Google replaces Gemini Notebook’s daily quotas with compute-based limits
Gemini Notebook will adopt five-hour and weekly compute-based limits for consumer accounts on September 2, replacing predictable daily feature quotas with a workload-sensitive meter.
Google will move Gemini Notebook consumer accounts to compute-based usage limits on September 2, 2026. Instead of counting each chat or generated artifact equally against a fixed daily quota, the new system measures prompt complexity, conversation length, source count, model choice and the feature being used.
Limits will refresh every five hours until the account reaches a separate weekly ceiling. Google has published plan multipliers, but it has not disclosed the underlying number of compute units or a stable conversion between those units and common tasks.
Key takeaways
- The change begins rolling out to consumer accounts on web and mobile on September 2.
- Usage will depend on prompt complexity, chat length, source count, models and features rather than a fixed generation count.
- The short-term allowance refreshes every five hours, while a weekly limit also applies.
- Google AI Plus receives twice the standard limit, AI Pro receives four times the standard limit, and AI Ultra receives five or 20 times the Pro limit depending on the subscription.
- Web users who exhaust their allowance can queue some Studio outputs with “Generate later” and receive a notification when processing completes.
What changes on September 2
Gemini Notebook previously presented separate daily or monthly caps for chats, Audio Overviews, reports and other generated artifacts. The new system combines AI activity into a workload-sensitive allowance. A short chat response and a Video Overview will therefore no longer have an implied one-to-one cost.
Google says all existing features remain available, but the amount of usage consumed by each request will vary. The rollout announcement names consumer accounts on the web and mobile apps; it does not establish the same September 2 schedule for every Workspace, Education or Cloud-managed deployment.
Users can see remaining capacity under Settings and Usage. Chat will also show a warning near the limit and the time when the five-hour allowance refreshes. Studio displays an expected-cost indicator before generation, giving users a relative signal when a requested artifact will consume more capacity.
How the new meter works
Google’s help page lists five inputs to the calculation: prompt complexity, the model selected, the features invoked, conversation length and specific feature requirements. The company’s announcement additionally points to the number of sources attached to a notebook.
This means repeated follow-up questions in a long, source-heavy notebook may cost more than similar prompts in a new, smaller context. Media-heavy outputs such as Video Overviews or Slide Decks may also consume more of the allowance than ordinary text interactions. Google says Gemini Notebook can suggest a cheaper alternative when a requested output would exceed the available limit.
When the five-hour allowance runs out, users must wait for its stated refresh time or upgrade. Refreshes continue only until the account reaches its weekly ceiling, after which the help page directs users to wait for the weekly reset or move to a plan with a larger multiplier.
Generate later softens some interruptions
On the web, an exhausted user can queue supported Studio artifacts using “Generate later.” Gemini Notebook processes the request after capacity becomes available and can notify the user when the output is ready. Google warns that queued generation may take a couple of hours.
The option reduces the need to repeatedly retry an expensive artifact, but it does not restore interactive work immediately. It is also initially limited to the web, so mobile users should not assume that every deferred-generation control will be present on September 2.
The missing number matters
Google publishes relative entitlements rather than the standard allowance itself. Accounts without a paid plan receive the standard limit; AI Plus receives twice that amount; AI Pro receives four times the standard amount; and the two Ultra configurations receive five or 20 times the Pro allowance.
Without the base number or a task-cost table, users cannot translate those multipliers into a reliable number of conversations, Audio Overviews, videos or slide decks. Independent coverage also notes that the old fixed daily counts disappear from the updated help material, making plan comparisons less predictable until real usage data accumulates.
Anyone relying on Gemini Notebook for teaching, research or deadline-driven production should inspect the Usage panel after rollout and test representative notebooks before assuming previous capacity still applies. The decisive evidence will be how quickly typical text, audio, video and slide workflows consume the five-hour and weekly meters.
Source check
- Google’s announcement confirms the September 2 consumer rollout, compute-based accounting and five-hour refresh cycle.
- Gemini Notebook Help documents the weekly ceiling, plan multipliers, usage indicators and Generate later workflow.
- ANTI/HYPE’s independent report compares the new meter with the previous fixed quotas and highlights the undisclosed base allowance.
Try the related loot
Put six hosted Workers AI models behind Cloudflare AI Search
