> ## Documentation Index
> Fetch the complete documentation index at: https://docs.prisme.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Carbon Footprint

> How the LLM Gateway estimates the energy and CO₂eq impact of every model call

Every completion served through the LLM Gateway is scored for its environmental impact: the energy it consumed (kWh) and the greenhouse-gas emissions it produced (kg CO₂eq). The estimate is attached to each call and aggregated in [Observability](./observability) (the **Carbon** headline card).

The method is **inspired by the [EcoLogits](https://ecologits.ai) methodology**: a per-token energy model, scaled by datacenter efficiency (PUE) and converted to CO₂ with a region-specific grid emission factor, reported with an uncertainty range. It is a **parametric estimate, not a measurement** — its accuracy depends entirely on the metrics declared on each model.

## Where the inputs come from

The estimate is driven by three fields you declare per model, in the **Metrics** section of the model declaration form (Governe → Models → Add/Edit model → Advanced):

<Frame>
  <img src="https://mintcdn.com/prismeai/BM7eLwrJPD0hfd-J/images/ai-governance-model-metrics.png?fit=max&auto=format&n=BM7eLwrJPD0hfd-J&q=85&s=c17a7523c0aa7eb6d84ed747e26f3214" alt="Model declaration — Metrics section (Energy/Token, PUE Profile, Eco Score)" width="1284" height="696" data-path="images/ai-governance-model-metrics.png" />
</Frame>

| Field                                           | Feeds               | Meaning                                                                                                                |
| ----------------------------------------------- | ------------------- | ---------------------------------------------------------------------------------------------------------------------- |
| **Energy/Token** (`energy_per_token`)           | GPU energy          | Energy drawn per output token, in **kWh/token**. Required — without it, no carbon estimate is produced (only pricing). |
| **PUE Profile** (`pue_profile`)                 | Datacenter overhead | `efficient`, `average` (default) or `inefficient`.                                                                     |
| **Region** (`region`)                           | Emission factor     | Grid carbon intensity of where the model runs. Defaults to `world`.                                                    |
| **Eco Score**, **Cost Tier**, **Training Date** | Display only        | Not used in the calculation.                                                                                           |

<Warning>
  **Unit matters.** `Energy/Token` is expressed in **kWh per token** (e.g. `0.0000103` kWh/token ≈ 10.3 mWh/token). Entering a value in Wh/token would make the reported footprint 1000× too high. The field is not unit-suffixed in the UI, so declare it in kWh/token.
</Warning>

## Calculation

For a call with `in` input tokens and `out` output tokens, and a model whose declared energy per token is `ept`:

### 1. GPU energy

Input tokens (encoding) are counted at \~33% of an output token's energy:

```
gpu_energy = (in × ept × 0.33) + (out × ept)        [kWh]
```

### 2. Server overhead

A fixed server draw of 0.1 kW over an estimated 0.02 s of processing per token:

```
processing_time = (in + out) × 0.02                 [s]
server_energy   = 0.1 kW × processing_time / 3600   [kWh]
```

### 3. Datacenter efficiency (PUE)

The raw energy is scaled by the Power Usage Effectiveness of the datacenter:

```
total_energy = (gpu_energy + server_energy) × PUE   [kWh]
```

| PUE Profile         | Multiplier |
| ------------------- | ---------- |
| `efficient`         | 1.1        |
| `average` (default) | 1.58       |
| `inefficient`       | 2.0        |

### 4. CO₂eq emissions

The energy is converted with the grid emission factor of the model's region:

```
gwp = total_energy × emission_factor                [kg CO₂eq]
```

| Region            | Emission factor (kg CO₂/kWh) |
| ----------------- | ---------------------------- |
| `world` (default) | 0.475                        |
| `eu-west`         | 0.276                        |
| `eu-north`        | 0.231                        |
| `us-east`         | 0.389                        |
| `us-west`         | 0.352                        |
| `asia`            | 0.638                        |

### 5. Uncertainty range

Both energy and emissions are reported with a **±20%** band (`min = value × 0.8`, `max = value × 1.2`).

### Condensed formula

```
E   = [ (in·ept·0.33 + out·ept) + (0.1 · (in+out)·0.02 / 3600) ] × PUE
CO₂ = E × emission_factor
```

## Worked example

A short call of 51 input + 18 output tokens on a model declared at `ept = 0.0000103` kWh/token, `average` PUE, `us-west`:

* `gpu_energy` = (51 × 0.0000103 × 0.33) + (18 × 0.0000103) ≈ **0.000359 kWh**
* `server_energy` = 0.1 × (69 × 0.02) / 3600 ≈ **0.0000038 kWh**
* `total_energy` = (0.000359 + 0.0000038) × 1.58 ≈ **0.000573 kWh**
* `gwp` = 0.000573 × 0.352 ≈ **0.0002 kg CO₂eq** (\~0.2 g CO₂eq)

## Result shape

Each call returns:

```json theme={null}
{
  "energy": { "value": 0.000573, "min": 0.000459, "max": 0.000688, "unit": "kWh" },
  "gwp":    { "value": 0.000202, "min": 0.000161, "max": 0.000242, "unit": "kgCO2eq" },
  "details": { "gpu_energy": 0.000359, "server_energy": 0.0000038, "pue": 1.58, "emission_factor": 0.352, "region": "us-west" }
}
```

<Note>
  The constants (0.33 input factor, 0.1 kW server draw, 0.02 s/token, PUE multipliers and regional emission factors, ±20% band) are fixed in the gateway. Only `energy_per_token`, `pue_profile` and `region` are per-model. A model with no `energy_per_token` reports pricing only and no footprint.
</Note>

## Related

* [Observability](./observability) — where the aggregated Carbon (kg CO₂eq) card is surfaced
* [Available Models](./available-models) — the governed model catalog
* [Model Governance](./model-governance) — access, routing and failover controls