Docker images for ComfyUI

English | 🀄 中文说明

This repo is for Docker images that runs ComfyUI - an AIGC GUI powering node-based workflow.

Quick Start - NVIDIA GPU

mkdir -p \
  storage-cache/dot-cache \
  storage-cache/dot-config \
  storage-nodes/dot-local \
  storage-nodes/custom_nodes \
  storage-models/models \
  storage-models/hf-hub \
  storage-models/torch-hub \
  storage-user/input \
  storage-user/output \
  storage-user/user-profile \
  storage-user/user-scripts

# Add sudo if needed
docker run -it --rm \
  --name comfyui-cu130 \
  --pull=always \
  --runtime=nvidia \
  --gpus all \
  -p 8188:8188 \
  -v "$(pwd)"/storage-cache/dot-cache:/root/.cache \
  -v "$(pwd)"/storage-cache/dot-config:/root/.config \
  -v "$(pwd)"/storage-nodes/dot-local:/root/.local \
  -v "$(pwd)"/storage-nodes/custom_nodes:/root/ComfyUI/custom_nodes \
  -v "$(pwd)"/storage-models/models:/root/ComfyUI/models \
  -v "$(pwd)"/storage-models/hf-hub:/root/.cache/huggingface/hub \
  -v "$(pwd)"/storage-models/torch-hub:/root/.cache/torch/hub \
  -v "$(pwd)"/storage-user/input:/root/ComfyUI/input \
  -v "$(pwd)"/storage-user/output:/root/ComfyUI/output \
  -v "$(pwd)"/storage-user/user-profile:/root/ComfyUI/user \
  -v "$(pwd)"/storage-user/user-scripts:/root/user-scripts \
  -e CLI_ARGS="" \
  yanwk/comfyui-boot:cu130-slim-v2

CUDA Compatibility

The supported CUDA versions for each GPU architecture are shown in the table below:

GPU Architecture	Blackwell	Hopper	Ada Lovelace	Ampere	Turing	Volta	Pascal	Maxwell
Example GPU	RTX 5090	H100	RTX 4090	RTX 3090	RTX 2080 GTX 1660	TITAN V	GTX 1080	GTX 980
cu130 ⭐	✔️	✔️	✔️	✔️	✔️	❌	❌	❌
cu128	✔️	✔️	✔️	✔️	✔️	❌	❌	❌
cu126	❌	❌	✔️	✔️	✔️	✔️	✔️	✔️

• CUDA 13.0 images are currently recommended.

  • ComfyUI’s performance library is currently developed based on CUDA 13.0, which is also considered one of the stable versions of PyTorch.

• If you are unsure about your NVIDIA GPU architecture, see this article.

Note	These CUDA compatibility limitations are due to the PyTorch toolchain, not the NVIDIA CUDA Toolkit. For more information, refer to the PyTorch build script.

CUDA Image Tags

Slim

The slim images start with only ComfyUI and ComfyUI-Manager, yet include many dependencies to make future Custom Node installation easier. Recommended for beginners.

• cu130-slim-v2 ⭐

  • CUDA 13.0, Python 3.13 (with GIL), No xFormers

• cu128-slim

  • CUDA 12.8, Python 3.12

• cu126-slim

  • CUDA 12.6, Python 3.12

MEGAPAK

The megapak images are all-in-one bundles, including development kits and dozens of Custom Nodes for ComfyUI.

• cu130-megapak-pt211 ⭐

  • CUDA 13.0, Python 3.13 (with GIL), GCC 14, PyTorch 2.11.0

• cu128-megapak

  • CUDA 12.8, Python 3.12, GCC 14, PyTorch 2.11.0

• cu128-megapak-pt29

  • CUDA 12.8, Python 3.12, GCC 11, PyTorch 2.9.1

• cu128-megapak-pt28

  • CUDA 12.8, Python 3.12, GCC 11, PyTorch 2.8.0

• cu126-megapak

  • CUDA 12.6, Python 3.12, GCC 13, PyTorch 2.9.1

Testing

• nightly

  • Using development version of PyTorch. For testing latest features.

ROCm Image Tags

• rocm

  • For AMD GPUs with ROCm 7.

  • Based on PyTorch’s builds (faster release).

• rocm7

  • For AMD GPUs with ROCm 7.

  • Based on AMD’s builds (comprehensive functionality).

• rocm6

  • For AMD GPUs with ROCm 6 (stable version).

XPU Image Tags

• xpu

  • For Intel GPUs with XPU.

More Image Tags

• cpu

  • A smaller image for CPU only.

• archived

  • Archived Dockerfiles of retired image tags.

License

Mulan Public License，Version 2

This open source license is written and valid both in Chinese and English, how good is that!