Google is transitioning its Gemini Notebook service to a new usage-tracking model that calculates limits based on the computational intensity of user requests. Beginning September 2, the current system of flat daily and monthly caps for features like chats, audio overviews, and video overviews will be retired in favor of a flexible, compute-specific approach. This change applies to both free and paid tiers, though specific capacity thresholds for each subscription level remain unspecified.
Under the new framework, usage will be determined by factors including the complexity of a prompt, the length of a chat session, the number of sources referenced, and the specific features utilized. By moving away from fixed request counts, Google aims to provide users with more granular control over their workflow. The company notes that this system will allow users to manage their compute budget more effectively, as the new limits refresh every five hours rather than on a daily cycle.
The update introduces a notification system designed to help users monitor their remaining capacity. If a user approaches their limit, the interface will suggest alternative outputs or allow them to defer resource-heavy tasks. For instance, users can choose to delay the generation of audio overviews or slide decks. These deferred tasks will generate automatically in the background, and users can opt-in to receive notifications once the content is ready for review.
This shift reflects a broader industry trend toward usage-based models for generative AI services, which often face high demand for agentic features that consume significant tokens. By implementing these dynamic limits, Google joins other providers that have moved away from static request counts to better manage server load and resource allocation. The update will roll out to all consumer accounts across both web and mobile platforms starting next week.