Fooocus Open source

What it is

Fooocus is a free, offline image generator by lllyasviel (the ControlNet author), released under GPL-3.0 at lllyasviel/Fooocus. It wraps Stable Diffusion XL in a Midjourney-like interface: type a prompt, get images, with prompt expansion and sensible defaults doing the tuning. The README states the project is in limited long-term support with bug fixes only and will not move to newer architectures such as FLUX; the last code push was in December 2025. It became popular in 2023-2024 because it hid the complexity of Stable Diffusion behind a handful of settings while still giving good results, and its presets and styles are still used by many hobbyists.

Key features

• One-click Windows package (download and run.bat); Linux and macOS via Python
• Automatic prompt expansion and quality presets
• Image prompt, inpaint, outpaint and upscale/vary tools
• Style presets and LoRA support for SDXL
• Runs on as little as 4 GB NVIDIA VRAM and 8 GB RAM
• Fully offline, no account

Models inside

Stable Diffusion XL checkpoints and LoRAs only. The README says Fooocus does not support newer models like FLUX and points users to other tools for them.

Pricing and rights

Fooocus is free and open source (GPL-3.0). Images are generated locally with no watermark or usage limits; commercial use depends on the SDXL checkpoint's licence (the base SDXL model uses the CreativeML Open RAIL++-M licence). There are no paid plans.

Who it is for

Beginners who want local SDXL images with minimal setup on a modest GPU, and users who already rely on existing SDXL LoRAs and checkpoints. It is less suitable for anyone who needs current models, accurate text in images, video or an API for automation.

Verdict

Fooocus remains the easiest way to get good SDXL results offline, but it is frozen in bug-fix mode on an older model family. For current models choose ComfyUI or InvokeAI.