Google Just Changed How It Counts Your Gemini AI Usage. Here Is What You Actually Get Now.
Google switched from counting requests to measuring computing power. That sounds technical, but it has a practical effect: you may run out of AI credits faster than you expect, and the new rules are deliberately vague.
Key points
- Google now measures Gemini AI usage by computing power consumed, not by the number of requests a user makes.
- US subscribers choose from four tiers priced at $0, $8, $20, or $100 to $200 per month.
- Free-tier users get a context window of 32,000 tokens, roughly 24,000 words, per conversation; AI Pro and Ultra users get up to one million tokens, about 750,000 words.
- Usage resets on two separate clocks: every five hours for daily use, and every seven days for a weekly cap.
- Google's own support documents warn that limits can change without notice, and free users may be cut back first.
Google spent much of this summer adding AI features across its apps, so that Gemini, its artificial intelligence assistant, now shows up almost everywhere in Google's product lineup. But alongside those upgrades, first reported by Wired AI, came a quieter change that is already catching users off-guard: Google rewrote the rules for how it measures what you use.
Previously, the system was simple. Ask for three images, use three credits. Now Google charges by computing effort. A short question to a basic model costs a little. A long, complex request to a smarter model, one capable of deeper reasoning, costs a lot more. Two users on the same plan could have very different experiences depending on how they prompt.
That approach makes financial sense for Google. It is essentially billing you closer to what you actually cost its data centres. For you as a user, it is murkier. There is no clean rule like "five video generations a day" to memorise.
Will you suddenly run out of AI mid-task?
Yes, that is possible, especially if you favour heavier tasks like generating videos or asking Gemini to write code. The complexity of your request, the AI model you choose, and the "thinking" level you set (Standard, Extended, or Deep Think) all drain your allowance at different rates.
The four plan options in the US are the free tier, AI Plus at $8 a month, AI Pro at $20 a month, and AI Ultra at either $100 or $200 a month. Google describes free usage limits only as "standard." AI Plus doubles them. AI Pro quadruples them. AI Ultra multiplies AI Pro limits by five or twenty, depending on which Ultra price you pay.
One concrete difference between plans is the context window, which is the maximum amount of text Gemini can hold in memory during a single conversation. Free users get the equivalent of roughly 24,000 words. AI Plus users get about 96,000 words. AI Pro and Ultra users get around 750,000 words, enough to feed an entire novel into a single chat.
To check where you stand, open the Gemini app on the web, click the cog icon at the lower left, then select "Usage limits." On Android or iOS, tap the menu at the top left, then the cog, then "Usage limits." You will see two progress bars: one for your five-hour rolling allowance, one for your weekly cap. Hit the weekly cap on a paid plan and you drop down to the most basic model until the counter resets.
Google's own support pages note that limits may shift without warning due to capacity pressures, and free users are the first in line for cutbacks.
Honest takeaway: Before upgrading your plan, spend a week checking the usage screen daily. Most casual users will not touch their weekly cap. If you do hit it regularly on free or Plus, the $20 Pro plan is where the limits jump meaningfully. Only pay for Ultra if you are running AI tasks long enough to fill a small library.
