Your AI usage has
a water bill.

Pluvi is a water-neutral proxy for AI APIs. Swap one URL, and every inference call is measured, offset, and certified.

// before
baseURL: "https://api.openai.com/v1"

// after
baseURL: "https://api.pluvi.ai/v1"
Request access See how it works
~10ml of water per GPT-4-class query 6.4 billion gallons — Microsoft’s disclosed water use in FY2023 0 major AI providers carry a verified water-neutral credential today ~10ml of water per GPT-4-class query 6.4 billion gallons — Microsoft’s disclosed water use in FY2023 0 major AI providers carry a verified water-neutral credential today
How it works

Four steps to water-neutral AI.

01

Route

Point your API calls at Pluvi. Same models, same response shape, same auth. We proxy to the provider and pass everything through.

02

Measure

Every request is logged by model and token count. We estimate the water footprint using energy-per-token coefficients and regional cooling data.

03

Offset

A rolling offset is applied on your behalf through verified watershed restoration and wetland projects via Alliance for Water Stewardship partners.

04

Certify

You get a verifiable certificate and a badge you can embed in your product, docs, or ESG disclosures. Proof your AI is water-neutral.

The credential

A badge worth showing.

Embed a verified badge in your product, README, or compliance reports. Each badge links to a live, auditable certificate.

Water-neutral AI — verified by Pluvi
<img src="https://pluvi.ai/badge/your-id" />
Water-Neutral Certificate
AccountAcme Corp
PeriodMarch 2026
Queries48,291
Est. water footprint312.4 L
Water offset312.4 L
Offset projectAWS Watershed — Chesapeake Bay
StatusVerified
Pricing

Simple pass-through pricing.

Starter
+8%
Over pass-through model cost. For teams exploring water-neutral AI.
  • Up to 1M tokens/month
  • Water footprint dashboard
  • Verified badge
  • Monthly certificate
Get started
Growth
+6%
For production workloads that need a water-neutral credential at scale.
  • Up to 50M tokens/month
  • Water footprint dashboard
  • Verified badge
  • Monthly certificate
  • Priority support
Get started
Enterprise
Custom
For organizations with compliance requirements or high-volume inference.
  • Unlimited tokens
  • Custom offset projects
  • Dedicated account manager
  • SLA & compliance reports
  • SSO & audit logs
Talk to us
Methodology

Published, versioned, auditable.

footprint = tokens × energy_per_token[model] // kWh per token, model-specific × water_per_kwh[datacenter_region] // liters per kWh for cooling × stress_index[watershed] // local water stress multiplier

Coefficients are derived from published datacenter PUE/WUE disclosures and regional water stress data from the World Resources Institute.