Google’s New Gemini Compute Limits Spark User Backlash Over Rapid Usage Caps- Users of Google’s AI services are raising concerns after the company introduced a new compute-based usage system for its Gemini platform, with some subscribers reporting that their allotted usage is being exhausted far faster than expected. The change has sparked frustration, particularly among paid users who say the new model is harder to predict and manage.
One Google AI Pro subscriber shared video evidence showing that a single failed video-generation prompt consumed nearly an entire five-hour usage allowance within minutes, raising questions about how compute costs are being calculated and displayed. The user claimed that even unsuccessful outputs were being heavily charged under the new system.
Google has acknowledged the complaints and said it is reviewing the issue, though it has not yet announced any changes to the rollout or pricing structure.
Shift to Compute-Based Usage Model
The recent update marks a significant shift in how Google allocates access to its AI tools. Instead of a fixed number of prompts or messages, the new system is based on “compute usage,” which measures the processing power required for each interaction.
Under this model, more complex tasks—such as long conversations, image generation, or video creation—consume more resources than simple text queries. The idea, according to Google, is to better align usage limits with the actual cost of running advanced AI workloads.
However, users argue that the system lacks transparency and makes it difficult to predict how quickly their limits will be used up, especially for creative or experimental prompts.
Why Users Are Frustrated
The backlash largely stems from unpredictability. Under the old system, users had a clearer sense of how many prompts they could send before hitting a cap. The new compute-based model, however, varies depending on the complexity of each request.
Some users report that even minor experimentation with multimodal features—such as image or video generation—can significantly reduce available usage time. Others say that failed or incomplete outputs still consume a large portion of their quota, which has raised concerns about fairness and efficiency.
Paid subscribers, particularly those on AI Pro plans, argue that the change feels like a downgrade in usability despite unchanged subscription pricing.
Google’s Response
Google has confirmed awareness of the issue and said it is investigating user reports. While the company has defended the new system as a more accurate reflection of computational costs, it has not ruled out adjustments to improve transparency or user experience.
The company has been rolling out updates across its AI ecosystem, gradually integrating compute-based limits into different tiers of its Gemini offerings. Google maintains that the system is designed to ensure fair access during periods of high demand while supporting more advanced features.
Broader Context: AI Monetisation Challenges
The controversy highlights a broader challenge facing AI companies: balancing advanced capabilities with sustainable pricing models.
As AI systems become more powerful—handling video generation, long-context reasoning, and multimodal inputs—computational costs increase significantly. Companies like Google are experimenting with usage-based models to manage infrastructure demands, but these systems can feel unpredictable to end users.
Industry analysts say this is part of a wider transition phase in AI monetisation, where firms are still searching for pricing structures that reflect real compute usage without alienating users.
What Comes Next
For now, users are waiting for clarification on how compute usage is calculated and whether failed or partial outputs will continue to count against quotas in the same way. Transparency and predictability are expected to be key demands if Google hopes to maintain trust among paid subscribers.
As competition in the AI space intensifies, user experience and pricing clarity could become just as important as model performance—especially as platforms like Gemini, ChatGPT, and others continue expanding into high-cost features like video and multimodal generation. Mushrooms as the New Meat Alternative: The Plant-Based Revolution | Maya
