People who’ve been using Gemini frequently lately may feel like their quota is running out faster than before—this isn’t just your imagination. Google quietly released an official announcement, completely changing how Gemini calculates usage. The old method of “fixed daily requests” has been replaced with “compute-based calculation,” meaning every request now counts how much compute resources you used to determine your remaining quota, rather than just counting how many messages you sent.
Google Gemini overhauls usage rules: switching from daily limits to compute-based billing
Previously, Gemini’s usage limits were based on how many prompts you sent. For example, the free tier allowed up to 5 prompts per day on the Pro model, while AI Pro users had 100 per day. The biggest change in the new system is that prompts are no longer counted as one each. Instead, usage is calculated based on actual computational resources consumed. This means if your prompt message contains more content, it will use more computational resources. Conversely, shorter content uses less computational resources.
According to Google’s official support documentationAt least four factors affect usage: the complexity of the prompt itself, which features you select (image generation, video generation, Deep Research, Pro model, extended thinking, Deep Think, etc.), the length of the entire conversation, and the size and number of files you upload. In other words, each feature no longer has a fixed quota. If you’ve used up your compute allocation, you won’t be able to use it anymore and will need to wait until the “compute reset.”
The reset cycle has also been changed to replenish the quota every 5 hours, but there’s still an overall weekly cap. This dual-track design (short-term + weekly cap) is almost identical to what Claude has been using for a long time, meaning Google has finally caught up with the industry standard.
As for the computational cost of each solution? Roughly as follows
- The free version has a standard quota
- Google AI Plus has 2x the standard quota
- Google AI Pro has 4 times the standard quota
- Google AI Ultra has the highest quota, reaching 20 times
The actual feel and difference really depends on each person’s personal experience, especially for those who use it heavily every day, who need to allocate the usage more efficiently, otherwise it will fill up quickly.
What’s strange is that, possibly due to being rushed, this Google announcement is written somewhat briefly, without detailed explanations. At the same time, the original Gemini Series Apps Info PageOr list the fixed quotas alongside, such as: “Pro 3.1: up to 100 per day,” “Thinking model: up to 300 per day,” etc.:
The new system has been gradually implemented starting from May 17, 2026, and previously there wereForeign media outlets discovered that Google is also developing usage dashboard functionality.It’s not online yet, but since the computational load has been changed, it should be visible soon.
In the web version of Gemini, you can find the “Usage limits” feature in the menu. After entering, you can check how much usage you have remaining, when the reset time is, and the weekly limit:
Source: KOCPC Chinese