TreadLightlyAI
AI Transparency Card
Models
Inference
Green4100% renewable energy (Free, Starter, Builder tiers)
Conventional2Carbon offsets (Advanced plan only)
Inference
Green Tiny and Green Small run the same model on the same host. What differs is how long one answer can be: Tiny is capped, Small is not.
Training
Training
| Creator | Energy |
|---|---|
| Google DeepMind | Unknown |
| OpenAI | Carbon Offsets |
| Anthropic | Carbon Offsets |
Compromises we make
Most energy numbers are estimates
Only one model reports its own energy use: Green Large. Until August 2026 the default model reported it too. It now runs on Infomaniak, which reports none, so most of what you see is a token-based estimate.
We made that trade knowingly. Infomaniak has the strongest environmental record of any host we use, and the model it serves passes our safety bar. Neither of those turns an estimate into a measurement, so we label every estimate and say which is which.
Training energy isn't green
Inference is only part of the overall footprint. Training is a significant cost too, and we are not aware of any large models trained on renewable power.
Anthropic infrastructure
Two Claude models are available on Advanced. We don't host them and can't certify them as renewable.
Use of Claude Code
Sonnet 4.5 and Opus 4.6 assist development. We use frontier models for software development.
Environmental
Water Use?
Not tracked by providers. The 2025 AI environmental cost landscape review flags water as the field's largest omitted metric. We would rather name the gap than invent a number.
Energy Methodology
- Measured
- Green Large only (GreenPT)
- Estimated
- Green Tiny, Green Small, Haiku, Sonnet
- Green tier rate
- 90 uWh/token, 108 with overhead
- Version
- Calibration
- 115.31 uWh/token measured (Green Large); 56.65 measured on a model retired from the lineup
- Overhead
- 1.2x infrastructure multiplier
Full energy methodology: the formula, the rates, and their limits
Ethics & Policy
Model Inclusion Standard
- We exclude
- Below C overall or any F on the FLI Safety Index.
Any provider that dissolved its safety team.
This applies to the model creator and the API host independently. A creator that fails is excluded even when someone else does the hosting. - We require
- At least one of: green-hosted, energy-efficient, safety-leading, or open-weight.
Commitments