How much energy⇄ does AI inference use?

This dashboard presents a new methodology for understanding the environmental impact of closed AI models. It combines the latest peer-reviewed science, new empirical testing across 1,700+ tests and 230 million generated tokens, and real-world application to corporate carbon inventories.

CLEER at the core

CLEER (Closed-model Latent Energy Estimation Range) estimates per-token energy for proprietary models, where direct measurement is not possible. We benchmark open models on known hardware, match each closed model to the closest open proxies based on observed speed and behaviour, then project onto those measured power curves. → HHAI

Use case

CLEER estimates energy per token, the unit used in a real inventory. → Reach out for per-token factors.

Because tokens are difficult to interpret, this dashboard reports whole sessions. The 2 profiles draw on public production data: chat uses ShareChat conversations and Qwen-Bailian serving traces; agentic uses AgentX, a corpus of real coding-agent sessions. Every model receives the same workload, so the differences shown reflect the model, not usage.

Verbosity differs between models. Asked the same question, models generate different numbers of tokens, and a more verbose model uses more energy in practice. This dashboard holds token counts constant across every model so the comparison isolates per-token energy. Real-world totals will differ by how much a model tends to write.

From chip to facility

A model's accelerator energy is only part of a data center's energy use. Reported figures account for the host server, provisioned-but-idle capacity, and cooling and power distribution. These factors vary by operator and site. The methodology can accommodate different assumptions for each; this dashboard shows 1 documented scenario. → Emissions methodology

Carbon

Carbon emissions depend on how and where electricity is generated, with results varying by more than 10× across plausible grids. The methodology is designed to work across different energy and emissions assumptions; the dashboard shows 1 scenario: behind-the-meter gas generation at 730 gCO₂e/kWh, used in place of location-specific grid data, together with US-average facility efficiency and embodied emissions from hardware and data center construction. Different grid assumptions can produce substantially different results. → Emissions methodology