Google eases Gemini AI limits after user backlash
What's the story
In response to user feedback about hitting limits too quickly, Google has announced changes to its Gemini app. The tech giant has introduced a new "compute-used" approach. The method refreshes every five hours until the weekly limit is met and considers prompt complexity, tool usage, and chat length.
Usage breakdown
New approach based on prompt complexity
Google's new approach to usage limits is based on the complexity of prompts. The company explained that "a simple text prompt uses far less compute than a complex video or coding prompt." This way, users will not hit their limits as quickly while using more complex features of the app. In a bid to further enhance user experience, Google will let Gemini app users purchase pay-as-you-go top-up AI credits.
Quota management
Quota limits for individual prompts
To prevent complex prompts with large files from quickly depleting limits, Google is capping the amount of quota a single prompt can use. The tech giant has also clarified that errors do not count against the usage limits. "If a request fails, you won't be charged," the company said. "Our system mistakes are on us, not you. Your quota is used only for successful completions."
Enhanced features
More detailed usage breakdowns and notifications for heavy tasks
For heavy tasks like Deep Research that "require more compute," Google is going to provide "more detailed usage breakdowns and notifications to help you maximize your limits." Currently, the gemini.google.com/usage dashboard only provides a high-level overview of usage. This change will give users a better understanding of their usage patterns and help them manage their quotas more effectively.
Prompt update
Flash-Lite prompts are now free of charge
Google has also made 3.1 Flash-Lite prompts free of charge and they won't count against your quota anymore. The company also clarified that when a specific model is selected, it remembers the choice across all future sessions unless manually changed or a cap is hit triggering an automatic switch to a lighter model. This way, users can customize their experience according to their needs without worrying about hitting limits too quickly.