How to install Stable Diffusion 3.5 Large model on ComfyUI

Updated Categorized as Tutorial Tagged , , 7 Comments on How to install Stable Diffusion 3.5 Large model on ComfyUI

Our old friend Stability AI has released the Stable Diffusion 3.5 Large model and a faster Turbo variant. Attempting to correct the flop of Stable Diffusion Medium, the SD 3.5 Large is a marked improvement. It generates diverse styles with good prompt adherence.

In this post, I will walk you through how to install and use the SD 3.5 Large models in ComfyUI.

Software

We will use ComfyUI, an alternative to AUTOMATIC1111. You can use it on Windows, Mac, or Google Colab. If you prefer using a ComfyUI service, Think Diffusion offers our readers an extra 20% credit.

Read the ComfyUI beginner’s guide if you are new to ComfyUI. See the Quick Start Guide if you are new to AI images and videos.

Take the ComfyUI course to learn how to use ComfyUI step-by-step.

Stable Diffusion 3.5 is currently not supported by AUTOMATIC1111 and Forge.

SD 3.5 Large model

Generation time

The following SD 3.5 models are available on ComfyUI. The timing below is what I got from my RTX4090.

  1. SD 3.5 Large – Full model requiring 20 sampling steps. 20 secs per image.
  2. SD 3.5 Large Turbo – Distilled model requiring 4 sampling steps. 10 secs per image.

Image sizes

The image standard resolution is 1024×1024. Like SDXL and Flux, you can use other aspect ratios that roughly stick to 1M pixels.

  • 1:1 – 1024 x 1024
  • 5:4 – 1152 x 896
  • 3:2 – 1216 x 832
  • 16:9 – 1344 x 768

Memory requirement

You should be able to run it with 12 GB VRAM.

License

SD 3.5 Large comes with a more permissive Community License, allowing small busineses to use them commercially.

SD 3.5 Large Full model

Step 0: Update ComfyUI

ComfyUI started supporting SD3.5 natively in October 2024. If you haven’t updated your ComfyUI since then, do so now.

The easiest way to update ComfyUI is through the ComfyUI Manager. Click Manager > Update ComfyUI.

Restart ComfyUI and reload the page.

Step 1: Download SD 3.5 Large model

Visit the model page. Fill in the agreement form.

Download the SD 3.5 Large checkpoint model. Put the it in the folder ComfyUI > models > checkpoints.

Step 2: Download the text encoders

You may have these models already if you have used Stable Diffusion 3 medium and Flux.

Before downloading the models, visit the SD 3 medium model page and complete the agreement form.

Download the models below:

If you have less than 32 GB RAM (CPU RAM, not GPU VRAM), you can use t5xxl_fp8_e4m3fn.safetensors instead of t5xxl_fp16.safetensors.

Put them in Comfyui > models > clip folder.

Step 3: Load the ComfyUI workflow

Download the SD 3.5 Large workflow JSON file below.

Drop it to your ComfyUI.

Press Queue Prompt to generate an image.

Stable Diffusion 3.5 Turbo

If the full model is too slow for you, you can try the Stable Diffusion 3.5 Turbo model. It is a distilled model that can produce high-quality images with four steps.

Step 0: Update ComfyUI

ComfyUI started supporting SD3.5 natively in October 2024. If you haven’t updated your ComfyUI since then, do so now.

The easiest way to update ComfyUI is through the ComfyUI Manager. Click Manager > Update ComfyUI.

Restart ComfyUI and reload the page.

Step 1: Download SD 3.5 Large Turbo model

Visit the model page and fill in the agreement form.

Download the SD 3.5 Large Turbo checkpoint model.

Put the model file in the folder ComfyUI > models > checkpoints.

Step 2: Download the text encoders

You may have them already if you have used Stable Diffusion 3 medium and Flux.

Before downloading the models, visit the SD 3 medium model page and complete the agreement form.

Download the models below:

If you have less than 32 GB RAM (CPU RAM, not GPU VRAM), you can use t5xxl_fp8_e4m3fn.safetensors instead of t5xxl_fp16.safetensors.

Step 3: Load the ComfyUI workflow

Download the SD 3.5 Large workflow JSON file below.

Drop it to your ComfyUI.

Press Queue Prompt to generate an image.

Reference

Avatar

By Andrew

Andrew is an experienced engineer with a specialization in Machine Learning and Artificial Intelligence. He is passionate about programming, art, photography, and education. He has a Ph.D. in engineering.

7 comments

  1. the tutorial should mention that the clip models should be put in ComfyUI/models/clip

    I mistakenly put them in the checkpoint folder.

  2. Yes, VERY new! But plugging along. It did take several attempts to get ComfyUI Manager to show up, finally have everything working – thanks for your great guides!

Leave a comment

Your email address will not be published. Required fields are marked *