# Stanford's Joules-to-GDP framework ranks AI models by economic value per joule

> Original title: New Stanford paper finds that a smaller AI model can return far more economic value per unit of energy than the most capable one.

- Source: Rohan Paul
- Published: 2026-11-04T21:29:20.000Z
- Bad Influence: https://badinfluence.news/items/dy9ol0trzkc9twvamgl62h001
- Original: https://x.com/rohanpaul_ai/status/2118093262082945044

## Summary

Stanford researchers introduce Joules-to-GDP, a framework that measures the dollar value of AI work minus running cost per joule of energy. Across 220 GDPval tasks in 44 occupations, Gemma 4 31B delivers 56% of GLM 5.2's net value with 23% of the energy. Moving Qwen3.6-27B from 1 to 8 H100s cuts task time 22% but uses 3.2x the energy.

## Post

New Stanford paper finds that a smaller AI model can return far more economic value per unit of energy than the most capable one.

Now that the world is limited by energy/power, how much energy each AI task burns should indeed by studied.

Here, the researchers built Joules-to-GDP, which measures the dollar value of an AI's work, minus its running cost, per joule of energy used. They tested 5 open models on 220 professional tasks from GDPval, covering 44 occupations.

> GLM 5.2 produced the most net value per task, but Gemma 4 31B got 56% of it with 23% of the energy. 

> Moving Qwen3.6-27B from 1 to 8 H100s cut task time by just 22% while using 3.2x the energy.

> GPT-OSS 120B used 68% less energy than GLM 5.2 despite nearly the same price and speed.

> Running the same model on 8 H100 chips instead of 1 made tasks only 22% faster but used 3.24x the energy, for nearly the same dollar value of work

https://x.com/jonsaadfalcon/status/2118079242340814896/video/1

![](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Fmedia%2FHWT5kNSaUAA5Q8F.jpg&mode=thumb&exp=1794096000&sig=15e9063e96d54f0d)

[Video](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Famplify_video_thumb%2F2118075316396822529%2Fimg%2Faw5x4OPTHVkZt8Vx.jpg&mode=thumb&exp=1794096000&sig=9bdecdfae24f9124)

## Quoted post @JonSaadFalcon

> AI is using more and more of the world’s energy. But how much economic value are we getting from AI?
> 
> We’re introducing Joules-to-GDP (J2G), a framework that measures how well AI turns energy into economic value and shows which tasks, models, and hardware give the best return.
> 
> Across 44 occupations in finance, engineering, law, administrative work, and more in @OpenAI's GDPval, we arrive at three findings:
> 
> 1) The tasks with the highest economic value are not the tasks with the highest economic value per joule. For example, CAD design earns ~10% more per task than building websites, but the latter earns ~27% more value per joule.
> 
> 2) The most powerful model isn’t always the best use of energy. Gemma 4 31B delivers 56% of GLM 5.2’s net value with 23% of the energy, for 2.5x the value per joule.
> 
> 3) More chips don’t guarantee proportionate returns. Moving the same model from 1 to 8 NVIDIA H100s uses 3.2x the energy but cuts completion time by only 22%, with about the same value.
> 
> With @SnorkelAI, we’re also releasing GDPWorks, a benchmark for economically valuable tasks across 36 additional occupations.
> 
> Joint work co-led by @amirzeinaalii and @Avanika15, with @Mersad_Abbasi, @Jon_Hartley_, John Hennessy, @HazyResearch, and @Azaliamirh.
> 
> More details below 🧵

https://x.com/JonSaadFalcon/status/2118079242340814896