Google has revised Gemini's usage limits, changing from counting quotas per command to billing based on actual processing volume.
At Google I/O 2026, Google announced a change in Gemini's usage model, shifting from usage limits counted per command to compute-used billing based on actual processing, similar to ChatGPT and Claude which had already moved away from per-command counting.
Under this new quota system, each Gemini use is not deducted per fixed prompt command but is assessed based on three main factors: the complexity of the prompt, the features used, and the length of the conversation.
For example, asking the AI to generate a video consumes more processing resources than a simple short text question. Therefore, users issuing basic commands will barely notice a difference, but heavy usage, such as video generation or coding, may reach limits more quickly.
Additionally, Google has introduced usage limits in 5-hour and weekly cycles. Heavy users will first encounter the 5-hour limit, followed by the weekly limit. Remaining quotas can be checked via the Gemini app or website.
Usage quotas vary by subscription package: free users have the smallest quota, AI Plus users get double that of free users, AI Pro users receive four times the free quota, and the premium AI Ultra package offers about 20 times the AI Pro quota.
When users exhaust their quota, Gemini switches to a smaller model. However, AI Pro and Ultra subscribers have the option to purchase pay-as-you-go credits to unlock limits and continue using high-end models. These credits will apply across Google Antigravity, Google Flow, and the Gemini app in the near future.