Google changes Gemini usage limits and introduces pay-as-you-go AI credits due to complaints

Google today announced a series of adjustments to usage limits for the Gemini app, responding to significant user feedback. The new policy, which considers the “computational power” used, had been introduced at I/O 2026 last week. The changes aim to provide a fairer and more efficient experience for all users of the artificial intelligence tool.

The company noted that the decision to review the parameters was taken following complaints about the rapid exhaustion of quotas. The computational power usage approach is designed to take into account the complexity of requests, the tools used and the duration of conversations, factors that directly influence resource consumption.

Detalhes of the new policy for computing power

Gemini’s “computational power usage” system now operates with updates every 5 hours, until the pre-established weekly limit is fully reached. Esta methodology seeks to more accurately reflect the real consumption of computing resources. Antes, users reported that intensive activities quickly consumed their quotas, preventing prolonged use of the AI ​​tool.

Google had already clarified that a simple text request consumes significantly less computational power than a complex one involving video or programming. Compreender this distinction is essential for users to manage their expectations and the way they interact with the application. With these adjustments, the company seeks to make consumption more transparent and predictable for the community.

One of the main new features is the introduction of an option for Gemini app users to purchase additional AI credits. Esses credits will be made available on a pay-as-you-go model, offering greater flexibility for those who require more capacity. Esta functionality aims to serve those who demand intensive use of Gemini and found barriers within existing limits.

Alterações specific to Gemini 3.1 Pro and Flash-Lite

Josh Woodward, Gemini lead, detailed that Google is limiting the amount of quota a single prompt can consume when using Gemini 3.1 Pro. Measure Esta was implemented in response to complex prompts that included large files, which exhausted usage limits too quickly for some users. Agora, the intention is to allow users of the Pro model to make better use of their quotas.

Google also made a point of clarifying that system errors will not be counted towards the user’s usage limits. The company stated that “if a request fails, you will not be charged,” emphasizing that “errors in our system are our responsibility, not yours.” Sua quota is only used for requests that complete successfully, which provides an important guarantee for the community.

Outra relevant change directly benefits users of the Flash-Lite 3.1 model. The prompts run in this model are now completely free and will not be deducted from the user’s computing power quota. Esta decision expands access to a lighter model, allowing more experimentation and basic use without concerns about credit consumption.

Leia Tambem

João Fonseca x Djokovic: where to watch the game live at Roland Garros

João Fonseca x Novak Djokovic in a decisive duel for the third round of Roland Garros

Conmebol defines Libertadores round of 16 clashes with six clubs from Brazil

Transparência and usage tracking improvements

Tarefas that require more robust processing, such as Pesquisa Avançada, were identified by Google as the largest resource consumers. In response to this reality, the company has committed to providing more detailed usage analytics as well as proactive notifications. Essas tools seek to help users maximize their limits and better understand how their quota is being used.

Atualmente, the dashboard available at gemini.google.com/usage only offers an overview of consumption. With the promised improvements, this platform is expected to become a more robust and informative tool, allowing for more granular management of the use of artificial intelligence. Transparency in resource allocation is a priority for the company, aiming to strengthen user trust.

Google also made an important clarification regarding the selection of specific models within the application. Once the user chooses a model, such as the Gemini 3.1 Pro or the Flash-Lite 3.1, that preference will be remembered and maintained across all future sessions. The change will only occur if the user manually adjusts this setting or if a limit is reached that triggers automatic return to a less demanding model.

Otimizações for Google AI users Ultra

Google has confirmed the fix of a significant bug that caused “certain people” quotas to exhaust early after viewing just one or two Omni videos. Este problem, which generated frustration among users, was promptly resolved by the development team. The correction ensures that resource consumption is fairer and aligned with what the system promises.

Adicionalmente, Google AI Ultra users now have an extra benefit: twice as many generations as Omni. Esta measure substantially increases the creative and interaction capacity for premium subscribers. The company continues to seek new opportunities to increase the amount of Omni available, reinforcing its commitment to valuing its most engaged users.