v0.1.3 preview · open source · macOS

Know what your AI
really costs.

AI Meterix measures the energy your Mac spends on AI, counts every token each model uses, and turns both into a clear daily cost. All on your machine.

InstallView on GitHub ↗
$ curl -fsSL https://github.com/apipas/aimeterix/releases/latest/download/install.sh | sh
The AI Meterix dashboard: cost, energy, tokens and calls for the last 7 days

Built for people who ship with AI

Developers, indie makers and small teams who run local models or pay for cloud AI APIs.

⚡

Energy

Reads your chip's power every 10 seconds and splits it by app and by local model (llama.cpp, Ollama).

🔢

Tokens

A local proxy counts input, output and cache tokens of every call, streaming included. Ten cloud providers built in.

💶

Cost

Turns energy and tokens into one daily cost with your electricity rate and dated public token prices.

How it works

Three small services on your Mac. Your apps keep their own API keys.

Your appsAI calls
Mac chipCPU, GPU, power
→
proxy127.0.0.1:8787
agentevery 10 s
→
Databaseon your Mac
Dashboard127.0.0.1:8788

Read the docs →

Private by design

Your data stays on your Mac.

  • Kept: model names, token counts, timings, energy
  • Never kept: prompts, answers, API keys
  • Local only: everything listens on 127.0.0.1
  • No account, no telemetry, no update check