GPT-4 Turbo: cost, energy and CO₂ per token
This model has been retired, or is being retired, by its provider (shutdown on 23 October 2026). The sheet stays online for reference and to quantify past usage; the current range is on the model sheets page.
What is the footprint of GPT-4 Turbo? GPT-4 Turbo emits about 2.3 kg of CO₂e, uses 26.8 L of water and 5.2 kWh of electricity per million generated tokens. Per million input tokens: 46 g of CO₂e, 543 mL of water, 105 Wh. Public price: €26 per million generated tokens, €9 for input.
Factors tokenclimate-v4-2026-07, updated on , high confidence. Sources and equations in the methodology.
Key figures, per million tokens
| Energy | CO₂e | Water | Price | |
|---|---|---|---|---|
| Input tokens | 105 Wh | 46 g | 543 mL | €9 |
| Generated tokens | 5.2 kWh | 2.3 kg | 26.8 L | €26 |
One million generated tokens represents about 16 km in a combustion car.
Frequently asked questions
How much CO₂e for 1 million tokens with GPT-4 Turbo?
About 2.3 kg per million generated tokens and 46 g per million input tokens (factors tokenclimate-v4-2026-07 of the TokenClimate registry).
Where do these figures come from?
Factors published in TokenClimate's versioned registry, each with its citation on this page. The full equations and parameters are in the public methodology, at tokenclimate.com/en/methodology.
What is the confidence level?
High confidence: published estimate specific to this model, fitted on its own points.
How can you measure your organisation's real usage?
The TokenClimate dashboard connects to your organisation's Anthropic, OpenAI or OpenRouter admin keys (stored encrypted) and measures cost and CO₂e per team and per model. You can see it in the demo: tokenclimate.com/demo. Locally, the open source plugin claude-carbon measures a Claude Code session.
Sources and method
Energy per token
The per-token energy factor behind the figures above, published with its citation. (Jegham et al., arXiv:2505.09598 (v6))
Infrastructure parameters
Datacenter parameters applied to this provider (the equations live in the methodology). (Jegham et al., arXiv:2505.09598 (v6))
| PUE | 1.12 |
| Grid carbon intensity (CIF) | 0.35 kgCO₂e/kWh |
| On-site water (WUE) | 0.3 L/kWh |
| Off-site water (EWIF) | 4.35 L/kWh |
| Amortised manufacturing (embodied) | 49 gCO₂e/kWh |
Price
Public provider prices. (developers.openai.com, pricing)
Full equations and assumptions in the methodology.