Google recently introduced a new compute-based usage system for its Gemini AI plans, shifting from a fixed prompt limit to a dynamic credit-style approach. Under this system, each task consumes a variable amount of the user's allowance depending on factors like prompt complexity, features used, and conversation length. The change was intended to make resource allocation more efficient, but it has instead sparked widespread frustration among subscribers.
The new limits apply primarily to Google AI Pro subscribers, who get a five-hour usage window that resets every five hours. Once that window is exhausted, users face a broader weekly cap. However, many are finding that the new system drains their allowance far quicker than expected, leading to premature lockouts.
The incident that caught attention
One Google AI Pro user, Ashutosh Shrivastava, took to social media to share a troubling experience. He reported that after providing a single prompt for video generation using Gemini's avatar feature, his entire five-hour rate limit was consumed within minutes. The video generation also failed, meaning he lost his entire allowance without getting a usable result. He posted a video demonstrating the rapid depletion: starting at 0% usage, the indicator jumped to 100% after just a few minutes of processing.
The complaint quickly gained traction on X (formerly Twitter), catching the eye of Josh Woodward, Google's Gemini lead. Woodward responded directly, saying "Yikes, let us take a look!" indicating that the company is aware of the issue and willing to investigate individual cases.
Understanding the compute-based system
Before this change, Gemini operated on a simpler model: a fixed number of prompts per time period. Users knew exactly how many queries they could make before hitting a limit. The new system replaces that with a variable credit model, where each prompt's resource consumption is determined by the underlying computation needed. Simple text queries may cost little, but tasks like video generation, large file analysis, or long conversations can consume credits rapidly.
While the intent is to align pricing with actual resource usage, the lack of transparency has become a major pain point. Users have no way to predict how many prompts they can make in a session, or which actions will drain the allowance quickly. This unpredictability undermines the premium experience that paying customers expect.
Broader community backlash
Complaints are not isolated. The Gemini subreddit has seen a flood of posts criticizing the new limits. Many users report similar experiences: after upgrading to the Pro plan, they find their usable time drastically reduced compared to the previous fixed limits. Some have even compared the new system unfavorably to competing AI services like ChatGPT Plus or Claude Pro, which offer more predictable usage tiers.
For example, ChatGPT Plus users have a fixed message limit per hour, which is clearly displayed. Similarly, Claude Pro offers a message cap that resets predictably. In contrast, Gemini's compute-based system leaves users guessing, causing frustration especially among power users who rely on the tool for work or creative projects.
Google's response and pressure to improve
Google has not yet publicly detailed any changes to the system, but the company's swift acknowledgment of Shrivastava's case suggests they are monitoring the feedback closely. Some users have noted that Google has already increased limits for certain "Antigravity" users, boosting quotas by up to 9x after a previous reduction. However, for most regular Pro subscribers, the caps remain tight.
The situation highlights a broader challenge for Google: balancing resource allocation with user experience. While compute-based pricing can be fairer in theory, it requires clear communication and real-time usage tracking. Without these, users feel nickel-and-dimed, especially when a single failed task wipes out an entire window.
Historical context and competitive landscape
Google's Gemini models, especially the Pro tier, are positioned as premium alternatives to OpenAI's GPT-4 and Anthropic's Claude. The initial launch faced criticism for limited availability and restrictive usage caps. Over time, Google increased limits, but the shift to a compute-based system appears to have undone some of that progress. Meanwhile, competitors have maintained more straightforward pricing: ChatGPT Plus charges $20 per month for a set hourly limit, and Claude Pro offers similar predictability.
The video generation capability is a relatively new feature, and its high computational cost may explain why it consumed the entire allowance. But as AI becomes more integrated into daily workflows, users demand reliability. A system where a single experimental prompt can exhaust a five-hour budget is seen as unacceptable for a paid service.
Analyzing the impact on users
For professionals and creators, time is money. Losing a full usage window to a failed attempt means lost productivity and potential revenue. The incident underscores the need for robust error handling and clear usage forecasts. Ideally, Gemini should warn users before engaging a high-cost operation, showing an estimate of the resources required. Currently, the interface does not provide such information.
Moreover, the lack of a rollback mechanism for failed tasks adds insult to injury. If a prompt fails due to a server error or model limitation, the user still forfeits the consumed credits. This is in stark contrast to cloud computing services like AWS, where failed instances may be refunded or excluded from billing.
Google could improve the system by implementing a credits-exhausted notification that allows users to pause or cancel high-cost operations. Another solution would be to provide a daily or weekly allowance breakdown, showing how many credits each type of task costs. Transparency would go a long way toward rebuilding trust.
As the AI industry matures, user expectations evolve paying customers want predictable, reliable access. Google's misstep here serves as a cautionary tale for other companies adopting dynamic pricing. The feedback loop is clear: Google is facing increasing pressure to adjust Gemini's usage limits, either by loosening restrictions, offering more granular controls, or reverting to a fixed prompt model. The company's next move will be watched closely by both users and competitors.
Source: Android Authority News