Media
1 item
Description
Estimate the carbon emissions of your ChatGPT usage in real time. Bring environmental awareness to your AI conversations: stay informed and empowered when using the tools you love. ChatGPT Carbon Estimator (CBE) helps you understand the environmental impact of each interaction with ChatGPT by providing real-time estimates of carbon emissions, water usage, and relatable real-world equivalents. Whether you're casually chatting or deeply exploring ideas with ChatGPT, this lightweight extension makes it easy to stay mindful of your digital footprint. Generative AI models use large-scale compute resources and energy. By visualizing the impact of your queries, ChatGPT Carbon Estimator helps raise awareness around AI sustainability, without adding friction. CBE is ideal for: 🖖Curious users who want more sustainable tech habits 🖖Educators teaching digital responsibility 🖖Developers and researchers exploring AI’s real-world costs 🖖Anyone who uses ChatGPT regularly What CBE does: 🖖Automatically calculates carbon emissions for each ChatGPT response 🖖Shows real-time equivalents: - Water consumed (in liters) - Emails sent - Car travel distance - Flight time - Google searches 🖖Tracks total emissions per session — live in your browser 🖖 Displays a clean, light-blue badge right on the ChatGPT page CBE is privacy-friendly and runs entirely in your browser. 🖖No tracking 🖖No data collection 🖖No API keys required 🖖No OpenAI account access You see exactly what the model outputs and we use that alone to estimate the environmental cost. Methodology and sources: Bring environmental awareness to your AI conversations: stay informed and empowered when using the tools you love. ChatGPT Carbon Estimator (CBE) helps you understand the environmental impact of each interaction with ChatGPT by providing real-time estimates of carbon emissions, water usage, and relatable real-world equivalents. Whether you're casually chatting or deeply exploring ideas with ChatGPT, this lightweight extension makes it easy to stay mindful of your digital footprint. Generative AI models use large-scale compute resources and energy. By visualizing the impact of your queries, ChatGPT Carbon Estimator helps raise awareness around AI sustainability, without adding friction. CBE is ideal for: 🖖Curious users who want more sustainable tech habits 🖖Educators teaching digital responsibility 🖖Developers and researchers exploring AI’s real-world costs 🖖Anyone who uses ChatGPT regularly What CBE does: 🖖Automatically calculates carbon emissions for each ChatGPT response 🖖Shows real-time equivalents: - Water consumed (in liters) - Emails sent - Car travel distance - Flight time - Google searches 🖖Tracks total emissions per session — live in your browser 🖖 Displays a clean, light-blue badge right on the ChatGPT page CBE is privacy-friendly and runs entirely in your browser. 🖖No tracking 🖖No data collection 🖖No API keys required 🖖No OpenAI account access You see exactly what the model outputs and we use that alone to estimate the environmental cost. Methodology and sources The assumptions are: 🖖 1 token = 0.0003 kWh on average for the most common model used today, GPT4o. Noting that, while GPT4o-mini is a smaller model and should use less energy, studies showed that it uses an A100-chip model (compared to H200 for GPT4o), which is less efficient and might counterbalance the energy savings of the model itself. 🖖 ChatGPT queries happen in Europe, which leads to a kg CO2 per kWh = 0.26 on average. This is an average across the whole region. In Australia, this variable would be 0.81. 🖖 A study from 2023 showed that running GPT-3-sized models consumes approximately 500 ml of water per 1,000 tokens. This includes both direct water use to cool the servers in the data center and indirect water use: water used by power plants to generate the electricity powering those servers 🖖On average 1 token = 1.33 words 💬 FAQ Q: Does this access my ChatGPT account? A: No. It only observes content rendered on the page. Q: Are these estimates accurate? A: They are approximate, based on the length of each response and published average energy metrics. The goal is not precision, but perspective. Q: Is my data stored or sent anywhere? A: No. Everything runs locally in your browser. No storage, no cloud.