Docs

How it works, where your data lives, and every command. The full manual is also built into the dashboard at 127.0.0.1:8788/#help.

🧭 How it works

🔀

proxy

Your apps send their AI calls here. It forwards each call unchanged and notes the model and token counts.

⚡

agent

Every 10 seconds it reads the chip's power and which apps use it, so a local model's energy is measured.

📈

dashboard

Turns the counts into costs with your prices and electricity rate.

Supported providers

OpenAI, Anthropic, Google Gemini, Mistral, Groq, Together AI, Fireworks AI, OpenRouter, DeepSeek and xAI, plus local llama.cpp and Ollama. Add any other with aimeterix providers add.

💾 Where your data lives

What	Where
Database	`~/Library/Application Support/aimeterix/meter.db` SQLite by default; PostgreSQL and DuckDB work too
Settings	`~/Library/Application Support/aimeterix/aimeterix.yaml`
Logs	`~/Library/Logs/aimeterix`

Kept: model names, token counts, timings, energy
Never kept: prompts, answers, API keys
Nothing leaves your Mac except your own calls to the provider you chose

🛠️ Commands

Run

Start everything in the background

$ aimeterix start

Stop everything

$ aimeterix stop

Check

Is everything running and measuring?

$ aimeterix health

Today's usage and costs in Terminal

$ aimeterix usage

What this Mac can measure

$ aimeterix doctor

Where the data, settings and logs are

$ aimeterix paths

Costs and settings

Your electricity price per kWh

$ aimeterix profile set --rate 0.30

The AI providers and their proxy addresses

$ aimeterix providers

Load public token prices

$ aimeterix prices apply-reference

Change ports or database before the first start (optional)

$ aimeterix init

Remove

Stop and delete all settings and data

$ aimeterix uninstall --purge

Help

Every command and option

$ aimeterix --help