The latest buzz in the tech world is all about OpenAI’s GPT-5, a model lauded for its unprecedented ability to handle complex tasks. However, away from the spotlight, a group of researchers is highlighting a critical, and unaddressed, issue: the model’s prodigious energy consumption. In a field where official data is scarce, these independent efforts are revealing a significant and troubling trend, where each new generation of AI demands more power, pushing the industry’s environmental footprint to new and potentially unsustainable levels.
The scale of this issue is quantified by researchers at the University of Rhode Island. Their benchmarking of GPT-5 found that a single, medium-length response from the model uses an average of 18 watt-hours. To put this in context, that’s enough energy to keep an incandescent light bulb on for 18 minutes. Given the sheer volume of queries that a service like ChatGPT processes—reportedly 2.5 billion per day—this individual cost balloons into a national-level concern. The total daily consumption could meet the electricity needs of 1.5 million US homes, underscoring the massive scale of AI’s power requirements.
The reason for this surge is straightforward: model size. While OpenAI has not released the parameter count for GPT-5, experts are confident it is substantially larger than its predecessors. This is a critical factor, as a study from the French AI company Mistral has shown a “strong correlation” between a model’s size and its energy consumption. A model that is ten times bigger, the study found, will have an impact one order of magnitude larger. This suggests that the trend of making larger, more capable models, which OpenAI’s CEO has championed, comes with a predictable and escalating environmental price.
Adding to this power demand are the new capabilities of GPT-5. While the model employs a “mixture-of-experts” architecture to improve efficiency, its ability to reason and to process video and images requires more intense computation. This “reasoning mode” can significantly increase the time and resources needed for a response, with some experts suggesting a five to ten-fold increase in resource use for the same answer. This combination of size and advanced functionality makes GPT-5 a landmark in AI capability, but also in its environmental impact, leading to urgent calls for greater transparency from its creators.