thread where every reply carries the history.
Chatting with 3 selected sources	Chatting with all 300 sources in a big notebook.
A Mind Map or a short Quiz	A Video Overview or a Slide Deck.
A focused, single-purpose prompt	A multi-step prompt asking for several outputs.

Before a Studio generation, a bar now shows the expected AI usage cost (9to5Google, 2026). Consequently, you can see which way a choice leans before you commit. Why guess when the app tells you? The cheapest request of all, in fact, is the one you never repeat: a prompt that works the first time, and an output you keep instead of regenerating.

Free vs paid: how much more usage do Plus and Pro get?

Every plan runs the same system with a bigger allowance. Google AI Plus gets 2x the standard limits, Pro gets 4x, and Ultra gets 5x or 20x more than Pro, depending on the subscription (Gemini Notebook Help, 2026). The free tier, in other words, is the baseline every other plan is measured against. Here’s how that works out.

Plan	Usage allowance	Compared with free
No plan (free)	Standard limits	The baseline.
Google AI Plus	2x standard	Twice the free allowance.
Google AI Pro	4x standard	Four times the free allowance.
Google AI Ultra	5x or 20x more than Pro	Roughly 20 or 80 times the free allowance.

Do the maths and the top Ultra tier works out to roughly 80 times the free allowance (20 times Pro’s 4x). That gap, notably, explains why a free account can feel cramped in exam week while an Ultra account never sees the warning. As of October 2026, Google AI Plus costs $4.99 a month in the US and Pro $19.99, but prices move more often than this structure. Our explainer on whether Gemini Notebook is free compares the plans and prices.

What were the old limits per day?

Before September 2, the free tier allowed 50 chats, 3 Audio Overviews and 10 each of reports, flashcards, quizzes and mind maps per day, and the 30TB Ultra plan allowed 5,000 chats (Gemini Notebook Help, 2026). That’s where the familiar “3 a day” number comes from. From our research into Google’s help pages on October 1, 2026, we found the plan comparison page still shows this per-day table. The usage limits page, however, says compute-based limits took over on September 2, so we read the table as history.

Plan	Chats a day	Audio Overviews a day	Video Overviews a day	Reports, flashcards, quizzes, mind maps
Free	50	3	3	10 a day each.
Plus	200	6	6	20 a day each.
Pro	500	20	20	100 a day each.
Ultra (20TB)	2,500	100	100	500 a day each.
Ultra (30TB)	5,000	200	200	1,000 a day each.

Therefore, if an older article quotes those numbers as today’s limits, it’s describing the old system. People still search that way, however. We analyzed 1,451 Google autocomplete suggestions for Kortex blog topics in October 2026, and for this topic they included “notebooklm limits per day” next to “gemini notebook compute limits” and “gemini notebook new limits”.

How many notebooks and sources can you have?

Capacity is separate from usage: it’s how much your account and each notebook can hold, and it stayed the same. The free tier still allows 100 notebooks with 50 sources each, Pro allows 500 notebooks with 300 sources each, and the 30TB Ultra plan allows 600 sources per notebook (Gemini Notebook Help, 2026). Each uploaded source can be up to 500,000 words or 200MB (Gemini Notebook Help, 2026). Similarly, sharing a notebook doesn’t change any collaborator’s source limit. The table at the top of this guide has the full ladder.

Hitting the notebook count rather than the usage limit? That’s an organization problem, in fact, not a plan problem. Our system for organizing 50+ Gemini notebooks covers naming, tagging and archiving.

What happens when you hit the limit?

You get a message instead of a silent failure. Near the edge it reads “You’re almost at your AI usage limit. Limit resets at 3:00 PM”, and at the edge it becomes “Limit reached. All features are available after 3:00 PM” (Gemini Notebook Help, 2026). From there, you’ve got three options.

1. Wait for the refresh. The message gives the exact time. With five-hour windows, that’s usually the same day.
2. Generate later. On the web, pick an artifact in the Studio panel and select Generate later. Google warns it “may take a couple hours”, and you’ll be notified if notifications are on.
3. Upgrade. The Upgrade button in chat or in a Studio generation moves you to a plan with a bigger multiplier.

In addition, Gemini Notebook may suggest an alternative output when your first choice would go over the limit (Google, 2026). For example, it might offer a different Studio format that covers the same need.

How do you make your limit last longer?

Google’s list of factors points straight at the habits that help: chat length and the number of sources both raise what a request costs (Google, 2026). Rewording and regenerating, meanwhile, is where usage quietly disappears. A few small changes go a long way:

• Select fewer sources per question. Tick only the sources a question needs; as a result, you get sharper citations and a smaller request.
• Start fresh chats for new topics. A thread that’s run for an hour drags its whole history into every reply.
• Write the prompt right the first time. A tested prompt from our 30 best Gemini Notebook prompts beats three improvised attempts.
• Queue heavy outputs. For example, put Video Overviews and Slide Decks on Generate later at the end of a session.
• Keep what you made. Download decks and audio, and keep a copy of chats you’ll need again.

If your notes live in Notion or Obsidian anyway, our guide to saving Gemini Notebook notes to Notion or Obsidian shows how an AI assistant can carry them there through Kortex MCP, a private beta for invited Kortex Pro members. Invites go out from the Kortex MCP waitlist.

How does Kortex help you stretch your limits?

Kortex is a free Chrome extension for Gemini Notebook, and we built it, so weigh this section accordingly. It can’t raise Google’s limits. That said, it helps you avoid spending them twice. Its free plan includes 10 exports and 10 imports a day, and Pro, at $9.99 a month or $4.92 a month billed yearly, removes both caps (Kortex pricing, 2026). Exports read what’s already in your notebook and save it as PDF, Markdown, text or ZIP, so saving a chat never asks the model to do new work. Its saved prompt library, similarly, keeps your best prompts one click away, which cuts down on rewording. Kortex’s own limits are separate from Google’s. Anything Kortex asks Gemini Notebook to generate, however, such as a scheduled Audio Overview, counts against your allowance like any other request. See how the pieces fit in Kortex vs Gemini Notebook.

In short, compute-based limits reward focused questions and outputs you keep, and that second part is the easy one to fix.

Frequently asked questions

What are the Gemini Notebook usage limits in 2026?

Since September 2, 2026, consumer accounts use compute-based limits. Your allowance depends on prompt complexity, chat length, the number of sources and the features you use, and it refreshes every five hours until you reach a weekly cap (Gemini Notebook Help, 2026). Capacity is unchanged: 100 notebooks and 50 sources per notebook on the free tier.

When do Gemini Notebook limits reset?

Every five hours, until you reach your weekly limit. After that, you wait for the weekly reset or upgrade your plan. The message under the chat box shows the exact reset time, for example “Limit resets at 3:00 PM”, and Settings, then Usage, shows your remaining usage.

How much more usage do paid plans get?

Google AI Plus gets two times the standard allowance and Google AI Pro gets four times. Google AI Ultra gets 5x or 20x more than Pro, depending on the subscription, which puts the top Ultra tier at roughly 80 times the free allowance.

What uses up Gemini Notebook limits fastest?

Google doesn’t publish a cost per action, but it names four factors: prompt complexity, chat length, the number of sources and the features you use. It singles out Video Overviews and Slide Decks as outputs to schedule for later, which suggests they’re among the heaviest requests.

What happens when I hit the Gemini Notebook limit?

You see a message such as “Limit reached. All features are available after 3:00 PM”. You can wait for the reset, queue the output with Generate later on the web and get notified when it’s ready, or upgrade to a plan with a bigger allowance.

Did the source limits change on September 2?

No. The change affected how daily usage is counted, not how much a notebook holds. Google’s plan table still lists 50 sources per notebook on the free tier, 100 on Plus, 300 on Pro and up to 600 on Ultra, with up to 500,000 words or 200MB per source (Gemini Notebook Help, 2026).

---

Editorial note: fact-checked against the official help pages and docs linked above on October 2, 2026. Spotted something out of date? Contact us and we will fix it, or read more about Kortex.