The era of seemingly unlimited free access to cutting-edge generative AI has abruptly ended. OpenAI and Google, the two titans of the artificial intelligence sector, have simultaneously imposed stringent new daily usage caps on their most computationally intensive tools OpenAI’s video generator, Sora, and Google’s latest image model, Nano Banana Pro.
The sudden restrictions, which disproportionately affect non-paying users, underscore a harsh reality: running state-of-the-art generative AI systems consumes monumental computing power and energy, straining even the world’s most sophisticated data centers.
Sora Limits Spark Viral ‘Melting GPUs’ Comment
The most dramatic cut came from OpenAI. Free users of its highly-anticipated video generation model, Sora 2, are now limited to just six video generations per day. This represents a drastic reduction from the estimated 30 or more generations users reported receiving in the weeks following the model’s public launch.
The severity of the constraint was confirmed by Bill Peebles, the head of Sora at OpenAI, who took to the social media platform X to explain the change. “Our gpus are melting, and we want to let as many people access sora as possible!” he posted, adding, “we’re setting usage limits for free users to 6 gens/day.”
While the “melting GPUs” comment is hyperbolic, it points to the genuine infrastructure crisis caused by AI video creation. Generating even a few seconds of high-fidelity video requires an exponential increase in Graphics Processing Unit (GPU) time and electricity compared to simple text or even static image generation. The new limit is a clear attempt to ration this precious resource, especially over the holiday period, which has seen usage spike.
Crucially, the new caps do not apply to paying subscribers of ChatGPT Plus and Pro, confirming the company’s dual strategy: use the free tier as a powerful demonstration platform while protecting the premium experience for its paying customer base. Paid users can also purchase additional video generation tokens as needed.
Google Tightens Access to Nano Banana Pro
Google, meanwhile, followed a quieter but equally significant path in tightening limits across its Gemini AI ecosystem.
Free users of Nano Banana Pro, the image generation and editing tool powered by the new Gemini 3 Pro architecture, have seen their daily allowance cut to just two generated images per day. This marks a reduction from the previously reported three images per day and a massive drop compared to the older Nano Banana model (powered by Gemini 2.5), which still allows for up to 100 images per day for free.
The company’s support pages offer a less dramatic, but equally telling, explanation, citing “high demand” for image generation and editing, and warning that “limits may change frequently and will reset daily.”
The restrictions extend beyond images. Access to the core conversational abilities of Gemini 3 Pro has also been scaled back for non-paying users. Where Google initially offered a clear allocation of free prompts per day, the company has now replaced this guarantee with ambiguous language, describing free usage as “basic access” where “daily limits may change frequently.”
The Economics of Generative AI
The synchronicity of these rollbacks by the two market leaders highlights the economic friction inherent in offering cutting-edge AI for free.
“The computational cost is the most immediate reason for these limits,” explains Dr. Lena Chen, an AI economist at the Future of Tech Institute. “Generating a single complex video with Sora is effectively subsidizing a huge chunk of GPU time for a non-paying user. As demand grows, that cost becomes unsustainable for the company if it wants to maintain service quality for everyone.”
The move is widely seen as a commercial pivot, transforming generative AI from an experimental “free buffet” into a more clearly monetized service. By limiting the free tier, both companies are creating a strong incentive for heavy users artists, content creators, and professionals to migrate to paid subscription models, which offer substantially higher, and more stable, usage ceilings. For these users, the value proposition is simple: pay for priority access and guaranteed compute resources.
For the casual user, the new limits force more careful rationing of daily allowances. While two images or six videos might be enough for light experimentation, creators who rely on high-volume iteration will now find their creative workflows severely constrained unless they upgrade their subscription.
The new limitations are a clear signal that the massive operational costs associated with powerful generative models have finally begun to outweigh the benefits of offering unrestricted free access, ushering in a more sustainable, if less generous, phase in the consumer AI landscape.
