Google and OpenAI have introduced new daily caps on free image and video generation as soaring global usage strains computing resources. The restrictions aim to balance accessibility, manage GPU load, and maintain service quality for millions of users.
The measures reflect a significant uptick in demand for generative AI tools, notably following the recent excitement around Google’s Nano Banana Pro image model and OpenAI’s Sora video generator. As usage surged, both companies found their current computing capacity inadequate to meet the growing user base.
OpenAI has implemented a limit of six free video generations per day for users of its Sora tool. Paying subscribers under the ChatGPT Plus and Pro plans will continue to enjoy unrestricted access, while additional video generations can be purchased as needed. Bill Peebles, Head of Sora at OpenAI, confirmed the new restrictions, stating on X that the infrastructure was under significant strain, noting that the system’s GPUs were “running hot.” He emphasized that the adjustments would facilitate broader access to the technology without service disruption.
Google restricts free image creation on Nano Banana Pro
Similarly, Google has restricted its Nano Banana Pro image generation for free users, limiting them to two image generations or edits per day. This model, driven by the Gemini 3 Pro Image, was launched with high expectations, which resulted in unexpectedly high traffic immediately following its release.
In conjunction with these restrictions, Google has introduced variable limits for the broader Gemini 3 Pro tier. The company’s support page now indicates that “daily limits may change frequently” for free users accessing the Gemini 3 Pro Thinking capabilities. However, users of the earlier Nano Banana version, which operates on the less demanding Gemini 2.5 architecture, still have the capacity to generate up to 100 images daily.
Compute pressure highlights AI’s growing resource demands
These recent changes come during a holiday weekend in the United States, a time typically characterized by heavy usage as users explore creative applications. Both tech giants have stated that these restrictions underline the considerable computational and energy resources required for high-quality image and video generation.
OpenAI’s CEO Sam Altman recently clarified prior comments about “melting GPUs,” explaining that while processing units can indeed run intensely under heavy loads, modern data centers are equipped with advanced cooling systems designed to prevent actual damage.
The newly implemented limits, which are expected to be temporary for some users, illustrate the ongoing challenges AI companies face as they scale increasingly sophisticated generative systems for a rapidly expanding global audience. As both Google and OpenAI adapt to meet the rising demands, the adjustments reflect not only the immediate need to optimize resources but also signify the broader implications for the future development of generative AI technologies.
See also
Trump’s Draft EO Aims to Override State AI Laws, Centralizing Federal Control Over AI Regulation
Google Deletes Infographic After Plagiarism Allegations from Food Blogger; Faces Backlash Over AI Practices
Accenture Partners with OpenAI to Upskill 50,000 Professionals with ChatGPT Enterprise
AI Pioneer Fei-Fei Li’s World Labs Achieves $1B Valuation, Pioneering Spatial Intelligence
Meta Platforms Surges 8% on Strong AI Investments, Targeting $841.42 Fair Value
