In a post on X, @testingcatalog reports that Gemini Notebook is getting a new usage system built around five-hour limit refreshes, a flexible Notebook-specific allocation, estimated usage for artifact generation and a background queue for work users defer. The post isn't a Google announcement, so exact quotas, eligibility and availability are still unconfirmed.

The reported changes would affect how users plan AI-generated artifacts such as slides and videos. They describe a usage model in which limits refresh more often, spending is shared across Notebook features and some generation requests can be deferred when the current allocation is unavailable.

What the reported Gemini Notebook usage limits change

The report describes four connected changes:

  • Limits would refresh every five hours rather than every 24 hours.

  • Feature-specific limits would be replaced by a flexible limit for Gemini Notebook.

  • The interface would estimate how much an artifact-generation request uses.

  • Users who are out of limits could defer some generations to a background queue.

Together, these changes suggest a shift from managing separate feature allowances toward managing a shared allocation that is replenished more often. The post does not specify how large that allocation is or whether the same rules apply to every account.

Limits would refresh every five hours

According to the report, Gemini Notebook limits would refresh every five hours instead of every 24 hours. That would divide usage into more frequent refresh periods, potentially giving users more opportunities to resume work during the day rather than waiting for a single daily reset. The report describes the five-hour refresh only as part of this new usage system, leaving open whether it covers every Notebook feature.

Gemini Notebook screen describing five-hour limit refreshes, flexible usage and a new background queue.
Gemini Notebook screen describing five-hour limit refreshes, flexible usage and a new background queue.

Image credit: @testingcatalog on X

One flexible Notebook allocation would replace feature-specific limits

The report also says Gemini Notebook is replacing limits assigned per feature with a flexible Notebook-specific limit. In practical terms, that would let users decide how to spend a shared allocation instead of receiving separate allowances for individual generation tools.

That could make the system easier to understand when users switch between artifact types. Someone creating a slide deck, for example, would be spending from the same Notebook-level allocation described in the report rather than checking a separate limit for every feature. The report describes that pool in general terms, leaving open whether every artifact type draws from it and how the allocation is calculated.

Usage estimation and the background queue

A screenshot shared with the report shows an “AI Usage” indicator that distinguishes “Already used” from “Estimated use.” It also shows “Generate later” and “Generate now” controls. A tooltip associated with the deferred option says it does not use the current limit, costs less and will be ready in hours. The report leaves open pricing and account-tier differences.

This suggests that Gemini Notebook may show an estimate before a user starts an artifact-generation task. The report does not explain how the estimate is calculated, how much usage a particular artifact consumes or whether the estimate changes based on the selected format, length or other settings.

Gemini Notebook dialog showing AI usage estimates and Generate later for a slide deck.
Gemini Notebook dialog showing AI usage estimates and Generate later for a slide deck.

Image credit: @testingcatalog on X

The same report describes a background queue for times when a user is out of limits. Users could defer generations such as videos and slides, with those tasks then generating automatically later. This would separate the decision to request an artifact from the point at which the system has enough available capacity to generate it.

The report does not say how long queued work takes, whether users can cancel it, how completion is communicated or which artifact types are supported. “Ready in hours” appears in the tooltip, but it should not be treated as a guaranteed completion time for every request.

Sources