Google is tweaking how Gemini Notebook handles usage limits in a way that should feel more natural for people who actually use the tool throughout the day. Starting September 2, the company is replacing rigid daily caps with flexible, compute-specific limits that refresh every five hours and scale based on what you’re doing inside a notebook.

The move follows a similar shift Google made to the main Gemini app earlier this year and signals a broader transition away from simple “X chats per day” rules toward a more nuanced model that accounts for prompt complexity, chat length, number of sources, and which features you’re using.

For many users, the old limits felt arbitrary. You could blow through your allowance with a couple of heavy, source-heavy notebooks in the morning, then be locked out until the next day—even if you just wanted to ask a quick follow-up question.

Under the new system, your overall usage limit is calculated dynamically. A short, simple query with one or two sources will cost far less than a long, multi-source analysis that asks the model to synthesize a dozen PDFs, generate an audio overview, and produce a structured report.

Google describes it as giving you “more control over your workflow” and your “compute budget.” In practice, that means:

• Limits refresh every five hours instead of once per day, so you can keep working across the day rather than hitting a hard wall at midnight PT.
• The interface will show your remaining usage below the chat and, if a request would exceed your limit, suggest lighter alternatives (for example, a shorter summary or fewer sources).
• Your plan still matters. Free accounts get “standard” limits; Google AI Plus, Pro, and Ultra subscribers get 2x, 4x, and up to 20x higher limits depending on tier.

This is effectively the same compute-based model Google rolled out to Gemini in May, now extended to Notebook as the company tries to align limits with actual resource consumption instead of a blunt count of interactions.

From Google’s side, the shift is about cost control and fairness. Running large-context, multi-source notebooks with advanced features is significantly more expensive than short Q&A sessions. A purely “per chat” limit subsidizes heavy users at the expense of light ones, while also making it harder to predict infrastructure load.

A compute-aware model lets Google:

• Tie limits more closely to token usage and model load, especially as newer Gemini 3.x Flash models roll out with different pricing tiers.
• Encourage more efficient prompting and source management without outright banning complex workflows.
• Create a clearer upgrade path: if you consistently need more compute, the answer is a higher-tier plan rather than gaming daily caps.

For power users—researchers, students, analysts, and creators who live in Notebook—the change should feel more forgiving. Instead of a single daily bucket that disappears after a couple of heavy sessions, you get multiple refresh windows across the day. That’s especially useful if your work is bursty: a big morning session, a lighter afternoon, then another push in the evening.

The trade-off is that limits are now less transparent. “Standard limits” don’t map cleanly to a fixed number of chats anymore; they depend on what you ask and how you ask it. Google’s help pages acknowledge this, noting that limits factor in “prompt complexity, the models and features you use, the length of your chat and specific features.”

Gemini Notebook (formerly NotebookLM) doesn’t have standalone pricing; it’s bundled into Google’s AI subscription tiers. After Google’s May 2026 reshuffle, those tiers look roughly like this for consumers:

• Free / no plan: standard limits
• Google AI Plus: 2x standard limits
• Google AI Pro: 4x standard limits
• Google AI Ultra: 5x or 20x higher than Pro, depending on the Ultra SKU

Source limits per notebook have also expanded over time, with Ultra plans supporting hundreds of sources and very large documents. The new compute-based limits sit on top of those structural caps, governing how much you can actually do with those sources in a given window.

For anyone already paying for AI Pro or Ultra, the practical effect is more headroom to run complex notebooks without constantly worrying about hitting a daily ceiling. For free users, the five-hour refresh should reduce the “all or nothing” feeling of the old system, even if the absolute ceiling remains lower.

If you mostly use Notebook for light tasks—summarizing a couple of articles, getting quick answers from uploaded PDFs, or generating short audio overviews—you might not notice much difference beyond seeing a usage meter and occasional suggestions to lighten a request.

If you regularly:

• Upload dozens of sources
• Run long, multi-step analyses
• Use advanced features like deep synthesis, structured outputs, or audio overviews

you’ll likely feel the change immediately. Heavy sessions will consume more of your limit, but you’ll also get more chances to resume work as the quota refreshes every five hours.

Google’s framing is all about flexibility and control, but the underlying message is clear: compute is the scarce resource, and limits will increasingly reflect that. For a tool designed around deep, source-grounded reasoning, that’s a more honest model than pretending every chat costs the same.

The rollout begins September 2 for consumer accounts on web and mobile, with the usage tracker and dynamic limits appearing directly in the Notebook interface.