Free & open source · Fewer than 3 clicks to first image

Fooocus Stable Diffusion UI

A Midjourney-style interface over Stable Diffusion XL no parameter tweaking, just type a prompt and generate.

Get it from GitHub

Opens the official lllyasviel/Fooocus repository · Windows

Author

lllyasviel

Cost

Free

Based on

SDXL + Gradio

Status

Bug-fix LTS

Watch out for fake sites: The project's own README specifically warns that fake "Fooocus" websites show up in Google search results. The GitHub link above is the only official source — avoid any other site claiming to offer it.

How to install it

  1. Open the GitHub repo and find the Windows download

    The button above takes you to the real repo — scroll to the "Download" → "Windows" section for the direct download link.

  2. Uncompress the downloaded file

    Right-click → "Extract All" to a folder with enough free space.

  3. Run run.bat

    Double-click it inside the extracted folder. First launch auto-downloads the SDXL base and refiner checkpoints.

  4. Type a prompt and generate

    No sampler settings, no CFG scale to tune — Fooocus handles those choices automatically.

System requirements

GPUNVIDIA with 4 GB+ VRAM minimum
RAM8 GB minimum; 16 GB recommended
OSWindows (also runs on Linux and via Colab)
Disk~10 GB for default models; +1.28 GB if you use inpainting

Good to know before you start

  1. SDXL only

    Fooocus is built specifically around Stable Diffusion XL — it doesn't run older SD 1.5 checkpoints.

  2. Anime and Realistic presets exist too

    Newer versions ship run_anime.bat and run_realistic.bat alongside the default run.bat.

  3. The project is in maintenance mode

    The author has said Fooocus is now limited to bug fixes rather than new features, so it's stable but won't gain major new capabilities.

Good to know: This is an independent overview, not affiliated with lllyasviel. The link above goes to the real, verified GitHub repository — this page never hosts the installer itself.