---
title: "5 Fooocus Alternatives for Newer Models and Edits"
url: https://www.imagequarry.com/alternatives/fooocus/
last_updated: 2026-10-20
---

# Fooocus alternatives now that it gets bug fixes only

Fooocus is an offline, open-source image generator built entirely on Stable Diffusion XL, with a simple prompt-first interface and its own inpaint and image prompt tools. Its README now lists it as Limited Long-Term Support with bug fixes only, says there are no plans to add newer model architectures, and points Flux users to WebUI Forge or ComfyUI. Locally it needs at least a 4GB Nvidia GPU. Mage suits people who want newer models hosted with typed edits, ComfyUI people staying offline, and Leonardo.Ai people who want phone apps. ImageQuarry suits people who want each described change kept as a version.

_We make ImageQuarry, one of the alternatives below. Fooocus and each alternative were compared on model support, project activity, hardware, how edits are made, how you start and price, using each vendor's public pages, checked October 20, 2026._

## Fooocus alternatives at a glance

| Tool | Best for | Price from | How it differs |
|---|---|---|---|
| Mage | Fooocus users who want SDXL plus newer models hosted, with edits written as instructions | Free plan with a 300 Gem bonus; Basic is $10 a month | SDXL and newer families in the browser, with Mango instruction edits |
| ComfyUI | Fooocus users staying offline who want Flux and other newer models | No charge locally; Comfy Cloud Standard is $20 a month on monthly billing | The engine the Fooocus README itself suggests for Flux |
| Leonardo.Ai | People who want many current models, a daily allowance and phone apps instead of a local install | Free plan of 150 fast tokens a day; Essential $12 a month on monthly billing | Hosted models from several labs, 150 tokens a day, iOS and Android |
| ImageQuarry | People who want to stop masking and simply say what to change, keeping each attempt | Starter at $9 monthly, or Pro at $29 monthly | Edits asked for in words, kept as versions you can branch from |
| Invoke | People who want another local open-source app that runs Flux and is still adding features | Open source (Apache 2.0) with no paid tier | Local Flux, SDXL and SD 1.5 with a layer canvas, actively released |

## What is Fooocus?

Fooocus calls itself "offline, open source, and free" and is "built entirely on the Stable Diffusion XL architecture". It ships three default SDXL models for general, realistic and anime images, accepts SDXL models from Civitai, and expands short prompts with an offline GPT-2 engine (the "Fooocus V2" style). Input Image tools cover inpaint and outpaint, Vary (Subtle) and Vary (Strong), upscaling, Image Prompt, FaceSwap and Describe. The README says the GitHub repo is the only official source.

_Fooocus, the GitHub project by lllyasviel, is not the look-alike fooocus.com, fooocus.ai or fooocus.org sites, which its README calls fake._

## What Fooocus does well

- **Short prompts go a long way**: An offline GPT-2 based prompt engine expands what you type, and three default SDXL models (juggernautXL for General, realisticStockPhoto for Realistic, animaPencilXL for Anime) cover common looks out of the box.
- **Its own inpaint and image prompt methods**: Fooocus uses its own inpaint algorithm and models, plus its own image prompt algorithm with ControlNet-style options, and the README says its inpaint results are more satisfying than software using the standard SDXL inpaint model.
- **Offline, with nothing to pay**: The software runs offline and the maintainer sells no hosted version, describing Fooocus as 100% non-commercial open-source software under GPL v3.
- **Quick variations and batches**: Vary (Subtle) and Vary (Strong) rework an image in one step, and Advanced settings set the number of images, styles, aspect ratios and a negative prompt.

## Where Fooocus falls short

- **Bug fixes only**: The README lists Limited Long-Term Support with bug fixes only; the latest release is v2.5.5 from August 12, 2024, and the repo's last push was on December 1, 2025.
- **SDXL and nothing newer**: There are no plans to migrate to or add newer model architectures, and for Flux the README sends you to WebUI Forge or ComfyUI/SwarmUI.
- **Built around Nvidia cards**: The minimum is a 4GB Nvidia GPU and 8GB of RAM; AMD needs 8GB and is in beta, Apple M1/M2 is about 9x slower than an RTX 3XXX, and CPU-only use needs 32GB of RAM.
- **No editing by instruction**: Edits run through inpaint, outpaint, upscale and variation tools; no instruction or conversational editing is described.
- **Thin support outside Windows**: Mac is "not intensively tested", the official Colab notebook was last tested on August 12, 2024, and sharing the UI with --listen or --share is unauthenticated by default.

## Why people look for a Fooocus alternative

- You want Flux or other newer models, and Fooocus says it has no plans to add newer architectures.
- You work on a Mac or an AMD card, where Fooocus calls AMD support beta and puts Apple M1/M2 at about 9x slower than Nvidia.
- You want to type an edit instruction, and Fooocus edits only through inpaint, outpaint and variation tools.
- You want active releases, and the newest Fooocus release, v2.5.5, dates from August 12, 2024.
- You ran Fooocus on Colab, and its official notebook was last tested on August 12, 2024.

## The 5 Fooocus alternatives

### 1. Mage

Best for: Fooocus users who want SDXL plus newer models hosted, with edits written as instructions
How it differs: Mage hosts SDXL alongside Flux 2, Chroma, Z-Image and Nano Banana, so the SDXL look you know and newer families sit in one menu with no GPU to manage. Mango 3 takes exact instructions for an edit while keeping the rest in place, and up to 10 reference images help hold one character across a set. Creations stay visible to your account only unless you post them.
Price: Free plan with a 300 Gem bonus; Basic is $10 a month (as of 2026-10)
Limitation: Mango 3 tops out at 2K output (higher only on Pro Plus and up), Gems expire, and the API is a limited beta.

### 2. ComfyUI

Best for: Fooocus users staying offline who want Flux and other newer models
How it differs: The Fooocus README points Flux users to ComfyUI (or SwarmUI or WebUI Forge). ComfyUI runs on the same 4GB VRAM floor, supports AMD, Intel and Apple Silicon GPUs, and lists Flux.2, Qwen Image, HiDream and editing models such as Flux Kontext. The trade is a node graph in place of Fooocus's single prompt box, though Comfy Agent can build a graph from a description.
Price: No charge locally; Comfy Cloud Standard is $20 a month on monthly billing (as of 2026-10)
Limitation: Each job is a node graph you build or load, and Comfy Cloud's no-cost start is 5 runs for new users.

### 3. Leonardo.Ai

Best for: People who want many current models, a daily allowance and phone apps instead of a local install
How it differs: Leonardo.Ai hosts FLUX.2 Pro, FLUX.1 Kontext, Ideogram 4.5, Nano Banana Pro and Seedream 5.0 Pro next to its own Lucid and Phoenix models, so the newer families Fooocus will not add are a menu choice. Paid plans take up to 6 guidance reference images, you can train a model from 10 to 20 images, and it runs on the web, iOS and Android.
Price: Free plan of 150 fast tokens a day; Essential $12 a month on monthly billing (as of 2026-10)
Limitation: Free plan creations are public, and its editing pages do not describe typing an instruction to change an image.

### 4. ImageQuarry

Best for: People who want to stop masking and simply say what to change, keeping each attempt
How it differs: Fooocus changes an image through inpaint and outpaint areas or the two Vary options. ImageQuarry works from requests instead: pick a version, write what should change and what should stay, and review the new version it adds. Uploaded sources and references stay in the project alongside every version.
Price: Starter at $9 monthly, or Pro at $29 monthly (as of 2026-10)
Limitation: Every plan is paid, so it cannot stand in for an offline tool with no running cost, and ImageQuarry publishes no Terms or output specifications.

### 5. Invoke

Best for: People who want another local open-source app that runs Flux and is still adding features
How it differs: Invoke is open source and offline like Fooocus, but it supports Flux as well as SDXL and SD 1.5, and v7.0.0 alpha builds appeared on October 4, 11 and 19, 2026. Its layer canvas adds control layers, regional references and inpaint masks, while boards and Canvas Projects keep work in order.
Price: Open source (Apache 2.0) with no paid tier (as of 2026-10)
Limitation: Its guidelines ask for 8GB of VRAM for SDXL and 10GB for FLUX.1, more than the 4GB Fooocus needs, and the hosted Invoke platform has shut down.

## Who should stay with Fooocus

- You want a no-cost, fully offline SDXL generator with one prompt box, and your Nvidia card runs it well.
- You like the GPT-2 prompt expansion and the three default SDXL models for general, realistic and anime looks.
- You use SDXL models from Civitai and have no need for newer architectures.
- You lean on Vary (Subtle), Image Prompt and FaceSwap for quick variations of one look.

## How to try an alternative alongside Fooocus

1. Find your outputs folder (the path_outputs key in the config) and copy the images you still want to work on.
2. Leave the image log on while you compare tools, since the --disable-image-log flag turns that record off.
3. Note which default model sat behind your favourites (General, Realistic or Anime) so you can look for a similar style in the new tool.
4. Rewrite one inpaint job as a sentence, for example "replace the sky with dusk light", and try it in a tool that takes typed edits, such as Mage or ImageQuarry.
5. Keep Fooocus installed for SDXL work during the trial, since it runs offline at no cost.

Does not carry over: Fooocus styles, its GPT-2 prompt expansion and its own inpaint models belong to Fooocus, so expect to adjust prompts that leaned on them.

## Fooocus alternatives: questions

### Is Fooocus still being updated?

Only with bug fixes. The README describes Limited Long-Term Support with bug fixes only and no plans to move to newer model architectures. The latest release is v2.5.5 from August 12, 2024, and the repo's last push was on December 1, 2025.

### What should I use for Flux instead of Fooocus?

Fooocus's own README recommends WebUI Forge or ComfyUI/SwarmUI for Flux. Invoke also runs Flux locally, and Mage (Flux 2) and Leonardo.Ai (FLUX.2 Pro) host Flux models in the browser.

### Is there a free online alternative to Fooocus?

Fooocus's official Colab notebook was last tested on August 12, 2024, and its README calls sites such as fooocus.com and fooocus.ai fake. In the browser, Mage offers a free plan with a 300 Gem sign-up bonus and Leonardo.Ai a free plan of 150 fast tokens a day. Every ImageQuarry plan is paid.

### Can I run Fooocus on a Mac or an AMD GPU?

Mac is not intensively tested, and the README puts Apple M1/M2 at about 9x slower than an Nvidia RTX 3XXX; AMD needs 8GB and is in beta. ComfyUI supports AMD, Intel and Apple Silicon GPUs, and a hosted tool such as Mage needs no local GPU at all.

### Where does Fooocus save my images, so I can take them to another tool?

Fooocus writes images to an outputs folder that you can change with the path_outputs key in its config, and keeps an image log unless you start it with --disable-image-log. Copy that folder before you uninstall or switch, since that folder is where Fooocus stores them.

### How does ImageQuarry compare with Fooocus?

Fooocus is offline SDXL with masks, variations and image prompts, at no cost on your own GPU. ImageQuarry is a paid hosted workspace where you describe the image, ask for each change in words, add references, and choose between four models, with Nano Banana 2.1 as the default.

## Sources

Checked 2026-10-20.

- [Fooocus: github.com/lllyasviel/Fooocus](https://github.com/lllyasviel/Fooocus)
- [Fooocus: github.com/lllyasviel/Fooocus/releases](https://github.com/lllyasviel/Fooocus/releases)
- [ComfyUI: comfy.org](https://comfy.org/)
- [ComfyUI: comfy.org/pricing](https://comfy.org/pricing/)
- [ComfyUI: github.com/Comfy-Org/ComfyUI](https://github.com/Comfy-Org/ComfyUI)
- [Invoke: github.com/invoke-ai/InvokeAI/releases](https://github.com/invoke-ai/InvokeAI/releases)
- [Invoke: invoke.ai](https://invoke.ai/)
- [Invoke: invoke.ai/features/canvas/layers-and-drops](https://invoke.ai/features/canvas/layers-and-drops/)
- [Invoke: invoke.ai/features/gallery](https://invoke.ai/features/gallery/)
- [Invoke: invoke.ai/start-here/system-requirements](https://invoke.ai/start-here/system-requirements/)
- [Leonardo.Ai: docs.leonardo.ai](https://docs.leonardo.ai/)
- [Leonardo.Ai: web.archive.org/web/20261003232705/https://www.leonardo.ai/pricing](https://web.archive.org/web/20261003232705/https://www.leonardo.ai/pricing?plan=team)
- [Leonardo.Ai: web.archive.org/web/20261011200646/https://leonardo.ai](https://web.archive.org/web/20261011200646/https://leonardo.ai/)
- [Mage: docs.mage.space](https://docs.mage.space/)
- [Mage: docs.mage.space/models/mango](https://docs.mage.space/models/mango)
- [Mage: docs.mage.space/privacy/private-by-default](https://docs.mage.space/privacy/private-by-default)
- [Mage: mage.space](https://www.mage.space/)
- [Mage: mage.space/membership](https://www.mage.space/membership)

## Related pages

- [ComfyUI alternatives](https://www.imagequarry.com/alternatives/comfyui/): Alternatives to ComfyUI for hosted generation and editing in words, without building a node graph.
- [Mage.space alternatives](https://www.imagequarry.com/alternatives/mage-space/): Alternatives to Mage (mage.space) for credits that last, a general API and phone apps.
- [Leonardo.Ai alternatives](https://www.imagequarry.com/alternatives/leonardo-ai/): Alternatives to Leonardo.Ai for typed edits, clearer free tier terms and one bill for API use.
- [Invoke alternatives](https://www.imagequarry.com/alternatives/invoke/): Alternatives to Invoke, compared.
- [AUTOMATIC1111 alternatives](https://www.imagequarry.com/alternatives/automatic1111/): Alternatives to the AUTOMATIC1111 Stable Diffusion web UI that are still maintained or need no install.
- [All AI image tool alternatives](https://www.imagequarry.com/alternatives/): Every alternatives page, grouped by kind of tool.

## Say what to change instead of masking it

Pick a version, write the change and what to keep, then compare the new version with the ones before it.