Sora and Nano Banana AI Tools Receive New Daily Limits as Demand Rises
Interest in the latest image and video generation tools has surged, prompting both companies to impose daily usage limits. As demand continues to rise, users are now facing new caps on how many generations they can produce. Free Sora users are limited to six videos per day, while paid subscribers maintain their existing allowances. At the same time, free users of Nano Banana Pro can generate or edit only two images per day, although earlier versions of the tool still allow up to 100 images daily.
These adjustments reflect a noticeable increase in system load. They also show how quickly user activity can push generation tools toward their operational limits, especially during high-traffic periods.
Increased Pressure on GPU Infrastructure
The new limits highlight how heavily these models rely on GPUs to process large amounts of data. Although such hardware is designed to operate under significant strain, prolonged spikes in activity still require careful management. With rising user engagement, systems must balance performance and availability, which often leads to temporary restrictions.
This situation also underscores the broader reality of modern AI tools: generating images and videos requires substantial compute and energy. As demand shifts, companies may adjust usage policies to maintain stability.
Variable Limits and User Impact
Nano Banana Pro’s updated rules include variable restrictions on Gemini 3 Pro access for non-subscribers, with the support page noting that daily limits may change frequently. This flexibility suggests that the company expects ongoing fluctuations in usage patterns. The timing of these changes coincides with a holiday period in the United States, when many users may have additional leisure time to experiment with generative features.
Although the limits may feel restrictive, they demonstrate how rapidly the adoption of creative AI tools is growing. They also show how providers must continuously balance user enthusiasm with the practical demands of running compute-intensive systems.
