Stable Diffusion 3.5 Environmental Impact
Latest open-source image model from Stability AI
- Architecture
- Multimodal Diffusion Transformer (MMDiT)
- Parameters
- 8B
- Provider
- Stability AI
Energy per query
3.2 Wh
11x more than a Google search (0.3 Wh)
CO2 per query
1.3 g
Global Average grid (475 gCO₂/kWh)
Water per query
6 mL
~167 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 3.5 compare?
Detailed Breakdown
Energy Consumption
Stable Diffusion 3.5 uses a Multimodal Diffusion Transformer (MMDiT) architecture at 8B parameters. At ~3.2 Wh per image, it's slightly more energy-intensive than SDXL due to the larger model but produces significantly higher quality output. The open-source nature means energy depends on deployment hardware.
Power Source & Carbon
Fully open-source. Can be run on consumer GPUs (RTX 4090 etc.) or cloud infrastructure. Self-hosted deployments can choose low-carbon regions.
Water Usage
At ~6.0 mL per image. When self-hosted on consumer hardware, water consumption drops to zero as personal hardware uses air cooling.
About Stable Diffusion 3.5
Stable Diffusion 3.5 is an open-source image generation model from Stability AI, released in October 22, 2024, that runs below the category average for energy consumption at 3.2 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 8B parameters, it latest open-source image model from stability ai.
These figures are estimates derived from hardware specifications and API benchmarks — Stability AI has not published official energy data for Stable Diffusion 3.5. Actual consumption may vary significantly depending on batching, quantisation, and infrastructure optimisations that we cannot observe from outside.
Stable Diffusion 3.5 in Context
Your yearly Stable Diffusion 3.5 footprint
At 25 queries per day, your annual Stable Diffusion 3.5 usage consumes 29.2 kWh — roughly what a fridge uses in a month. That produces 11.9 kg of CO₂.
Image vs text: the energy gap
Generating one image with Stable Diffusion 3.5 uses 11x 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
Stability AI Stable Family
How energy efficiency has evolved across versions.
What does your Stable Diffusion 3.5 usage cost the planet?
Use our calculator to estimate your personal environmental footprint based on how often you use Stable Diffusion 3.5.
Calculate My ComputeFrequently Asked Questions
How much energy does Stable Diffusion 3.5 use per query?
Each Stable Diffusion 3.5 query consumes approximately 3.2 Wh of energy. This is 11x more than a traditional Google search (~0.3 Wh).
What is Stable Diffusion 3.5's carbon footprint?
Based on the carbon intensity of Self-hosted (varies), each query produces approximately 1.3 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 3.5 use?
Each query consumes approximately 6 mL of water, primarily used for cooling the data centers that process the request.
How does Stable Diffusion 3.5 compare to a Google search?
A Stable Diffusion 3.5 query uses 11x more than a Google search in terms of energy. A Google search uses approximately 0.3 Wh, while Stable Diffusion 3.5 uses 3.2 Wh.
Technical Details
Architecture
Multimodal Diffusion Transformer (MMDiT)
Parameters
8B
Release date
2024-10-22
Open source
Yes