We quantify CO₂-equivalent emissions based on the energy used during model inference, including compute cycles, server cooling, and a proportional share of datacenter infrastructure. These are quantified values, not metered measurements.

Our quantification methods are based on published efficiency benchmarks and cloud provider data. Exact values depend on infrastructure variables that are not always publicly disclosed. Actual carbon emissions for AI usage may vary.

It means the carbon is not just offset on paper, but permanently prevented from leaking into the atmosphere through independently audited methane projects with publicly retired credits on recognised registries.

CrbonFree AI is currently in beta. The core chat and tracking features are free. Carbon elimination is a voluntary action priced at cost.

Crbonfree.ai is currently designed for individual users to use and track personal AI usage and support carbon elimination in amounts that match their actual AI carbon footprint. Crbon Labs is developing tools to track and offset AI for use by small and large organizations. If you are interested in solutions for your organization to understand and offset its carbon footprint from AI usage, check out beta.crbonfree.com.