Google has changed how it meters Gemini use, replacing clearer per-action limits with a system based on how much computing power a request is expected to consume. For Gemini users, that means the old mental math, such as a fixed number of image or video generations per day, is less useful.
The company’s AI apps have gained broader capabilities across Google services this summer, but the bill for those features is now being expressed in fuzzier units. A short request for a weather forecast should count less than a long coding prompt or a complex video generation, according to Google’s usage model. That may match Google’s data-center costs more closely. It also makes the user experience more opaque, which is very on-brand for subscription software that would prefer you stare at a progress bar and consider upgrading.
How the new metering works
Google now counts Gemini usage by resource demand rather than by the number of prompts alone. The main variables are the user’s plan, the length and complexity of the prompt, and the Gemini model selected in the prompt box.
Google’s support documentation says Gemini access “is subject to change or may be limited based on testing, experimentation or availability.” In practical terms, the amount of Gemini a user can run may vary as Google adjusts capacity or experiments with limits.
The company makes several Gemini models available, including Flash-Lite, Flash and Pro. Google says more capable models count more heavily against usage. The models also offer different thinking settings: Standard, Extended and Deep Think. Those settings affect quality, speed and usage consumption, according to Google.
What each Gemini plan gets
In the US, users can stay on Gemini’s free tier or pay for AI Plus, AI Pro or AI Ultra. AI Plus costs $8 a month, AI Pro costs $20 a month, and AI Ultra is listed at either $100 or $200 a month depending on the level of service.
Google does not publish a precise quota for the free tier in the material described here. It calls those limits “standard.” AI Plus gets twice the standard limits, while AI Pro gets four times the standard limits. AI Ultra gets either five times or 20 times the AI Pro limits, depending on which Ultra payment level the user has.
Context windows also differ by plan. That is the amount of material Gemini can consider inside one conversation thread. Free users get 32,000 tokens, which Google equates to about 24,000 words. AI Plus raises that to 128,000 tokens, about 96,000 words. AI Pro and AI Ultra users get a one million token window, roughly 750,000 words.
How to check remaining Gemini usage
Google exposes usage status inside the Gemini app, although the company still does not reduce the system to one neat number. On the web version, users can click the cog icon in the lower-left corner and select Usage limits. On Android or iOS, users can open the menu from the top-left corner, tap the cog, and then select Usage limits.
The screen shows two bars. The first shows current usage and resets every five hours. If that bar is exhausted, Gemini tells the user when the next reset will occur. The second bar tracks weekly usage and resets once a week.
Paid users who hit their limits are shifted down to the most basic Gemini model until the next reset, according to the described behavior. Google’s support documents also say limits can change without notice because of capacity constraints, and free users may be the first affected when Google tightens access.
This story draws on original reporting from WIRED.
