"Notice given: Energy usage for each AI query is disclosed by ChatUI"
In a groundbreaking move, Hugging Face, a leading company specialising in open-source AI applications, has unveiled ChatUI - an experimental tool designed to estimate the energy footprint of messages written using AI models like Meta's Llama 3.3 70B and Google's Gemini 3. The project is part of a broader transparency initiative within the open-source community, with the ultimate goal of making the energy consumption of digital tools as accessible as nutritional information on food.
Julien Delavande, an engineer at Hugging Face, is the brainchild behind ChatUI. The tool works by comparing the energy consumption of AI prompts to that of common household appliances, providing users with an easy-to-understand measure of the environmental impact of their digital interactions. For instance, generating an email via Llama 3.3 70B would consume approximately 0.1841 Wh, equivalent to a microwave running for 0.12 seconds.
ChatUI is integrated into an interface and displays the energy consumption in watt-hours or joules for each prompt, offering users a clear picture of their digital footprint. However, it's important to note that as an experimental tool, ChatUI's estimates may not be perfectly accurate.
Small energy savings from the choice of model or the length of the response can have a significant environmental impact, according to Delavande and his co-creators. The initiative is modeled after the Nutriscore system, which rates food based on its nutritional value, aiming to make the energy consumption of digital tools as visible as nutritional information on food and its ecological impact.
While the accuracy of ChatUI’s energy consumption estimates for Llama 3.3 70B and Gemini 3 is not publicly documented or independently validated in the current literature or release notes, users are advised to treat these estimates as approximate unless corroborated by hardware-level measurements or dedicated benchmarking studies.
If precise energy consumption figures are critical for your use case, you may need to complement ChatUI estimates with direct monitoring tools or specialized benchmarking frameworks designed for AI workloads. The initiative is a significant step towards increased transparency in the energy consumption of digital tools, encouraging users to make informed decisions about their digital footprint.
Artificial-intelligence models like Meta's Llama 3.3 70B and Google's Gemini 3 are measured for their energy consumption by the ChatUI tool, developed by Julien Delavande at Hugging Face, which provides users with an approximation of the environmental impact of their digital interactions. This initiative, modeled after the Nutriscore system, strives to make the energy consumption of digital tools as visible as nutritional information on food, thereby encouraging users to make informed decisions about their digital footprint.
