Prompt Power

Every AI prompt has a hidden cost. Prompt Power makes it visible.
Prompt Power runs silently in the background as you use Claude, ChatGPT, Gemini, and Perplexity — tracking the energy consumed, water used, and tokens processed with every
conversation.
What you get:
- Real-time token counting for every prompt and response
- Energy estimates (Wh) based on published AI research data
- Water usage tracking for data center cooling
- Relatable comparisons — see your usage in seconds of a hair dryer, minutes of a TV, or number of Google searches
- Platform breakdown showing which AI tools consume the most
- Full history of every interaction
How it works:
- Detects prompts and responses directly on each AI platform
- Estimates tokens using a client-side heuristic — no data leaves your browser
- Calculates energy using per-token rates from peer-reviewed research (Luccioni et al., 2023)
- Water estimates based on UC Riverside data center cooling research
Why it matters:
A single AI query uses 10x more energy than a Google search. Image generation uses 60x more. AI data centers are projected to consume as much electricity as entire countries by
2027, using billions of liters of fresh water for cooling.
This isn't about stopping — it's about being aware. Use AI intentionally.
Prompt Power is 100% local. No accounts. No tracking. No data sent anywhere. Your usage stays on your device.
Supports: Claude (claude.ai), ChatGPT (chatgpt.com), Gemini (gemini.google.com), and Perplexity (perplexity.ai)