OpenAI’s GPT-5 has arrived, bringing with it a suite of new, powerful capabilities. But as the tech world marvels at its performance, a group of researchers is sounding the alarm about its immense energy consumption. With no official data from the company, these independent findings are providing a crucial glimpse into the model’s environmental footprint, revealing a dramatic increase in power demands that raises serious questions about the long-term sustainability of the AI industry.
The numbers are eye-opening. According to a team at the University of Rhode Island’s AI lab, a medium-length response from GPT-5 consumes an average of 18 watt-hours. This is a substantial increase from previous models and is “significantly more energy than GPT-4o,” according to a researcher in the group. To put this in perspective, 18 watt-hours is enough to power an incandescent light bulb for 18 minutes. Given that ChatGPT handles billions of requests daily, the total energy consumption could reach the daily electricity demand of 1.5 million US homes, a staggering figure that underscores the scale of the problem.
A key factor driving this dramatic increase is the model’s size. While OpenAI has not released the parameter count for GPT-5, experts believe it is “several times larger than GPT-4.” This is consistent with a study from the French AI company Mistral, which found a “strong correlation” between a model’s size and its resource consumption. The study concluded that a model ten times bigger would have an impact one order of magnitude larger. This suggests that the trend of building ever-larger AI models will continue to drive up resource usage at an alarming rate.
The new capabilities of GPT-5 also play a significant role in its high energy demands. Its advanced “reasoning mode” and ability to process video and images require more intensive computation. A professor studying the resource footprint of AI models noted that using the reasoning mode could increase resource usage by a factor of “five to 10.” This means that while a “mixture-of-experts” architecture offers some efficiency, the new, more complex tasks are driving the overall energy footprint to new heights. This raises serious questions about the long-term sustainability of the AI industry and the need for greater transparency.