How much GPU memory does stable-diffusion-3.5-large-turbo need?

About 47.9 GB to draw images: 43.94 GB of weights and the runtime's own memory (an estimate: its settings are behind its licence).

What is the cheapest GPU to run stable-diffusion-3.5-large-turbo?

Right now one Nvidia A40 at $0.62 an hour; RTX 4090 at $0.63 an hour is the best value for its speed. Prices are live offers from GPU clouds, checked every 15 minutes.

Can stable-diffusion-3.5-large-turbo run on an RTX 4090?

Not on one: it needs about 47.9 GB and an RTX 4090 has 24 GB. Use a bigger GPU, or a smaller (quantized) version of the model.

Models / stable-diffusion-3.5-large-turbo

stabilityai/stable-diffusion-3.5-large-turbo

Text to image · served with Diffusers · ↓ 5.8K downloads a month · View on Hugging Face

stabilityai/stable-diffusion-3.5-large-turbo needs about 47.9 GB of GPU memory to draw images (an estimate: its settings are behind its licence). It runs on one Nvidia A40 from $0.62/h right now; best value: RTX 4090 at $0.63/h.

The model's publisher asks users to accept its licence (other). GridPup cannot download this one yet: choose an ungated version below.

Weights: 44 GB
GPU memory needed: about 47.9 GB (estimate)
Memory · disk: 60.7 GB · 89 GB
Architecture: StableDiffusion3Pipeline
Licence: other

Hardware and deploy

The publisher has not let GridPup download this licence-gated model yet. An ungated version of it (above) can be deployed.

Questions

How much GPU memory does stable-diffusion-3.5-large-turbo need?: About 47.9 GB to draw images: 43.94 GB of weights and the runtime's own memory (an estimate: its settings are behind its licence).
What is the cheapest GPU to run stable-diffusion-3.5-large-turbo?: Right now one Nvidia A40 at $0.62 an hour; RTX 4090 at $0.63 an hour is the best value for its speed. Prices are live offers from GPU clouds, checked every 15 minutes.
Can stable-diffusion-3.5-large-turbo run on an RTX 4090?: Not on one: it needs about 47.9 GB and an RTX 4090 has 24 GB. Use a bigger GPU, or a smaller (quantized) version of the model.

Loading…