AI Meter — Measure tokens. Estimate energy and water.

Private by default

Measure your AI usage. Locally.

AI Meter measures token usage across your local coding agents and estimates the associated electricity and direct cooling-water use. No accounts. No usage data leaves your Mac.

INSTALL FROM TERMINAL

$ curl -fsSL https://ai-meter.app/install.sh | sh

Reads local provider files. The only network request is a public update check.

AI Meter macOS widget showing local token usage, human-scale electricity and water equivalents, Cursor support, and an interactive usage chart

TRANSPARENT ESTIMATION

From measured tokens to estimated impact.

AI Meter follows a SCI for AI-aligned consumer boundary. These are local scenarios based on published research—not measurements from OpenAI, Anthropic, or their data centers.

01 / USAGE#

Provider-reported tokens

Codex, Claude Code, Cursor, OpenCode, and Gemini CLI token counts are read from local records, deduplicated, and combined on your Mac.

02 / ELECTRICITY⚡

Tokens ÷ 1M × 0.39 kWh

The default inference factor is 0.39 facility kWh per million tokens, with an indicative 0.20–0.75 kWh uncertainty range.

03 / WATER◒

Energy ÷ PUE × WUE

Direct cooling water uses a 1.20 PUE and 0.45 L per IT kWh site WUE. Both assumptions are adjustable inside AI Meter.

What is excluded: model training, embodied hardware, off-site electricity water, grid carbon, networking, and user-device energy.

Read the full methodology