Developers
API
One endpoint, free, CORS-open, no key. It returns the same numbers the calculator shows, with all three scenarios and the coefficient version. We ask for a link back to aiwateruse.org when you publish figures from it.
Endpoint
GEThttps://aiwateruse.org/api/v1/estimate?model=claude-sonnet&tokens=86000000&scenario=mid
Parameters
- model
- Site id (claude-sonnet) or API name (claude-sonnet-4). Required. Invalid values return the full model list.
- tokens
- Total tokens: 86000000, 86,000,000 and 86M all parse. For image/video models this is the generation count.
- prompts / period
- Alternative to tokens: prompts per day, with optional period=day|month|year (default day).
- scenario
- low | mid | high (default mid). Selects which scenario the equivalents block describes; all three are always returned.
Attribution
Free for any use. When you publish numbers from this API, link to aiwateruse.org. It keeps the data maintained.
Example
{
"input": {
"model": "claude-sonnet",
"modelLabel": "Claude Sonnet",
"class": "standard",
"units": 86000000,
"unitKind": "tokens",
"scenario": "mid"
},
"energyKwh": {
"low": 34.4,
"mid": 68.8,
"high": 129
},
"water": {
"low": {
"liters": 20.64,
"gallons": 5.45251
},
"mid": {
"liters": 344,
"gallons": 90.8752
},
"high": {
"liters": 2064,
"gallons": 545.251
}
},
"carbonKg": {
"low": 3.44,
"mid": 24.08,
"high": 34.4
},
"equivalents": {
"scenario": "mid",
"water": [
{
"id": "almond",
"count": 86,
"sediment": true,
"kind": "water"
},
{
"id": "toilet-flush",
"count": 57.3333,
"sediment": false,
"kind": "water"
},
{
"id": "coffee",
"count": 2.64615,
"sediment": true,
"kind": "water"
},
{
"id": "home-day",
"count": 1.10968,
"sediment": false,
"kind": "water"
}
],
"energy": [
{
"id": "ev-mile",
"count": 229.333,
"sediment": false,
"kind": "energy"
},
{
"id": "household-day",
"count": 2.37241,
"sediment": false,
"kind": "energy"
}
],
"carbon": [
{
"id": "car-mile",
"count": 60.2,
"sediment": false,
"kind": "carbon"
},
{
"id": "tree-year",
"count": 1.14667,
"sediment": true,
"kind": "carbon"
}
]
},
"meta": {
"coefficientsVersion": "1.3.1",
"scenarios": {
"low": "cooling only",
"mid": "cooling + electricity generation",
"high": "full lifecycle: + training amortization + hardware manufacturing"
},
"note": "Estimates, not measurements. Energy band reflects benchmark variance; water/carbon are computed from mid energy times the scenario factor.",
"methodology": "https://aiwateruse.org/methodology",
"attribution": "Please link to https://aiwateruse.org when citing these figures."
}
}