Skip to main content

ComfyUI vs AUTOMATIC1111: v0.39 against v1.10.1

ComfyUI and AUTOMATIC1111's Stable Diffusion web UI are both free, open-source programs that run image models on your own computer. They no longer move at the same speed. ComfyUI v0.39.2 is dated 8 October 2026, and its maker, Comfy Org, now also ships a desktop installer and a paid cloud. AUTOMATIC1111 is still on v1.10.1, built from code committed on 27 July 2024. This page compares them as they stand today: releases, models, install, hardware, licence and hosting.

Reviewed against 32 sources. Prices, models, policies and terms change, so check the linked pages before you buy, build, publish or trade.

The short answer

For a new install in October 2026, pick ComfyUI. It is still being released, it runs current open models such as Flux.2, Qwen Image and Wan 2.2, and it comes with an installer and an official cloud. AUTOMATIC1111 still does its original job: Stable Diffusion 1.5, SDXL and SD3 images from a simple form. Choose it only if that is all you need and a last release from 2024 does not worry you.

Choose ComfyUI if

  • You want models released since mid 2024. The ComfyUI README names 20 image families, including Flux.1, Flux.2, Qwen Image and Z-Image; the newest family AUTOMATIC1111 added is Stable Diffusion 3, in v1.10.0.
  • You want software that is still shipping. ComfyUI v0.39.2 is dated 8 October 2026; the last commit on the AUTOMATIC1111 master branch is from 27 July 2024.
  • You want video, audio or 3D in the same tool, or an installer that brings its own Python. Comfy Desktop runs on Windows 10+, macOS 13+ and Linux.
  • You have no suitable GPU. Comfy Cloud runs ComfyUI on 96 GB cards, with 5 free runs and plans from $20 a month.

Choose AUTOMATIC1111 if

  • You only make still images with Stable Diffusion 1.5, SDXL or SD3 and prefer one page of fields to a node graph.
  • You already have a working install. The clone failure reported on 20 December 2025 only stops fresh installs.
  • You rely on an extension or script written for it. Its official index lists 353 extensions, and the program they target has not changed since v1.10.1.

The facts, side by side

Each row shows the date it was checked and numbered links to the pages it was read from.

marks the side the facts favour on that row. Rows without one are a draw or a matter of fit.

Swipe the table sideways to see AUTOMATIC1111.

ComfyUI and AUTOMATIC1111 compared, fact by fact
ComparedComfyUIComfy OrgAUTOMATIC1111Stable Diffusion web UI
Latest release421819Advantage: v0.39.2, dated 8 October 2026 in the changelog. The newest GitHub release with portable downloads is v0.39.0, published 5 October 2026.v1.10.1. GitHub dates the release 9 February 2025; the code it points to was committed on 27 July 2024.
Last code change31920Advantage: 11 October 2026 on the master branch, the newest commit when this page was checked27 July 2024 on master. The dev branch last changed on 2 March 2026.
Who maintains it1121731Comfy Org, based in San Francisco, which also sells Comfy Cloud. The code is at github.com/Comfy-Org/ComfyUI.The developer AUTOMATIC1111, with collaborators merging fixes on GitHub. The README names no company and nothing for sale.
Interface117A node graph: every step is a box you wire to the next. App Mode puts a simple form over a finished workflow.A Gradio web page of tabs and fields: txt2img, img2img, Extras, PNG info, Checkpoint Merger, Training and Settings.
Image model families117212324Advantage: Stable Diffusion 1.5, SDXL, SD3.5, Flux.1, Flux.2, Qwen Image, Z-Image, Hunyuan Image 2.1, HiDream and Chroma, among 20 the README namesStable Diffusion 1.x and 2.x, SDXL (since v1.5.0), Stable Diffusion 3 (since v1.10.0), SD 2.1 Turbo, Segmind SSD-1B and Alt-Diffusion. FLUX appears nowhere in the README, changelog or wiki.
Video, audio and 3D117Advantage: Video (Wan 2.1 and 2.2, LTX-Video 2 and 2.3, HunyuanVideo 1.5), audio (ACE-Step 1.5, Stable Audio 3), 3D (Hunyuan3D 2.1) and text modelsNone listed. The README covers still images: generation, inpainting, outpainting and upscaling.
Easiest install917Advantage: Comfy Desktop: one installer for Windows 10+, macOS 13+ on Apple Silicon, or Linux (.deb and AppImage). It sets up Python and the dependencies itself.Install Python 3.10.6 and git, clone the repository, then run webui-user.bat on Windows or webui.sh on Linux and macOS.
Other install routes182531Windows portable builds for NVIDIA, AMD and Intel GPUs, comfy-cli, or a manual git clone on any system. Python 3.13 is recommended.A Windows zip, sd.webui.zip. A fresh git install from master has failed since December 2025 unless you set a replacement repository or switch to the dev branch; the zip includes that fix.
GPUs it runs on8717252627Advantage: NVIDIA (GTX 900 series onward; 30 series or newer recommended), AMD on Linux and Windows, Intel Arc, Apple Silicon, Ascend, Cambricon and Iluvatar chips, or CPU only with --cpuNVIDIA recommended; RTX 50 series cards need the dev branch. AMD on Linux through ROCm, AMD on Windows only through a DirectML fork. Apple Silicon runs most features. Intel through an external OpenVINO guide.
Minimum memory stated191723284 GB of VRAM with 8 GB of RAM, per the README. Comfy Desktop asks for 8 GB of RAM (16 GB recommended) and 4.85 GB of disk per installation.4 GB of VRAM for 512 x 512 images, with reports of 2 GB working, and 16 GB of RAM to run smoothly
Licence5922GPL-3.0. Comfy Desktop is dual-licensed: AGPL-3.0-or-later, or a commercial licence from Comfy Org.AGPL-3.0
Hosted version141230Advantage: Comfy Cloud, run by Comfy Org on RTX 6000 Pro cards with 96 GB of VRAM. 5 free runs, then plans from $20 a month.None from the project. Its wiki links third-party hosts: Paperspace, Runpod, Kaggle, Google Colab, Azure ML, SageMaker Studio Lab and vast.ai.

Where each one is stronger

ComfyUI Comfy Org

  • Still released every week or two

    GitHub lists 30 ComfyUI releases between 21 March and 5 October 2026, and the README describes a weekly cycle with a new stable version roughly every 2 weeks. AUTOMATIC1111 has tagged nothing since v1.10.1. 2118

  • Runs current open models, and more than images

    The README names 20 image families, including Flux.2, Qwen Image and Z-Image, plus video, audio, 3D and text models. Version 0.38.0 on 29 September 2026 added five open-model integrations in one release. The AUTOMATIC1111 list ends at Stable Diffusion 3. 1421

  • An installer and an official cloud

    Comfy Desktop installs ComfyUI with its own Python, updates itself and can run several versions side by side. When a job is too big for your card, Comfy Cloud runs the same workflow file on a 96 GB GPU, limited to the custom nodes it has installed. AUTOMATIC1111 has neither. 911121430

AUTOMATIC1111 Stable Diffusion web UI

  • One page instead of a graph

    Prompt, negative prompt, sampler, size and seed are fields on one tab, with inpainting, outpainting, hires fix and X/Y/Z plots built in. You get a first image without wiring anything. 17

  • Stable Diffusion tools in the box

    It ships a Training tab for embeddings and hypernetworks, a Checkpoint Merger for up to 3 checkpoints, face restoration with GFPGAN and CodeFormer, and prompt editing part way through a generation. 17

  • A fixed target for scripts and extensions

    Nothing on master has changed since 27 July 2024, so an extension or API script that worked on v1.10.1 still meets the same program. The official index lists 353 extensions. 1932

Which to pick, by job

  • Setting up local image generation for the first time

    Pick ComfyUI

    Comfy Desktop is one download that brings its own Python. A fresh AUTOMATIC1111 install needs Python 3.10.6 and git, and since December 2025 a git install from master stops at a missing repository until you apply the fix. 91731

  • Running FLUX, Qwen Image or another model released since mid 2024

    Pick ComfyUI

    ComfyUI lists Flux.1, Flux.2, Qwen Image and Z-Image as natively supported. The AUTOMATIC1111 README, changelog and wiki never mention FLUX, and the last family it added was Stable Diffusion 3 in v1.10.0. 1172124

  • Making short AI video clips

    Pick ComfyUI

    ComfyUI lists Wan 2.1 and 2.2, LTX-Video and HunyuanVideo 1.5 for video and ACE-Step 1.5 for audio. AUTOMATIC1111 lists no video or audio model. On Comfy Cloud, a 5 second Wan 2.2 clip at default settings uses roughly 11 credits. 11714

  • Quick SDXL or SD 1.5 images on a machine that already runs AUTOMATIC1111

    Pick AUTOMATIC1111

    An existing install is not touched by the December 2025 clone failure, the form needs no wiring, and SDXL has been supported since v1.5.0 with flags for 4 GB, 8 GB and 12 GB cards. Moving gains you nothing until you want a newer model. 312329

  • No GPU, or a laptop too small for the model

    Pick ComfyUI

    Comfy Org runs Comfy Cloud: 5 free runs, then Standard at $20 a month with 4,200 credits, on cards with 96 GB of VRAM. AUTOMATIC1111 has no hosted version; its wiki only links third-party hosts. 1430

What it costs

Both programs are free, so this table covers the one thing either side charges for, Comfy Org's hosted services, and what a local install needs instead. Prices are in US dollars from Comfy's pricing page. Comfy Cloud credits are spent only while a workflow is running, and plan credits do not roll over.

Swipe the table sideways to see AUTOMATIC1111.

ComfyUI and AUTOMATIC1111 prices
ComparedComfyUIComfy OrgAUTOMATIC1111Stable Diffusion web UI
The software1152217Free under GPL-3.0. Comfy says there are no feature gates, trials or paid tier for core functions.Free under AGPL-3.0. The README lists nothing for sale.
Free hosted trial1430Comfy Cloud: 5 free runs on real GPUs, no credit cardNo hosted version from the project
Paid hosted plans1430Standard $20, Creator $35 or Pro $100 a month. Billed yearly: $16, $28 or $80 a month ($192, $336 or $960 up front).None. The wiki links third-party hosts such as Paperspace, Runpod and vast.ai, which set their own prices.
What a plan includes144,200, 7,400 or 21,100 credits a month. Runs stop at 30 minutes, or 1 hour on Pro. Creator and Pro can import models from Civitai and Hugging Face.No plan. A local install has no credits, run limit or charge per image.
Paid extras inside the local app1516117Optional. Partner Nodes (closed models such as Nano Banana) and Comfy Cloud nodes spend prepaid credits. Launching with --offline disables the paid nodes.None listed in the README or wiki
Renting GPU time by the hour1430Comfy API: an RTX PRO 6000 with 96 GB of VRAM at $4.54 an hour, billed by the second, plus $0.20 per GB a month for model storageNo official service. You rent a machine from a third-party host and install the web UI on it.
Hardware for a local install982829Comfy Desktop: 8 GB of RAM minimum, 16 GB recommended, 4.85 GB of disk per installation, and an NVIDIA or AMD GPU recommended16 GB of RAM to run smoothly. For SDXL the wiki pairs 12 GB or more of VRAM with --xformers alone, 8 GB with --medvram-sdxl and 4 GB with --lowvram, and asks for at least 24 GB of RAM on Windows 11.

What changed recently

  1. ComfyUI v0.39.2 added partner nodes, including the Clarity AI Crystal Upscaler and Vidu Q4 Preview. 4

  2. ComfyUI v0.39.0 added the --offline and --disable-partner-nodes launch flags. 4

  3. Comfy Desktop v1.1.6 was published, the latest stable desktop build. 10

  4. ComfyUI v0.38.0 added open-model support including Ming Image 0.1 Design and ControlNet for Qwen-Image 2.1. 4

  5. ComfyUI v0.34.5 added beta Comfy Cloud nodes, which run Flux 2 and Z-Image Turbo workflows on Comfy's GPUs from a local install. 4

  6. Last commit to the AUTOMATIC1111 dev branch, an install fix. 20

  7. An AUTOMATIC1111 collaborator reported that fresh installs fail to clone the Stability-AI/stablediffusion repository, and put the fix on the dev branch. 31

  8. GitHub's date for the AUTOMATIC1111 v1.10.1 release. 18

  9. AUTOMATIC1111 v1.10.0 was released with Stable Diffusion 3 support. The v1.10.1 code, one bug fix, was committed the same day. 21

Questions people ask

Is AUTOMATIC1111 still being updated?

Not on its main branch. The last commit to master is from 27 July 2024, and the latest release is still v1.10.1, which GitHub dates 9 February 2025. The dev branch took install fixes until 2 March 2026, including the PyTorch version that RTX 50 series cards need, but no new version has been tagged from it. 19182025

Does AUTOMATIC1111 support FLUX?

Its own pages do not list it. The README, changelog and wiki name Stable Diffusion 1.x and 2.x, SDXL, Stable Diffusion 3, SD 2.1 Turbo, Segmind SSD-1B and Alt-Diffusion, and FLUX appears in none of them. ComfyUI lists Flux.1, Flux.2, Flux Kontext and Flux.2 Klein as natively supported. 172123241

Is ComfyUI harder to learn than AUTOMATIC1111?

The first hour is harder, because you connect nodes instead of filling in a form. Two things shorten it: ComfyUI ships workflow templates that load a finished graph, and App Mode can put a simple form over one. AUTOMATIC1111 keeps prompt, sampler and size on one tab, which is quicker to start with. 11317

How much VRAM do ComfyUI and AUTOMATIC1111 need?

Both name 4 GB as the floor. The ComfyUI README says it can run even the largest open models on 4 GB of VRAM with 8 GB of RAM by streaming weights, and its GPU guide recommends NVIDIA 30 series or newer. AUTOMATIC1111 supports 4 GB cards for 512 x 512 images; for SDXL its wiki suggests --lowvram at 4 GB, --medvram-sdxl at 8 GB and --xformers alone at 12 GB or more. 172329

Is ComfyUI free, and what does Comfy Cloud cost?

ComfyUI is free and open source under GPL-3.0, and Comfy Org says it will always be free to run locally. Comfy Cloud is the optional paid service: 5 free runs, then Standard at $20 a month with 4,200 credits, Creator at $35 with 7,400 or Pro at $100 with 21,100. Billed yearly, the same plans work out at $16, $28 and $80 a month. 111214

Can I use my AUTOMATIC1111 models in ComfyUI?

Yes, and you do not have to copy them. ComfyUI ships a file called extra_model_paths.yaml.example with a ready section for AUTOMATIC1111. Rename it to extra_model_paths.yaml, uncomment the a1111 section and set base_path to your stable-diffusion-webui folder. ComfyUI then reads the checkpoints, VAEs, LoRAs, embeddings, upscalers and ControlNet models in place. 16

Sources

Each page was read on the date beside it. The links open the current version, which may have changed since.

  1. 1Comfy Org: ComfyUI README on GitHub
  2. 2Comfy Org: ComfyUI releases on GitHub
  3. 3Comfy Org: ComfyUI commits on master
  4. 4Comfy Org: ComfyUI changelog
  5. 5Comfy Org: ComfyUI licence file
  6. 6Comfy Org: extra_model_paths.yaml.example
  7. 7Comfy Org: which GPU to buy for ComfyUI
  8. 8Comfy Org: ComfyUI system requirements
  9. 9Comfy Org: Comfy Desktop overview
  10. 10Comfy Org: Comfy Desktop releases on GitHub
  11. 11Comfy Org: Comfy Desktop download page
  12. 12Comfy Org: Comfy Cloud
  13. 13Comfy Org: Comfy Cloud documentation
  14. 14Comfy Org: Comfy Cloud pricing
  15. 15Comfy Org: Partner Nodes overview
  16. 16Comfy Org: Comfy Cloud nodes
  17. 17AUTOMATIC1111: Stable Diffusion web UI README
  18. 18AUTOMATIC1111: releases on GitHub
  19. 19AUTOMATIC1111: commits on master
  20. 20AUTOMATIC1111: commits on the dev branch
  21. 21AUTOMATIC1111: changelog
  22. 22AUTOMATIC1111: licence file
  23. 23AUTOMATIC1111 wiki: Features
  24. 24AUTOMATIC1111 wiki: all 33 pages, searched for FLUX
  25. 25AUTOMATIC1111 wiki: Install and run on NVidia GPUs
  26. 26AUTOMATIC1111 wiki: Install and run on AMD GPUs
  27. 27AUTOMATIC1111 wiki: Installation on Apple Silicon
  28. 28AUTOMATIC1111 wiki: Troubleshooting
  29. 29AUTOMATIC1111 wiki: Optimum SDXL usage
  30. 30AUTOMATIC1111 wiki: Online services
  31. 31AUTOMATIC1111: discussion 17212 on the failed Stable Diffusion clone
  32. 32AUTOMATIC1111: extension index the app loads (353 entries counted)

Take the free Starter Kit.

Five working files from the IImagined programs: 32 Instagram hooks, an AI character template, a pricing worksheet and two calculators. Free, and they open in your browser.

Get the free Starter Kit

Related program

AI Influencers. Learn reference testing, model selection, quality control, disclosure, rights checks, production, and measurement. Included in All Access.

IImagined publishes this comparison and sells the All Access membership.