Google recently changed how it calculates usage limits for Gemini.The shift from the simpler previous model of a fixed number of uses per day to calculating usage based on computational load—such as prompt complexity, features used, and conversation length—does better reflect the actual costs of AI services. After all, text-based Q&A, Deep Research, image generation, and video generation consume different amounts of computing power. But from the user’s perspective, it’s now hard to predict how much quota each sent message will consume; they can only keep an eye on the percentage bar showing how much is left, and many people are really struggling to get used to it.
If the usable quota is roughly the same as before, there’s no problem. However, after Google changed the computational quota, disputes kept arising. Many people complained that they hit the quota shortly after using it, and even the official team has recognized this.Significantly increase Antigravity’s Gemini quota by 9xRecently, another outrageous case has surfaced: a user complained on social media that after entering just one video-generation prompt, not only did the generation fail, but Gemini’s five-hour usage quota went straight from 0% to 100%.
Gemini’s new usage limits spark controversy: AI Pro user claims one failed video prompt drained a 5-hour quota.
According to foreign media Android Authority According to the report, Google AI Pro user Ashutosh Shrivastava recently shared a video on X saying that when he used Gemini’s Avatar video generation feature, his 5-hour usage limit initially showed 0%. He then entered a simple video generation prompt; the process ran for about 3–4 minutes, but the video generation ultimately failed, while the usage limit jumped straight to 100%—meaning he received nothing and exhausted his entire 5-hour quota:
Google Gemini head Josh Woodward also noticed the post and later replied below, essentially saying, “Oops, let us take a look.” However, as of now, there has been no further explanation on whether this is simply a bug, a calculation anomaly in a specific Avatar video feature, or whether the new usage system could really deduct credits this way in certain scenarios:
Google has changed Gemini’s usage limits to be compute-based, without detailing how much each feature might consume. It only noted that advanced models, higher reasoning levels, video or music generation, Deep Research, and similar features use up more quota, leaving users to figure it out through trial and error. Getting results is fine, but if you run into a situation like Ashutosh Shrivastava’s, it’s enough to make you want to smash your computer—you have to wait 5 hours for the limit to reset before you can continue.
However, I suspect Ashutosh Shrivastava’s situation is a bug — it’s unlikely that generating one video would max out the usage quota. That said, generating videos does consume a lot: a single Veo 3 video generation uses about 26%, and that’s with just a simple prompt. I also tested Deep Research with a roughly 1,000-word prompt, and it consumed a huge amount too — the five-hour usage quota jumped straight from 0% to 45%. Gemini’s current usage allowance really is a bit too small:
Gemini has just changed how it calculates computational usage, so this is essentially a teething period. Once Google gets more user feedback, it should gradually adjust to more reasonable quotas—something to look forward to.
Source: KOCPC Chinese