rakam-frugal-ai 2026.6.15

Carbon intensity data registry for LLM and embedding models with live Electricity Maps updates

rakam-frugal-ai

Carbon intensity data registry for LLM and embedding models, with live updates from the Electricity Maps API.

Ships a precomputed, weighted-by-region probability table for popular models (GPT-4o, GPT-4o-mini, Claude, embedding models, …) and refreshes the underlying datacenter carbon intensities on demand from a free Electricity Maps API key.

Install

pip install rakam-frugal-ai

Usage

import rakam_frugal_ai as ci

m = ci.model("gpt-4o")
print(m.CI)              # weighted carbon intensity, gCO2eq/kWh
print(m.size_billion_params)
print(m.datacenter_regions)

API key

Live refreshes require a (free) Electricity Maps API key. Set it in your environment:

export ELECTRICITY_MAPS_API_KEY=your-key-here

Without the key the library still works using the bundled snapshot; it just can't refresh values.

Python support

Python 3.10, 3.11, 3.12.

License

MIT — see LICENSE.