In short
Look for evidence rather than labels: published energy or emissions figures per request with a stated method, a choice of smaller models, information on where it runs, and independent ratings such as the AI Energy Score. Few tools publish all of this yet, so treat "green" or "sustainable" without numbers as a claim, not a fact.
Steps
- Look for energy or emissions figures per request, and check what they include.
- Check whether the tool offers smaller or lighter model options.
- Find out where it runs and whether the provider publishes data-centre information.
- Look up the model on independent ratings such as the AI Energy Score.
- Ask the provider directly when the information isn't published.
A green AI tool can show its numbers. There is no official energy label for AI tools yet, so the best signal is how much a provider publishes and how clearly it explains its method.
What good evidence looks like
- Per-request figures with a boundary. Google, for example, reports a median Gemini text prompt at 0.24 Wh, and explains what it counted. Mistral AI published a life-cycle assessment putting a 400-token response at 1.14 g CO₂e and 45 mL of water. The figures differ partly because they count different things.
- Model choice. Tools that offer smaller or faster models for simple tasks let you use less.
- Hosting information. Where it runs, how efficient the data centres are, and how the electricity is sourced.
- Independent ratings. Hugging Face's AI Energy Score rates many open models from one to five stars for energy efficiency on common tasks.
Warning signs
- "Green", "eco" or "carbon neutral" with no figures or method behind them.
- Claims based only on offsets or annual renewable certificates.
- A single impressive number with no boundary or date.
In the EU, Directive (EU) 2024/825 bans generic environmental claims to consumers that cannot be proved.
Common mistakes
- Treating a provider's own figure as independent. Most are self-reported and unverified.
- Comparing figures with different boundaries as if they were like for like.
For the questions to ask, see ask an AI company for its energy figures, and read energy labels for AI for where ratings are heading.
Sources
- AI Energy Score, Hugging Face (2025)
- Measuring the environmental impact of delivering AI at Google scale, Elsworth et al., Google (2025)
- Our contribution to a global environmental standard for AI, Mistral AI (2025)
- Directive (EU) 2024/825 on empowering consumers for the green transition, European Union (2024)
Related
How do I spot greenwashing in AI marketing?
Warning signs of greenwashing in AI products and services, from vague 'sustainable AI' labels to offset-based neutrality claims, and the questions to ask.
How do I ask an AI company for its energy figures?
What to ask an AI provider about energy, carbon and water per request, how to judge the answer, and why asking helps even when the reply is incomplete.
How do I read an AI Energy Score rating?
What the AI Energy Score's one-to-five-star ratings mean, what they measure, how to use them to compare models and what they don't tell you.
Energy labels and benchmarks for AI
Where AI energy ratings stand in 2026: AI Energy Score, MLPerf Power, EU AI Act documentation and data-centre rating, and what labels can show.