DOALocal AI › Generate AI images locally with ComfyUI
Generate AI images locally with ComfyUI
ComfyUI is the reference tool to generate images (and video) with open models on your own computer. It works with nodes: each step of the process is a box you connect to the next.
Checked on 13 Oct 2026. Versions change every week: check the official site before installing.
| Operating systems | Windows, macOS and Linux (desktop app, portable build or manual install, per the README) (source) |
|---|---|
| Licence and price | GPL-3.0 · free |
| Ease of use | Technical |
| Current version | v0.39.0 (5 Oct 2026) |
| Official site | comfy.org · GitHub |
How to install it
For new users the README recommends the desktop app (comfy.org/download). On Windows there is also a portable package, which the README says is not meant for regular users. With comfy-cli: pip install comfy-cli and comfy install. Manual install: git clone https://github.com/Comfy-Org/ComfyUI, pip install -r requirements.txt and python main.py; then open the address the terminal shows.
Which models
Comfy-Org publishes ready-to-use files for models such as FLUX.2-dev and the components for FLUX Klein 4B. Qwen-Image is among the most downloaded models on Hugging Face.
What hardware you need
Image generation is much more demanding than chat: in practice you need a dedicated graphics card or a Mac with plenty of memory. We do not publish VRAM figures for FLUX.2 or Qwen-Image because we could not verify them in an official source; check each model's page before downloading.
Keep reading
- Local AI: how to run AI on your computer offline (2026)
- Tools to run AI locally: Ollama, LM Studio, Jan and more
- How to install local AI on Windows, step by step
- How to install local AI on a Mac, step by step
- How to install local AI on Linux, step by step
- Which local AI model to pick for your RAM: 8, 16, 24, 32 GB
- Local AI on a Mac with Apple Silicon: guide and models
- Local AI vs. cloud AI: privacy, cost and limits
- Local AI: frequently asked questions