Image Generation

Stable Diffusion XL Environmental Impact

StandardPeer-reviewed

Open-source image model — higher energy per image

Architecture
Latent Diffusion Model (U-Net based)
Parameters
6.6B
Provider
Stability AI
2.9 Wh
Energy per query
1.2 g
CO₂ per query
5 mL
Water per query
10x more than
vs Google search

Energy per query

2.9 Wh

10x more than a Google search (0.3 Wh)

CO2 per query

1.2 g

Global Average grid (475 gCO₂/kWh)

Water per query

5 mL

~185 queries to fill 1 litre

Processing location

Self-hosted (varies)

Provider

Stability AI

Category

Image Generation

Grid carbon intensity

475 g CO2/kWh (27% renewable)

How does Stable Diffusion XL compare?

Ranked #93 of 152 models by energy per query

0 Wh2 Wh4 Wh6 Wh8 WhImagen 3Stable DiffusionXLDALL-E 3MidjourneyGoogle search (0.3 Wh)

Detailed Breakdown

Energy Consumption

Stable Diffusion XL consumes approximately 2.9 Wh per image at standard settings (30 steps, 1024x1024) based on the Luccioni et al. benchmark on A100 hardware. High-quality configurations (50+ steps, upscaling) can reach 11+ Wh. Being open-source, it's often run on consumer GPUs which may be less power-efficient than data center hardware.

Power Source & Carbon

As an open-source model, Stable Diffusion XL is deployed across diverse infrastructure — personal PCs, cloud providers (RunPod, Replicate, AWS, etc.), and on-premises servers. The carbon impact varies enormously. A user running it on a home PC in Sweden (30 g CO2/kWh) produces roughly 25x less carbon than someone running it on a cloud server in India (700 g CO2/kWh). Stability AI's own hosted service has faced financial difficulties and defaulted on payments to AWS and Google Cloud.

Water Usage

Water consumption is approximately 5.4 mL per image when run in a data center. When run locally on a personal computer, water consumption for cooling is effectively zero since PCs use air cooling. This is one of the key environmental advantages of self-hosted open-source models.

About Stable Diffusion XL

Stable Diffusion XL is an open-source image generation model from Stability AI, released in July 26, 2023, that runs below the category average for energy consumption at 2.9 Wh per query. Because its weights are publicly available, it can be self-hosted on any infrastructure — meaning its carbon footprint depends entirely on where and how you choose to run it. At 6.6B parameters, it open-source image model — higher energy per image.

Stable Diffusion XL in Context

26.5 kWh
per year

Your yearly Stable Diffusion XL footprint

At 25 queries per day, your annual Stable Diffusion XL usage consumes 26.5 kWh — roughly what a fridge uses in a month. That produces 10.9 kg of CO₂.

10x
vs a text query

Image vs text: the energy gap

Generating one image with Stable Diffusion XL uses 10x more energy than a text conversation with ChatGPT. The diffusion process that creates images requires many iterative passes through the model, each refining the image from noise — far more compute per output than generating text one token at a time.

Key Insights

Open-source weights — can be self-hosted on infrastructure you control

What does your Stable Diffusion XL usage cost the planet?

Use our calculator to estimate your personal environmental footprint based on how often you use Stable Diffusion XL.

Calculate My Compute

Frequently Asked Questions

How much energy does Stable Diffusion XL use per query?

Each Stable Diffusion XL query consumes approximately 2.9 Wh of energy. This is 10x more than a traditional Google search (~0.3 Wh).

What is Stable Diffusion XL's carbon footprint?

Based on the carbon intensity of Self-hosted (varies), each query produces approximately 1.2 g of CO2. The grid in this region has a carbon intensity of 475 g CO2/kWh with 27% renewable energy.

How much water does Stable Diffusion XL use?

Each query consumes approximately 5 mL of water, primarily used for cooling the data centers that process the request.

How does Stable Diffusion XL compare to a Google search?

A Stable Diffusion XL query uses 10x more than a Google search in terms of energy. A Google search uses approximately 0.3 Wh, while Stable Diffusion XL uses 2.9 Wh.

Technical Details

Architecture

Latent Diffusion Model (U-Net based)

Parameters

6.6B

Release date

2023-07-26

Open source

Yes