Image Generation

Imagen 3 Environmental Impact

StandardEstimated

Google's latest image generation model

Architecture
Cascaded Diffusion Model
Provider
Google
2.5 Wh
Energy per query
0.33 g
CO₂ per query
3 mL
Water per query
8x more than
vs Google search

Energy per query

2.5 Wh

8x more than a Google search (0.3 Wh)

CO2 per query

0.33 g

Google Global Network grid (300 gCO₂/kWh)

Water per query

3 mL

~370 queries to fill 1 litre

Processing location

Google global network (64% renewable)

Provider

Google

Category

Image Generation

Grid carbon intensity

300 g CO2/kWh (64% renewable)

How does Imagen 3 compare?

Ranked #89 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

Imagen 3 consumes approximately 2.5 Wh per image. Running on Google's efficient TPU infrastructure, it benefits from the same hardware optimizations as Gemini, resulting in lower energy consumption than comparable GPU-based image models.

Power Source & Carbon

Runs on Google's global TPU infrastructure with 100% renewable energy matching and a PUE of 1.10.

Water Usage

At approximately 2.7 mL per image, Imagen 3 benefits from Google's efficient cooling systems with a site WUE of around 0.26 L/kWh.

About Imagen 3

Imagen 3 is the most energy-efficient image generation model in our dataset at 2.5 Wh per query — ranked #1 of 16 models in its category. Google's model, released in August 1, 2024, google's latest image generation model. Built on Cascaded Diffusion Model.

Imagen 3 benefits from running in Google Global Network, one of the cleaner grid regions in our dataset at 300 gCO₂/kWh with 64% renewable energy. The same model running in a coal-heavy region would produce significantly more carbon per query.

These figures are estimates derived from hardware specifications and API benchmarks — Google has not published official energy data for Imagen 3. Actual consumption may vary significantly depending on batching, quantisation, and infrastructure optimisations that we cannot observe from outside.

Imagen 3 in Context

22.8 kWh
per year

Your yearly Imagen 3 footprint

At 25 queries per day, your annual Imagen 3 usage consumes 22.8 kWh — roughly what a fridge uses in a month. That produces 3.0 kg of CO₂.

8x
vs a text query

Image vs text: the energy gap

Generating one image with Imagen 3 uses 8x 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.

What does your Imagen 3 usage cost the planet?

Use our calculator to estimate your personal environmental footprint based on how often you use Imagen 3.

Calculate My Compute

Frequently Asked Questions

How much energy does Imagen 3 use per query?

Each Imagen 3 query consumes approximately 2.5 Wh of energy. This is 8x more than a traditional Google search (~0.3 Wh).

What is Imagen 3's carbon footprint?

Based on the carbon intensity of Google global network (64% renewable), each query produces approximately 0.33 g of CO2. The grid in this region has a carbon intensity of 300 g CO2/kWh with 64% renewable energy.

How much water does Imagen 3 use?

Each query consumes approximately 3 mL of water, primarily used for cooling the data centers that process the request.

How does Imagen 3 compare to a Google search?

A Imagen 3 query uses 8x more than a Google search in terms of energy. A Google search uses approximately 0.3 Wh, while Imagen 3 uses 2.5 Wh.

Technical Details

Architecture

Cascaded Diffusion Model

Release date

2024-08-01

Open source

No