Google Gemini Now Meters AI Usage by Computing Power, Not Requests

Google has fundamentally changed how it measures and limits usage of its Gemini AI platform, shifting from a simple request-based system to one that tracks the computational resources each interaction consumes. The change affects all users across free and paid subscription tiers, making it harder to predict when you'll hit your daily or weekly limits.

Under the previous system, users knew they could generate a specific number of images or videos per day. Now, usage depends on the complexity and resource intensity of each request. A simple weather query consumes far less of your quota than asking Gemini to code a functional mini-application. Two complex video generations might exhaust the same budget that previously allowed three simpler ones.

Why it matters

This shift reflects the real economics of AI infrastructure—different models and tasks have vastly different computational costs—but it transfers uncertainty to users. Businesses and individuals relying on Gemini for consistent daily workflows now face less predictable access. Google also reserves the right to adjust limits without notice based on capacity constraints, with free users likely affected first during resource crunches.

How the new usage system works

Three primary factors now determine how quickly you consume your Gemini quota. First, your subscription tier sets your baseline allowance. Second, the specific AI model you select matters—Gemini offers Flash-Lite, Flash, and Pro models, each with different capabilities and resource requirements. Third, the complexity and length of your prompts directly impact consumption.

Google also introduced "thinking" levels within each model: Standard, Extended, and Deep Think. More sophisticated thinking modes produce higher-quality responses but count more heavily against your limits.

Subscription tiers and multipliers

The free tier provides what Google calls "standard" limits, though the company doesn't publish specific numbers. The AI Plus plan, priced at eight dollars monthly, doubles those standard limits. AI Pro at twenty dollars monthly quadruples them. The AI Ultra tier offers either five times or twenty times the AI Pro limits, depending on whether you pay one hundred or two hundred dollars monthly.

All tiers access the same AI models, but paid subscribers get substantially larger context windows—the amount of information Gemini can process in a single conversation thread. Free users work within a 32,000-token window (roughly 24,000 words), while AI Plus users get 128,000 tokens. AI Pro and Ultra subscribers can use up to one million tokens, approximately 750,000 words.

Tracking your usage

Users can monitor their consumption through the Gemini interface. On the web app, click the settings icon in the lower left corner and select Usage limits. Mobile users should tap the menu button, then settings, then Usage limits.

The interface displays two progress bars. The first shows your current usage against a limit that resets every five hours. The second tracks your weekly quota. If you exhaust your limits on a paid plan, Gemini demotes you to the most basic model until the next reset period rather than blocking access entirely.

These details were first reported by WIRED, which noted that Google's support documentation acknowledges limits may change without notice due to capacity constraints.