Fooocus

Free plan APILinuxMacSelf-hostedWindows

What Fooocus is

Fooocus is free, offline, open-source image-generation software based on Gradio and Stable Diffusion XL. It creates images from text, accepts reference images, supports image editing, and offers API access. Its interface includes styles, guidance, quality, image count, negative prompts, aspect ratios, and sampling controls. An offline GPT-2-based prompt processor is designed to reduce detailed prompt engineering and parameter tuning. Fooocus supports SDXL models from Civitai and describes InsightFace use for FaceSwap. Output formats include PNG, JPG, JPEG, WebP, AVIF, and GIF. The project provides Windows downloads launched through run.bat, Linux installation instructions, and an unofficial Apple silicon guide. It can also run in Google Colab, where basic text-to-image is available on the free tier. Local hardware requirements vary; some Nvidia configurations list 4 GB GPU memory and 8 GB system memory. Support is limited to bug fixes, with no current plan to move to newer model architectures.

Who it suits

Fooocus suits users wanting local or self-hosted image generation with text, reference-image, and editing workflows. It is also relevant to people using Windows, Linux, macOS, or Google Colab.

What it does well

Free, offline, and open source
Supports text, reference images, and editing
Offers API access
Exports six listed image formats
Includes styles and sampling controls

What to know first

Local hardware requirements vary
Support is limited to bug fixes
No current migration to newer architectures
Network sharing is unauthenticated by default

Limits

The free plan is local, offline software and requires suitable hardware; some Nvidia configurations list minimums of 4 GB GPU memory and 8 GB system memory. Google Colab's free tier provides basic text-to-image, while project support is limited to bug fixes.

Plans

Plan	Billed and limits	Price
Fooocus	Free, offline, open-source software · local hardware requirements apply	Free

As github.com lists them, read 12 Oct 2026. The maker's own page has the final word.

Getting started

Open the Fooocus GitHub project
On Windows, download and extract the project, then launch it with run.bat
Follow the provided Linux installation instructions, or use the unofficial Apple silicon Mac guide
Run Fooocus locally, through an API, or in a self-hosted setup
Alternatively launch it in Google Colab for basic text-to-image on Colab's free tier
Configure basic authentication before exposing a local listener or Gradio share endpoint

Verdict

Fooocus provides a broad local image-generation workflow with simple controls and multiple output formats. Its hardware requirements, limited support scope, and default unauthenticated sharing endpoints deserve attention before deployment.

Questions

Is Fooocus free?

Yes. Fooocus is listed as free and is described as free, offline, and open source.

Which platforms does Fooocus support?

It supports API, Linux, macOS, Windows, and self-hosted setups. The project also describes Google Colab use.

Can Fooocus edit images and use references?

Yes. It supports image editing and reference images, alongside text-to-image generation.

What models and integrations does it use?

Fooocus is built on Stable Diffusion XL, supports SDXL models from Civitai, uses Gradio as its base, and describes InsightFace for FaceSwap.

What hardware is required?

Requirements vary by platform and GPU. Some Nvidia configurations list 4 GB GPU memory and 8 GB system memory as minimums.

Are shared endpoints protected by authentication?

The local listener and Gradio share endpoint are unauthenticated by default. Basic authentication can be configured.

Facts from Fooocus's own pages

API access: Yes · github.com, 19 Sept 2026
Free plan: Yes · github.com, 19 Sept 2026
Hardware limits: The project lists minimum local requirements that vary by platform and GPU, including 4 GB GPU memory and 8 GB system memory for some Nvidia configurations. · github.com, 12 Oct 2026
Hosted use: The project describes running Fooocus in Google Colab, with basic text-to-image available on Colab's free tier. · github.com, 12 Oct 2026
Image editing: Its listed functions include image variation and upscaling, inpainting, outpainting, and image prompts. · github.com, 12 Oct 2026
Integrations: The README identifies Gradio as its base and describes using InsightFace for FaceSwap and Civitai SDXL models. · github.com, 12 Oct 2026
Linux and macOS: The README provides Linux installation instructions and an unofficial guide for Apple silicon Macs. · github.com, 12 Oct 2026
Models: Fooocus is built on Stable Diffusion XL and supports SDXL models from Civitai. · github.com, 12 Oct 2026
Network access: The UI can be exposed with a local listener or a Gradio share endpoint, and both are unauthenticated by default unless basic authentication is configured. · github.com, 12 Oct 2026
Offline and open source: The project describes Fooocus as offline, open-source, and free software. · github.com, 12 Oct 2026
Project support: Fooocus is in limited long-term support focused on bug fixes, with no current plans to migrate to newer model architectures. · github.com, 12 Oct 2026
Prompting: Fooocus includes an offline GPT-2-based prompt processing engine and is designed to work without much prompt engineering or parameter tuning. · github.com, 12 Oct 2026
Purpose: Fooocus is image-generating software based on Gradio. · github.com, 12 Oct 2026
Reference images: Yes · github.com, 19 Sept 2026
Security: GitHub reports that the repository has no security policy file and no published security advisories. · github.com, 12 Oct 2026
Styles and controls: The interface includes styles, guidance, quality, image count, negative prompts, aspect ratios, and sampling controls. · github.com, 12 Oct 2026
Text-to-image: Yes · github.com, 19 Sept 2026
Windows installation: The project provides a Windows download whose extracted folder is launched with run.bat. · github.com, 12 Oct 2026