SwarmUI Auto Installer & Ultimate Image and Video Model Downloader (Patreon)
Downloads
Content
SwarmUI auto installer with extensions plus a unified AI models downloader for SwarmUI, ComfyUI, Automatic1111 and Forge, with 105 ready SwarmUI presets.
Supports SD 1.5, SDXL, FLUX, FLUX Krea, FLUX SRPO, Wan 2.1, Wan 2.2, Qwen Image, Qwen Image Edit, LTX 2.3, Krea 2 and MiniMax H3 with full Int8 ConvRot quants. Runs on Windows, RunPod, Massed Compute and SimplePod; this ultimate compilation also covers master tutorials and the V195 LongCat Avatar 1.6 lip sync preset.
Automatic SwarmUI Installer with Extra Extensions + Unified AI Models Downloader for SwarmUI, ComfyUI, Automatic1111 and Forge Web UI, Forge Web UI Classic and Neo - Supports SD 1.5, SDXL, FLUX, FLUX Krea, FLUX SRPO, Wan 2.1, Wan 2.2, Qwen Image, Qwen Image Edit Plus 2509, LTX 2.3, Krea 2, MiniMax H3, Full Int8 ConvRot Quants + SwarmUI presets to generate images and videos and more
Patreon exclusive posts index to find our scripts easily, Patreon scripts updates history to see which updates arrived to which scripts and amazing Patreon special generative scripts list that you can use in any of your task.
Join discord to get help, chat, discuss and also tell me your discord username to get your special rank : SECourses Discord
Please also Star, Watch and Fork our Stable Diffusion & Generative AI GitHub repository and join our Reddit subreddit and follow me on LinkedIn (my real profile)
=======
Latest zip file : SwarmUI_Model_Downloader_v195.zip
ComfyUI Backend (mandatory for SwarmUI) : https://www.patreon.com/posts/105023709
Recommended ComfyUI backend args (important!) :
--disable-pinned-memory --use-sage-attention --enable-triton-backend --disable-comfy-compiler
Sage Attention may reduce quality or cause black output on some models rarely
Tutorials
SwarmUI Complete Master Tutorial with ComfyUI Backend (28 October 2026) : https://youtu.be/0HmIeeI1iBo
How to update SwarmUI mini tutorial (31 October 2026) : https://www.youtube.com/watch?v=sdD3--WbHZk
Install Tutorial on Cloud (RunPod + MassedCompute) : https://youtu.be/bBxgtVD3ek4
SimplePod Tutorial : https://youtu.be/yOj9PYq3XYM
SimplePod Register : https://simplepod.ai/ref?user=secourses
Qwen Image Models Training Tutorial : https://youtu.be/DPX3eBTuO_Y
ComfyUI Master installation and how to use tutorial (17 October 2026) : https://youtu.be/nxsCBTcCVp4
ComfyUI Back-end Installer (Mandatory) with Torch 2.13 CUDA 13 and xFormers, TorchAO, Sage Attention, Segment Anything, Segment Anything 2 (SAM2) and SAM3, MSLK (xFormers won't work properly without this), InsightFace, Flash Attention https://www.patreon.com/posts/105023709
Definitely use our ComfyUI solo backend installer otherwise some of the newest stuff may fail
Example existing ComfyUI backend : E:\Comfy_UI_V105\ComfyUI\main.py
Sage Attention is optional
If you use Sage Attention and get black output, enable Display Advanced Options of SwarmUI and go Advanced Sampling and make Preferred DType = Default (16 bit)
Highest quality attention is --use-pytorch-cross-attention
New recommended backend is --enable-triton-backend
Recommended to add --disable-pinned-memory until its bugs fixed to not have any OOM error
Our Sage Attention is newest compiled and should not produce low quality or black outputs
If your GPU starts using shared VRAM for any reason, add this command to make it avoid that --reserve-vram 3
This command will preserve 3 GB VRAM for other tasks
--cache-none is useful when you work with dual models or keep switching between models
--novram is useful when you are getting OOM, especially longer video generation like LTX 2, it reduces VRAM usage and can be used together with --cache-none
Some VRAM savings --disable-smart-memory, --lowvram
Use --disable-smart-memory if you are getting stuck or OOM
Image and Video slider app : https://www.patreon.com/posts/133935178
18 November 2026 Update V195
Tutorial video : https://youtu.be/b_3BnPXouGc
New in SwarmUI: LongCat-Video-Avatar 1.6 Really Fast. A new preset turns one photo and any audio into a lip-synced 720p talking or singing video of any length, with 3 of the official 8 distilled steps. One new premerged model now runs every LongCat preset, and the downloader shows every bundle in one tab.
Amazing SwarmUI Presets moves to V92 with 105 presets: 1 new and 2 updated.
All 105 presets in the Presets tab, the new one marked NEW and the updated ones UPDATED:
The three LongCat presets up close, each with its own thumbnail:
Windows_Preset_Delete_Import.bat brings all of them into your SwarmUI; the steps are in How to update at the end.
NEW PRESET — LONGCAT AVATAR 1.6, 720P, 3 STEPS REALLY FAST
"LongCat Avatar 16 - Audio + Image To Lip Sync Video - 720p - 3 Steps Really Fast - 261118"
Your photo and your voice become a 720p talking or singing video of any length, with 3 of the official 8 steps. Apply the preset, attach the speech or song to the prompt, set the person's photo as Init Image and click Generate. All models come in the LongCat-Video-Avatar 1.6 Core Bundle, first in the new ComfyUI & SwarmUI Bundles tab of the downloader:
The bundle's new default model is our premerged INT8 ConvRot: both distillation LoRAs are merged in, so no LoRA work happens while sampling, and one model runs every LongCat preset. The plain INT8 ConvRot and BF16 models stay available in the downloader.
One photo and a 15-second voice recording, made with the same 3-step recipe at 720p in our ComfyUI:
The same man, glasses, T-shirt and room from the first second to the last, while his face, mouth and hands move with the words.
The same moments of the official 8 steps and the new 3-step preset, made from the same photo, speech, seed and size:
In our test with this 15-second 720p clip, the 3-step preset made the video about 2.9 times faster than the official 8 steps, with the same person, glasses, room and lip sync.
New in the LongCat Avatar group: Distilled Step Count (8, 4 or 3), Shorten Last Segment, which makes the last segment only as long as your audio needs, and int8 attention. The 3-step preset sets them for you. Prompt List, Mouth Enhancement and your SwarmUI LoRAs work as before.
Where to find it: Presets tab → filter LongCat → apply the 3-step preset, attach your audio, set your photo as Init Image and click Generate. Every finished video waits in History:
Reuse Parameters on a finished video loads its settings again for your next one.
UPDATED PRESETS — LONGCAT AVATAR 1.6, 8 STEPS
"LongCat Avatar 16 - Audio + Image To Lip Sync Video - 480p - 8 Steps - 261117"
"LongCat Avatar 16 - Audio + Image To Lip Sync Video - 720p - 8 Steps - 261117"
The official 8-step recipe now runs on the same premerged model. Same look: on the same photo, audio and seed it makes the same video as before, and you keep only one LongCat model on disk.
NEW IN THE MODEL DOWNLOADER
One 📦 ComfyUI & SwarmUI Bundles tab with every bundle: the SwarmUI bundles first, the LongCat-Video-Avatar 1.6 Core Bundle on top.
LongCat Video Avatar 1.6 Premerged DMD OPSD INT8 ConvRot in the Core Bundle and under Video Generation Models → LongCat-Video-Avatar 1.6 Models.
HOW TO UPDATE
Download SwarmUI_Model_Downloader_v195.zip from this post and extract it over your downloader folder.
Close SwarmUI and run Windows_Install_or_Update_SwarmUI.bat. It updates SwarmUI and our extensions, with the new LongCat Avatar options.
Update the ComfyUI backend: close SwarmUI again (step 2 starts it) and run Windows_Install_Or_Update_ComfyUI.bat from our ComfyUI V154.
Restart SwarmUI, then run Windows_Preset_Delete_Import.bat to import Amazing_SwarmUI_Presets_v92.json. It backs up your current presets first, then replaces them all, your own presets too.
Run Windows_Start_Download_Models_App.bat and download the LongCat-Video-Avatar 1.6 Core Bundle.
ComfyUI Installer and Presets: https://www.patreon.com/posts/105023709
17 November 2026 Update V194
Tutorial video : https://youtu.be/b_3BnPXouGc
New in SwarmUI: LongCat-Video-Avatar 1.6. Turn one photo and any audio into a lip-synced talking or singing video of any length, on your own GPU. The video follows your whole audio: it is made in segments of about 4 seconds that continue each other, with your photo as the identity reference, so a one-minute talk stays the same person in the same room from the first to the last frame.
Amazing SwarmUI Presets moves to V91 with 104 presets: 2 new.
All 104 presets in the Presets tab, the two new ones marked NEW:
The two new presets up close: 480p for speed, 720p for sharper detail, each with its own thumbnail:
Windows_Preset_Delete_Import.bat brings all of them into your SwarmUI; the steps are in How to update at the end.
NEW PRESETS — LONGCAT-VIDEO-AVATAR 1.6, LIP SYNC VIDEO OF ANY LENGTH
"LongCat Avatar 16 - Audio + Image To Lip Sync Video - 480p - 8 Steps - 261117"
"LongCat Avatar 16 - Audio + Image To Lip Sync Video - 720p - 8 Steps - 261117"
Your photo and your voice become a talking or singing video, as long as your audio. Apply a preset, attach the speech or song to the prompt, set the person's photo as Init Image and click Generate. All models come in the new LongCat-Video-Avatar 1.6 Core Bundle, the first bundle in the downloader:
The default model is our INT8 ConvRot version: about 17 GB instead of almost 30 GB for BF16, as the downloader shows them, and its sampling runs 1.7x faster than BF16 in our RTX 5090 test (one 93-frame segment at 512 x 768, 8 steps). BF16 stays available in the downloader.
The same presenter photo and the same 10-second narration at 1280 x 1920, made with LongCat-Video-Avatar 1.6, MiniMax H3 lip sync and AvatarForever in our ComfyUI:
Same photo, same audio, same size: only the model changes. All three keep the man, his glasses and the library of the photo, and LongCat-Video-Avatar 1.6 also moves his hands as he talks.
Mouth Enhancement is one switch in the new LongCat Avatar group: our CodeFormer mouth restoration runs on every frame, for cleaner, more even teeth and smoother lip edges. The same frames zoomed 4x, off on the left and on on the right:
Vocal separation is on by default, as in the official pipeline: background music does not move the mouth, and the saved video keeps your original audio.
Real jobs at 720p, each from one photo and one voice line: a chef presenting the special of the day, a real-estate agent showing a listing, a family doctor with a health tip and a fitness coach:
A 58-second talk made in one run. Fourteen frames from the first second to the last: the same man in the same room the whole time, while his face and hands keep moving with the words:
The same one-minute talk made with AvatarForever too, both in our ComfyUI: the same photo, the same narration, the same seed and the same 512 x 768 size, with mouth enhancement on in both. The same seven moments of both videos:
Both keep the man, his glasses, his T-shirt and the room for the whole minute. LongCat-Video-Avatar 1.6 moves his head and hands more as he talks, and AvatarForever keeps a steadier framing. The full side-by-side video with sound is attached to this post.
Prompt List in the LongCat Avatar group makes one video per prompt from the same voice and picture, your SwarmUI LoRAs apply, and Condition Cache is an exact speed-up for long videos.
Where to find it: Presets tab → filter 261117 → apply a preset, attach your audio, set your photo as Init Image and click Generate. Every finished video waits in History:
Reuse Parameters on a finished video loads its settings again for your next one.
NEW IN THE MODEL DOWNLOADER
LongCat-Video-Avatar 1.6 Core Bundle, first in the bundle list: all 11 files in one click.
LongCat-Video-Avatar 1.6 Models, first under Video Generation Models: the INT8 ConvRot model and the BF16 model one by one.
HOW TO UPDATE
Download SwarmUI_Model_Downloader_v194.zip from this post and extract it over your downloader folder.
Close SwarmUI and run Windows_Install_or_Update_SwarmUI.bat. It updates SwarmUI and our extensions, and adds the new LongCat Avatar extension.
Update the ComfyUI backend: close SwarmUI again (step 2 starts it) and run Windows_Install_Or_Update_ComfyUI.bat from our ComfyUI V153.
Restart SwarmUI, then run Windows_Preset_Delete_Import.bat to import Amazing_SwarmUI_Presets_v91.json. It backs up your current presets first, then replaces them all, your own presets too.
Run Windows_Start_Download_Models_App.bat and download the LongCat-Video-Avatar 1.6 Core Bundle.
ComfyUI Installer and Presets: https://www.patreon.com/posts/105023709
14 November 2026 Update V193
New in SwarmUI: MiniMax H3 Image with KJNodes' empty-image method, for cleaner skin, hands and fine detail. A new preset shows the model an empty picture during the first steps, a method from kijai's KJNodes. On many prompts it gives more natural skin, cleaner hands and sharper small objects than the standard Text To Image preset, at the same 16 MP and with the same prompt enhancer. It is also a new switch, MiniMax H3 Image KJNodes Empty Image, so you can turn it on with your own settings.
The prompt enhancer is now repeatable, too: a fixed seed gives the same caption and the same image every time.
Amazing SwarmUI Presets moves to V90 with 102 presets: 1 new and 5 updated.
All 102 presets in the Presets tab, the new one marked NEW and the updated ones UPDATED:
The new and updated presets up close; the new one has its own great hornbill thumbnail:
Windows_Preset_Delete_Import.bat brings all of them into your SwarmUI; the steps are in How to update at the end.
NEW PRESET — TEXT TO IMAGE WITH KJNODES EMPTY IMAGE
"MiniMax H3 Image - Text To Image 16 MP - KJNodes Empty Image - 6 Steps - 261114"
Cleaner skin, hands and fine detail from the same short prompt. Apply it, write your idea and click Generate, exactly like the standard Text To Image preset. It uses the same models, all in the MiniMax-H3 Core Bundle:
Already using MiniMax H3 Image? There is nothing new to download. The MiniMax-H3 Low VRAM Bundle has the smaller W4A8 versions.
The same caption and seed, the standard preset on the left and the new one on the right, made with the same presets in our ComfyUI, with details at their real pixels below. A watchmaker at his bench: the watch case, the gears and the loupe come out clean:
A chef plating a dish for a restaurant ad, with clean hands and a detailed plate:
An olive farmer for a brand campaign: natural, weathered skin instead of a crackled texture:
The empty picture also changes the composition, like a new seed, so keep both presets and use the one you like for each prompt.
Where to find it: Presets tab → filter 261114 → apply the preset, write your idea and click Generate:
The switch works for text to image; when you attach images to the prompt, the preset edits them as usual.
UPDATED — MINIMAX H3 IMAGE TEXT TO IMAGE AND IMAGE EDIT
A fixed seed repeats your image: the prompt enhancer now writes the same caption for the same seed, so the image repeats too.
Both presets keep the new switch off, so they work as before.
UPDATED — KREA 2
All three Krea 2 presets now select the BF16 text encoder of the Krea 2 Core Bundle under its own file name, so they always get it instead of SwarmUI's smaller default. Already using Krea 2? Download the Krea 2 Core Bundle once more to add it.
NEW IN THE MODEL DOWNLOADER
IP-Adapter And FaceID Bundle (SD 1.5, SDXL, Kolors): every IP-Adapter, FaceID adapter, FaceID LoRA and image encoder in one click, for SwarmUI's IP-Adapter feature and for ComfyUI.
Wan 2.2 Core 4 Steps Bundle now includes RIFE 4.7 frame interpolation.
LTX 2.5 Upscaler bundle: the smaller INT8 text encoder that SwarmUI uses by default, plus the Gemma 4 E2B prompt enhancer.
Download queue: shows how many downloads have finished, failed or been cancelled.
SwarmUI installer: if SwarmUI cannot be installed, it says why and what to do next.
HOW TO UPDATE
Download SwarmUI_Model_Downloader_v193.zip from this post and extract it over your downloader folder.
Close SwarmUI and run Windows_Install_or_Update_SwarmUI.bat. It updates SwarmUI and our extensions.
Update the ComfyUI backend: close SwarmUI again (step 2 starts it) and run Windows_Install_Or_Update_ComfyUI.bat from our ComfyUI V152.
Restart SwarmUI, then run Windows_Preset_Delete_Import.bat to import Amazing_SwarmUI_Presets_v90.json. It backs up your current presets first, then replaces them all, your own presets too.
If you do not have it yet, run Windows_Start_Download_Models_App.bat and download the MiniMax-H3 Core Bundle, or the MiniMax-H3 Low VRAM Bundle for less memory.
ComfyUI Installer and Presets: https://www.patreon.com/posts/105023709
11 November 2026 Update V192
MiniMax H3 Image is now in SwarmUI: 16-megapixel images in 6 steps. Two new presets turn a few words into images up to 6272 pixels wide and edit your photos with up to 16 images at once. MiniMax's own prompt enhancer runs locally inside SwarmUI: a short idea in any language becomes the detailed caption the model was trained on, and the enhancer even picks the best shape for your image.
Amazing SwarmUI Presets moves to V89 with 101 presets: 2 new, every earlier preset unchanged.
Real runs of the new presets in SwarmUI:
Every example in this post comes from a real run of the shipped presets in SwarmUI with their default settings.
The new presets up close, each with its own animal thumbnail:
Windows_Preset_Delete_Import.bat brings all of them into your SwarmUI; the steps are in How to update at the end.
To run them, download the MiniMax-H3 Core Bundle. It includes MiniMax H3 Image, its VAE and the prompt enhancer model; the text encoder is the one MiniMax H3 video already uses, so H3 users download only the new files:
Every format is in the catalog: INT8 ConvRot for the Core Bundle, W4A8 for the Low VRAM Bundle and BF16 in original precision, plus the prompt enhancer model in INT8 ConvRot and W4A8:
Where to find it: Image Generation Models → MiniMax-H3 Image Models, or the two MiniMax-H3 bundles under SwarmUI Bundles.
NEW PRESET — TEXT TO IMAGE AT 16 MEGAPIXELS
"MiniMax H3 Image - Text To Image 16 MP - 6 Steps - 261111"
Type an idea, click Generate, get a 16-megapixel image. The prompt enhancer writes the full description and chooses the aspect ratio. For advertising, put the headline, the brand and the price in quotes, and every letter comes out as typed:
Product photos for online shops, and packaging mockups with your own brand text:
Portraits at full detail. A farmer in a wheat field at golden hour, 3648 x 4576:
And wildlife: a snowy owl in falling snow, 4096 x 4096:
Where to find it: Presets tab → filter 261111 → apply the preset, write your idea and click Generate:
The prompt the model received is saved in the image's metadata as h3_image_prompt_received. Set MiniMax H3 Image Prompt Enhancer to Off to use your own long caption.
1 MP VS 4 MP VS 16 MP
The same caption and seed at 1, 4 and 16 MP, made with the same model in our ComfyUI presets, with the face from each cropped to the same share of the frame:
16 MP shows fur, whiskers and frost that 1 MP can only blur, and it stays fast. Seconds per image on an RTX 5090, also with only 24, 16 or 12 GB of its memory in use:
A 16 MP image takes about 16 seconds with all memory and about 20 seconds with 12 GB; the rest streams from system RAM. Lower Megapixels whenever you want a quick draft.
NEW PRESET — IMAGE EDIT WITH UP TO 16 IMAGES
"MiniMax H3 Image - Image Edit Up To 16 Images - 6 Steps - 261111"
Edit your photo with one sentence, or combine up to 16 images. Attach your images beside the prompt and call them @image1, @image2 and so on in the instruction. The enhancer looks at your images and writes the full edit description, and the result keeps the shape of image 1 at 16 MP. Virtual staging for real estate, an empty bedroom and three product photos:
Renovation previews from one photo, with the building, the windows and the camera kept:
A plain food or product photo becomes a finished advertisement, with the same pizza and the same bottle:
Where to find it: Presets tab → filter 261111 → apply the preset, attach your images, write the change and click Generate:
Without images it works as a text to image preset, so both presets do both jobs.
OUR PROMPT SYSTEM, BUILT IN
The official prompt enhancer in three modes: with reasoning (best), fast without reasoning, or off for captions you wrote yourself.
The prompt the model received is saved in every image's metadata, so you can read and reuse it.
Auto size: the enhancer picks the aspect ratio at your Megapixels; set MiniMax H3 Image Size to Exact to keep your Width and Height.
@image1 to @image16 for the images you attach, {red|blue} choices, and prompt batches from Add prompt, .txt files or a whole folder.
Less memory: pick the W4A8 model; BF16 is there for original precision.
Our new MiniMax H3 Image LLM guide in the zip makes ChatGPT, Claude or Gemini write long captions for the enhancer's Off mode.
HOW TO UPDATE
Download SwarmUI_Model_Downloader_v191.zip from this post and extract it over your downloader folder.
Close SwarmUI and run Windows_Install_or_Update_SwarmUI.bat. It updates SwarmUI to its new MiniMax H3 Image support and installs our MiniMaxH3Image extension.
Update the ComfyUI backend. With our ComfyUI installation, close SwarmUI again (step 2 starts it), run Windows_Install_Or_Update_ComfyUI.bat from our ComfyUI V150, then Windows_Custom_Nodes_Bundles_Installer.bat and install bundle 100. With SwarmUI's own ComfyUI, step 2 already updates our nodes.
Restart SwarmUI, then run Windows_Preset_Delete_Import.bat to import Amazing_SwarmUI_Presets_v89.json. It backs up your current presets first, then replaces them all, your own presets too.
Run Windows_Start_Download_Models_App.bat and download the MiniMax-H3 Core Bundle, or the MiniMax-H3 Low VRAM Bundle for less memory.
ComfyUI Installer and Presets: https://www.patreon.com/posts/105023709
10 November 2026 Update V190
MiniMax H3 Image in the downloader: 16-megapixel images in 6 steps. The MiniMax-H3 Core and Low VRAM bundles now include MiniMax H3 Image, the new model for text to image and image editing with up to 16 images. One click downloads the model, its VAE and the prompt enhancer model. The text encoder is the one MiniMax H3 video already uses, so H3 users download only the new files.
Real results:
Every image here comes from a real run of the two new MiniMax H3 Image presets of our ComfyUI V150.
The MiniMax-H3 Core Bundle, now with MiniMax H3 Image:
Every format is in the catalog: INT8 ConvRot for the Core Bundle, W4A8 for the Low VRAM Bundle and BF16 in original precision, plus the prompt enhancer model in INT8 ConvRot and W4A8:
Where to find it: Image Generation Models → MiniMax-H3 Image Models, or the two MiniMax-H3 bundles under SwarmUI Bundles.
WHY 16 MEGAPIXELS
The same caption and seed at 1, 4 and 16 MP, with the face from each cropped to the same share of the frame:
16 MP shows fur, whiskers and frost that 1 MP can only blur, and it stays fast. Seconds per image on an RTX 5090, also with only 24, 16 or 12 GB of its memory in use:
A 16 MP image takes about 16 seconds with all memory and about 20 seconds with 12 GB; the rest streams from system RAM.
IMAGE EDITING WITH UP TO 16 IMAGES
One sentence and a photo: a moss-covered wreck comes back to life, while the forest, the branches and the camera stay:
The presets for text to image and image editing, with MiniMax's own prompt enhancer built in, come with our ComfyUI V150. The zip also has a new MiniMax H3 Image LLM guide that makes ChatGPT, Claude or Gemini write detailed captions for it.
HOW TO UPDATE
Download SwarmUI_Model_Downloader_v190.zip from this post and extract it over your downloader folder.
Run Windows_Start_Download_Models_App.bat and download the MiniMax-H3 Core Bundle, or the MiniMax-H3 Low VRAM Bundle for less memory.
Get the two MiniMax H3 Image presets from our ComfyUI V150 post below.
ComfyUI Installer and Presets: https://www.patreon.com/posts/105023709
4 November 2026 Update V189
Edit real videos with MiniMax H3 in SwarmUI. Five new presets: replace the background of any video from one image, relight a whole clip, put a new face on someone, replace a person or object while every other pixel stays as filmed, and make long Pika H3 Flex videos that follow a story, scene by scene.
Amazing SwarmUI Presets moves to V88 with 99 presets: 5 new since V86, every earlier preset unchanged.
Real footage and real runs of the shipped presets:
Every result in this post comes from a real run of the shipped presets in SwarmUI.
All 99 presets in the Presets tab, the new ones marked NEW:
The new presets up close, each with its own animal thumbnail:
Windows_Preset_Delete_Import.bat brings all of them into your SwarmUI; the steps are in How to update at the end.
To run them, download the MiniMax-H3 Core Bundle. It includes the FaceSwap LoRA; if you already have the bundle, only the missing files download:
All the new presets use this one bundle.
Our live token counter is above the prompt: it counts your text, references, audio and video as you work and shows them against the MiniMax H3 limit, in green, amber and red, so you know what a run needs before you click Generate. With long video windows it counts per window, because each window is one run.
NEW PRESET — BACKGROUND REPLACE: YOUR VIDEO, A NEW BACKGROUND, NOTHING ELSE
"MiniMax H3 Background Replace - Pika H3 Flex 6 Steps - 261104"
Fully automatic: attach your video and a background image, and click Generate. The person is found by itself and stays exactly as filmed: every pixel of the face, hair, clothes and hands, the lip movements, the timing and the sound. Everything around them becomes the place in your image and moves with your camera.
Where to find it: Presets tab → filter 261104 → apply the preset, attach both files beside the prompt and click Generate. There is nothing to write or set. Long clips work too; we tested 31 seconds (736 frames). The video is made in overlapping windows, so a long clip needs no more GPU memory, only more time. The result is about 1 megapixel. People of similar size all stay, and distant passers-by go.
NEW PRESET — VIDEO RELIGHT
"MiniMax H3 Video Relight - Pika H3 Flex 6 Steps - 261104"
Relight a whole video. Attach your clip and click Generate: the prompt already holds a complete neon noir lighting plan. The people, the motion, the camera and the set stay.
Where to find it: Presets tab → filter 261104. Change the colours or directions in the Lighting plan lines of the prompt. H3 follows the plan through the prompt, so it is not pixel-exact light control; try two or three seeds.
NEW PRESET — FACE SWAP
"MiniMax H3 Face Swap - Ref2VA 8 Steps - 261104"
Put a new face on the person in your video. Attach your clip and a clear close-up of the new face, leave the prompt empty and click Generate. Clothes, setting and camera stay as in your clip.
H3 makes the whole clip anew, so the timing of the motion can drift a little. For your clip's exact timing and sound, use the Masked Edit preset below with Genjutsu Target set to head.
NEW PRESET — GENJUTSU MASKED EDIT: CHANGE ONE THING, KEEP EVERYTHING ELSE
"MiniMax H3 Genjutsu - Masked Edit - Pika H3 Flex 6 Steps - 261104"
Replace a person or an object and keep the rest of your clip exactly as filmed. Attach your clip and the new character and click Generate: the person is found and the prompt is written for you. The street, the traffic, the signs and the camera move stay untouched.
Where to find it: Presets tab → filter 261104. To replace something else, name it in Genjutsu Target: a car, a dog, a man in a blue suit.
NEW PRESET — PIKA H3 FLEX LONG VIDEO THAT TELLS A STORY
"Pika H3 Flex - Long Text to Video 20 Seconds - 261104"
A long video that develops instead of repeating. Write your description, then a line with ---, then one scene, another ---, the next scene. Each part of the video follows its own scene.
Five scenes in one prompt, one continuous 20-second flight:
Where to find it: Presets tab → filter 261104. About 4 seconds per scene works well; set Text2Video Frames for other lengths (481 = 20 s, 719 = 30 s). Without --- every window reads the same prompt, so motion can repeat. Scenes work in text runs, not with reference media.
HOW TO UPDATE
Download SwarmUI_Model_Downloader_v189.zip from this post and extract it over your downloader folder.
Close SwarmUI and run Windows_Install_or_Update_SwarmUI.bat.
Update the ComfyUI backend. With our ComfyUI installation, close SwarmUI again (step 2 starts it) and run Windows_Install_Or_Update_ComfyUI.bat from our ComfyUI V149. With SwarmUI's own ComfyUI, open Server → Backends, edit the ComfyUI Self-Starting backend and set AutoUpdate to Always Update for one start.
Restart SwarmUI, then run Windows_Preset_Delete_Import.bat to import Amazing_SwarmUI_Presets_v88.json. It backs up your current presets first, then replaces them all, your own presets too.
Run Windows_Start_Download_Models_App.bat and download the MiniMax-H3 Core Bundle.
ComfyUI Installer and Presets: https://www.patreon.com/posts/105023709
30 October 2026 Update V187
Pika H3 Flex: cinematic video with synchronized sound in 6 steps. Three new presets bring Pika Labs' new Pika H3 Flex to SwarmUI: text to video, beginning and ending frame to video and references to video, each with dialogue, sound effects and music made together with the picture.
Amazing SwarmUI Presets moves to V86 with 94 presets: 3 new, every earlier preset unchanged.
Our three examples: a ship in a bottle that sails out onto a stormy ocean, a blacksmith whose golden eagle lands on his anvil, and an airship captain made from just two images:
Every result in this post is a real run of the shipped presets in SwarmUI on our RTX 5090.
All 94 presets of Amazing SwarmUI Presets V86 in SwarmUI's Presets tab, the new ones marked NEW:
The 3 new presets up close, each with its own animal thumbnail:
Windows_Preset_Delete_Import.bat brings all of them into your SwarmUI; the steps are in How to update at the end.
NEW PRESETS — PIKA H3 FLEX: VIDEO WITH SOUND IN 6 STEPS
"Pika H3 Flex - Text to Video 6 Steps - 261030"
"Pika H3 Flex - Beginning and Ending Frame to Video 6 Steps - 261030"
"Pika H3 Flex - References to Video 6 Steps - 261030"
Pika's tuned MiniMax H3, in 6 steps. A 5-second 1344 × 768 clip with stereo sound takes about a minute on an RTX 5090. Init Audio, face inpainting and your own MiniMax H3 LoRAs stay available, as in every MiniMax H3 preset.
To run them, download the MiniMax-H3 Core Bundle. It now includes the two Pika H3 Flex LoRAs; if you already have the bundle, only those 3.6 GB download:
Beginning and Ending Frame: give it two images and it makes the shot between them. Our ship in a bottle becomes the real ship on the open sea in one continuous camera move:
Where to find it: Presets tab → filter Pika → Direct Apply Pika H3 Flex - Beginning and Ending Frame to Video 6 Steps - 261030. Put the beginning frame in Init Image, tick Display Advanced Options and put the ending frame in Image To Video → Video End Image, then click Generate:
Start the prompt with Pika's timing line; the preset description has it ready to copy. Leave Video End Image empty for first frame to video.
Text to Video: one prompt gives the whole scene, with the blacksmith's line, the hammer, the hiss of steam and the eagle's wingbeats in the soundtrack:
References to Video: attach a character image and a place image beside the prompt; our captain walks his sky harbour with his beard, goggles, green scarf and compass kept:
The MiniMax H3 prompt guide now covers Pika H3 Flex, so ChatGPT, Claude or Gemini can write your prompts.
ALSO IN V187
The MiniMax-H3 Low VRAM Bundle includes the two Pika H3 Flex LoRAs too.
SwarmUI's scheduler list gains pika_h3_flex, Pika's official 6-step schedule, after the update below.
HOW TO UPDATE
Download SwarmUI_Model_Downloader_v187.zip from this post and extract it over your downloader folder.
Close SwarmUI and run Windows_Install_or_Update_SwarmUI.bat.
Update the ComfyUI backend. With our ComfyUI installation, close SwarmUI again (step 2 starts it) and run Windows_Install_Or_Update_ComfyUI.bat from our ComfyUI V147. With SwarmUI's own ComfyUI, open Server → Backends, edit the ComfyUI Self-Starting backend and set AutoUpdate to Always Update for one start.
Restart SwarmUI, then run Windows_Preset_Delete_Import.bat to import Amazing_SwarmUI_Presets_v86.json. It backs up your current presets first, then replaces them all, your own presets too.
Run Windows_Start_Download_Models_App.bat and download the MiniMax-H3 Core Bundle.
ComfyUI Installer and Presets: https://www.patreon.com/posts/105023709
28 October 2026 Update V186
Anima character transfer, MOSS-Image layout control, stronger MiniMax H3 Genjutsu and faster FastH3 decoding. Put one character into any scene with Anima, place and pose people with simple color and depth maps in MOSS-Image, swap characters in your clips more convincingly with the new Genjutsu task LoRAs, and decode FastH3 Omni videos faster with LightVAE.
Amazing SwarmUI Presets moves to V85 with 91 presets: 5 new, 2 updated, and 24 others now switch the new options off when you apply them.
With the adapter's example maps, both new MOSS-Image presets put the two men where the mannequins stand, in their poses; without maps, the same prompt and seed picks its own layout:
Every result in this post is a real run of the shipped presets in SwarmUI on our RTX 5090.
All 91 presets of Amazing SwarmUI Presets V85 in SwarmUI's Presets tab, the new ones marked NEW and the updated ones UPDATED:
The 5 new and 2 updated presets up close:
Windows_Preset_Delete_Import.bat brings all of them into your SwarmUI; the steps are in How to update at the end.
NEW PRESETS — ANIMA: TAKE YOUR CHARACTER INTO A NEW SCENE
"Anima - Advanced Text To Image - 261028"
"Anima - Reference Character And Style - 261028"
One image of your character, any new scene. The Anima reference adapter carries the face, hair and outfit, and part of the drawing style, from your image, while your prompt picks the pose, place and light. Every final image is a clean 2048 × 2048.
To run them, download the Anima Core Bundle (3.84 GB) from the model downloader:
Our example: one character made with text to image, then four new scenes from that single image with one click on Generate prompt batch:
Where to find it: Presets tab → filter Anima → Direct Apply Anima - Reference Character And Style - 261028. Upload your character in Anima Reference Image, add your scene prompts and click Generate prompt batch:
The new Anima_Enchance_Prompt_Feed_For_LLMs.txt lets ChatGPT, Claude or Gemini write your prompts or a whole scene series.
NEW PRESETS — MOSS IMAGE COLOR + DEPTH CONTROL: PLACE YOUR PEOPLE WITH A MAP
"MOSS Image - Color + Depth Control - 261027"
"MOSS Image - Turbo 8 Steps Color + Depth Control - 261027"
Draw the layout, MOSS-Image follows it. A color map marks where each person or object goes, a depth map shows what is near and far, and your prompt says who they are. Turbo makes a 1024 × 1024 image in about 4 seconds.
To run them, download the MOSS-Image Core Bundle. It now includes the color + depth adapter; if you already have the bundle, only that 0.29 GB file downloads:
Where to find it: Quick Tools → Reset Params to Default, then Presets tab → filter Color + Depth → Apply. Add your maps with the Upload buttons of MOSS Image Color Map and MOSS Image Depth Map, describe each person and press Generate:
Part 8 of the MOSS-Image prompt guide writes the prompt for your maps.
NEW PRESET — FASTH3 OMNI WITH THE LIGHTVAE VIDEO DECODER
"FastH3 Omni - LightVAE References to Video 8 Steps - 261027"
The same FastH3 Omni video, decoded faster. LightVAE by LightX2V decoded and saved our 5-second video in about 4 seconds instead of about 6.5, with frames that look the same.
To run it, download the FastH3 Omni Core Bundle. It now includes the LightVAE decoder and the three Genjutsu files:
Where to find it: Quick Tools → Reset Params to Default, then Presets tab → filter LightVAE → Apply. The new MiniMax H3 Video Decoder parameter in Core Parameters picks the decoder for any MiniMax H3 or FastH3 Omni model:
Set it to None for the original decoder; every other H3 preset does that for you.
UPDATED PRESETS — MINIMAX H3 GENJUTSU: CHARACTER SWAP AND MOTION TRANSFER LORAS
"MiniMax H3 Genjutsu - FastH3 Omni 8 Steps - 261024"
"MiniMax H3 Genjutsu - Ref2VA 20 Steps - 261024"
Swap the person in any clip for your character, now with task LoRAs trained for exactly this job. Character Swap v1.1 and Motion Transfer v1.0 by Alisson Anjos bring your character's look over much more closely, and clips with mono sound now work.
The FastH3 Omni 8 Steps preset uses the FastH3 Omni Core Bundle above. The Ref2VA 20 Steps preset uses the MiniMax-H3 Core Bundle, which now includes the three Genjutsu files:
Same warrior, same shadowboxing clip, same seed: the original method and the new task LoRA default:
The task LoRA brings his big scale shoulder guards and heavy build over. original Genjutsu is still one click away in Genjutsu Method.
Where to find it: Quick Tools → Reset Params to Default, then Presets tab → filter H3 Genjutsu → Apply. Attach the clip and your character with Add References and press Generate:
Low VRAM mode is on by default with the same output.
ALSO IN V186
The MiniMax-H3 Low VRAM Bundle and the Complete Image Generation and Editing Bundle include the new models too.
The Krea 2 and FLUX 3 Klein bundles no longer download their shared text encoder again on every run.
MiniMax H3 Long Lip Sync - References no longer adds a second pass after an image-to-video preset: about 5.5 minutes instead of 11.
HOW TO UPDATE
Download SwarmUI_Model_Downloader_v186.zip from this post and extract it over your downloader folder.
Close SwarmUI and run Windows_Install_or_Update_SwarmUI.bat.
Update the ComfyUI backend. With our ComfyUI installation, close SwarmUI again (step 2 starts it) and run Windows_Install_Or_Update_ComfyUI.bat from our ComfyUI V146. With SwarmUI's own ComfyUI, open Server → Backends, edit the ComfyUI Self-Starting backend and set AutoUpdate to Always Update for one start.
Restart SwarmUI, then run Windows_Preset_Delete_Import.bat to import Amazing_SwarmUI_Presets_v85.json. It backs up your current presets first, then replaces them all, your own presets too.
Run Windows_Start_Download_Models_App.bat and download the core bundles shown above.
ComfyUI Installer and Presets: https://www.patreon.com/posts/105023709
24 October 2026 Update V185
This is a MiniMax H3 Genjutsu update. V185 adds two presets that replace the person in any video with your own character: attach your clip and one image of the new character, press Generate, and the result keeps the clip's motion, timing, camera, length, aspect ratio and original sound. Our MiniMaxH3References extension gets the new MiniMax H3 Genjutsu group, and the model downloader adds SAM 3.1 and Qwen3-VL 8B to the MiniMax H3 and FastH3 Omni bundles.
Amazing SwarmUI Presets moves to V84 with 86 presets: the two new presets, and the other 84 presets unchanged.
Our test in SwarmUI: this armoured warrior image and a 9.8-second phone clip of a man shadowboxing in a gym. Both presets put the warrior into the clip, punch for punch and step for step, in the clip's portrait format and with its own sound:
Exact V184 → V185 zip audit:
Presets: Amazing_SwarmUI_Presets_v83.json was replaced by Amazing_SwarmUI_Presets_v84.json. Its 84 presets are unchanged in every setting, description and thumbnail; 2 presets are new.
6 changed files: the model catalog and its size data (SAM 3.1 and Qwen3-VL 8B in the MiniMax-H3 Core, MiniMax-H3 Low VRAM and FastH3 Omni Core bundles), update_swarmui.py (a premium extension that could not be replaced during an update is installed again by the next update), the downloader app (shows v185), the regenerated Which_Bundles_Downloads_Which_Preset_Models.html and the MiniMax H3 prompt guide (new Genjutsu section).
NEW PRESETS — MINIMAX H3 GENJUTSU: PUT YOUR CHARACTER INTO ANY VIDEO
"MiniMax H3 Genjutsu - FastH3 Omni 8 Steps - 261024"
"MiniMax H3 Genjutsu - Ref2VA 20 Steps - 261024"
How it works: SAM 3.1 finds the person in every frame of your clip, and only that person is colour-inverted. MiniMax H3 then copies the clip's motion, timing, camera and lip movements, but not the performer's face, hair or clothes. A local Qwen3-VL 8B (Auto Describe) looks at your character image and six frames of the clip and writes the prompt for you: the new outfit, the action, the setting, the lighting and the framing. The clip's soundtrack is locked into MiniMax H3's audio stream for lip sync, and the saved video carries the original sound.
Where to find it: Quick Tools → Reset Params to Default, then the Presets tab → type Genjutsu in the filter → the preset card's menu → Direct Apply. Attach the clip and the character image with Add References (or Add A Reference With Trim to pick the part of the clip) and press Generate. Here is the FastH3 Omni preset after a real run of our example:
MiniMax H3 Genjutsu (on): turns the first reference video into the SAM 3.1 motion guide.
Genjutsu Background: video (character replacement) keeps the clip's location, lighting and framing; image (motion transfer) keeps your character image's background and takes only the motion.
Genjutsu Target: who to replace, in plain words: person, or man in red:1 to pick one of several people.
Genjutsu Auto Describe (on): the local Qwen3-VL 8B writes the outfit, action, setting, lighting and framing into the prompt. Leave the prompt empty, type short notes, or paste a complete MiniMax H3 prompt from ChatGPT, Claude or Gemini; a complete prompt is used as written.
Genjutsu Spoken Words (optional): the words spoken in the clip, for word-by-word lip sync.
Genjutsu Lock Clip Audio (on) and Genjutsu Match Clip Aspect (on): the clip's sound drives the lip movements, and the output takes the clip's aspect ratio at your Width x Height area.
Genjutsu Motion Guide Megapixels (0.15): the guide size; advanced options set the detection threshold and the mask grow.
The output follows the clip: our 9.8-second 1080x1920 clip gave 226 frames (9.4 seconds) at 768x1344, and the token meter beside the prompt shows that length and size before you press Generate.
Speed: FastH3 Omni 8 steps made the 9.4-second example in 4.5 minutes and Ref2VA 20 steps in 9.9 minutes in SwarmUI, first runs with model loading on an RTX 5090.
MODEL DOWNLOADER — SAM 3.1 AND QWEN3-VL 8B IN THE MINIMAX H3 BUNDLES
The MiniMax-H3 Core Bundle, the MiniMax-H3 Low VRAM Bundle and the FastH3 Omni Core Bundle now download everything the Genjutsu presets use:
SAM 3.1 Multiplex FP16 (1.63 GB, saved to Models/Stable-Diffusion): finds the person to replace in every frame.
Qwen3-VL 8B INT8 ConvRot (8.71 GB, saved to Models/text_encoders): writes the Genjutsu prompt. It is the same file as the Qwen Image 2.1 encoder, so if you already have it, the downloader checks it and keeps it.
New totals: MiniMax-H3 Core Bundle 79.91 GB (19 files), MiniMax-H3 Low VRAM Bundle 64.21 GB (19 files), FastH3 Omni Core Bundle 60.12 GB (11 files), sizes as the downloader shows them.
HOW TO UPDATE
Download SwarmUI_Model_Downloader_v185.zip from this post and extract it over your existing downloader folder.
Close SwarmUI and run Windows_Install_or_Update_SwarmUI.bat. It installs MiniMaxH3References 1.22.1 and our FoleyExtension with the Genjutsu nodes.
Keep the ComfyUI backend up to date. If SwarmUI uses our ComfyUI installation, close SwarmUI again (step 2 starts it, and a running backend blocks the update) and update that ComfyUI with our ComfyUI installer V145 (Windows_Install_Or_Update_ComfyUI.bat), which also brings the same Genjutsu presets to ComfyUI. If SwarmUI uses its own ComfyUI, open Server → Backends, edit the ComfyUI Self-Starting backend and set AutoUpdate to Always Update for one start.
Restart SwarmUI and its ComfyUI backend so the new extension and nodes register.
With SwarmUI running, run Windows_Preset_Delete_Import.bat to import Amazing_SwarmUI_Presets_v84.json. It backs up your current presets, thumbnails included, to utilities\presets_backups, deletes them all (your own presets too) and imports the new pack.
Run Windows_Start_Download_Models_App.bat and download the MiniMax-H3 Core Bundle or the FastH3 Omni Core Bundle (the files you already have are checked and kept).
ComfyUI Installer and Presets: https://www.patreon.com/posts/105023709
23 October 2026 Update V184
This is a Fibo 2.0, MOSS Image Edit and FastH3 Omni update. V184 adds four presets for Bria's Fibo 2.0, which generates and edits images from long JSON captions, two MOSS Image Edit presets that edit your photos with a short instruction, and four FastH3 Omni presets that turn your images, videos and sounds into video with synchronized audio, stills or audio in 8 steps. The model downloader gets the FastH3 Omni and Fibo 2.0 Core Bundles and a redesigned, much faster interface, keeps the model files whose header SwarmUI rewrote instead of downloading them again, and fixes the Qwen Image 2.1 text encoder that SwarmUI downloaded a second time.
Amazing SwarmUI Presets moves to V83 with 84 presets: 10 new presets, corrected descriptions for the four FLUX 2 8-step presets and the fixed text encoder in the two Qwen Image 2.1 Unified presets; the other 68 presets are unchanged. The new presets need the new FastH3 Omni Core Bundle, the Fibo 2.0 Core Bundle and the updated MOSS-Image Core Bundle of the model downloader.
All 84 presets as they appear in the SwarmUI Presets tab; the new presets are in the last rows:
Every preset still has its own animal. The ten new presets on their own:
Four Fibo 2.0 presets (text to image and image editing, Dev and Turbo), two MOSS Image Edit presets (30 steps and Turbo 8 steps) and four FastH3 Omni presets (references to video, long lip sync, text to image and audio only), each with its own artwork.
Exact V183 → V184 zip audit:
Presets: Amazing_SwarmUI_Presets_v81.json was replaced by Amazing_SwarmUI_Presets_v83.json. Of the 74 presets it keeps, 68 are unchanged in every setting, description and thumbnail; the four FLUX 2 8-step presets have a corrected description and the two Qwen Image 2.1 Unified presets select the fixed text encoder.
9 changed files: the model catalog and its size data (the FastH3 Omni and Fibo 2.0 models and bundles, the MOSS-Image edit LoRA, the Qwen Image 2.1 encoder names and the new source of the MiniMax H3 video VAE), the downloader app (new interface, shows v184), HF_model_downloader.py (keeps files whose header SwarmUI rewrote), update_swarmui.py (the new Fibo2 premium extension), the bundle report generator and the regenerated Which_Bundles_Downloads_Which_Preset_Models.html, and the MOSS-Image and MiniMax H3 prompt guides.
New: the Fibo 2.0 prompt guide in Automatic_Prompt_Generate_Enhance_To_Video_And_Image_Models_LLM_Instructions.
Windows_Install_or_Update_SwarmUI.bat brings SwarmUI with its native Fibo 2.0 support, the new Fibo2 premium extension, MossImage with Image edit mode, MiniMaxH3References 1.21.0, QwenImage21 1.0.2, SECoursesAudioTools and ComfyUI-TeaCache. The ComfyUI backend is updated separately (step 3 of HOW TO UPDATE).
NEW PRESETS — FIBO 2.0: TEXT TO IMAGE AND IMAGE EDITING, DEV AND TURBO
"Fibo 2 Dev - Advanced Text To Image - 261022"
"Fibo 2 Dev - Advanced Image Edit - 261022"
"Fibo 2 Turbo - Image Edit 4 Steps - 261022"
"Fibo 2 Turbo - Text To Image 4 Steps - 261022"
Fibo 2.0 by Bria is an 8B image model that generates and edits images. It was trained on long structured JSON captions that describe every object, the lighting, the camera and the text in the image, so a JSON caption gives the most control; plain text also works. All four presets load the models in INT8 ConvRot with the Qwen3-VL 4B INT8 ConvRot text encoder and the FLUX 2 VAE, set 1024×1024 and turn on our new Fibo2 extension. Measured in SwarmUI on an RTX 5090 at 1024×1024, for a new prompt with the models already loaded:
Dev - Advanced Text To Image: Bria's official recipe: 50 Euler steps, CFG 5 and Bria's JSON negative prompt from the model card. 17.3 seconds per image.
Turbo - Text To Image 4 Steps: Bria made the 4-step Edit Turbo model for editing, but in our tests it also renders text to image very well, lettering included: 4 steps, CFG 1, 1.9 seconds per image.
Dev - Advanced Image Edit: the official editing recipe with 50 steps and CFG 5: 34.7 seconds.
Turbo - Image Edit 4 Steps: Bria's distilled editing model, 4 steps, CFG 1: 2.5 seconds.
About 1 megapixel is the trained size (1024×1024, 1424×800, 1248×832 and the other shapes); 2 megapixels such as 1664×1248 also work well.
Where to find it: Presets tab → type Fibo 2 in the filter → click a preset card → Apply. Its settings fill the new Fibo 2.0 group of the parameters.
Fibo 2 Advanced (on): encodes your prompt with ComfyUI's native Fibo 2.0 text encoder exactly like Bria's pipeline. SwarmUI's generic encoder wraps Fibo 2.0 prompts differently and leaves edit images out of the text encoder.
Fibo 2 Sigma Shift: Bria's official shift, which changes with the image size.
Fibo 2 Edit Size: edits keep the size of the first image from 1 to 4 megapixels.
Editing: attach 1 to 4 images to the prompt with Add Image / Video / Audio (or drag them onto the prompt box) and write the instruction; @image1, @image2 refer to them. Masked edit: an Init Image plus a mask painted in the image editor (white = change).
Prompt syntax: a JSON caption in Bria's format (the new Fibo 2.0 LLM guide writes them for you) or plain text. {red|blue} picks one at random following the seed; (word:1.2) weights are removed because Qwen3-VL would read them as text.
Fibo 2 Prompt List: prompt batches from typed prompts, .txt files or a whole folder (one prompt per file, JSON captions included), like the MOSS-Image prompt batches. Generate prompt batch runs your main prompt first, then every extra prompt.
The extension acts only on Fibo 2.0 models. Apply any other preset and its own workflow runs unchanged.
NEW — ONE JSON PROMPT: FIBO 2 DEV 50 STEPS VS TURBO 4 STEPS
The same JSON caption and seed (261023) through both text-to-image presets. The caption asks for a cafe on a rainy evening with a green sign that reads "MORNING BREW" and a sidewalk chalkboard with "Fresh Coffee" and "Open Late":
Both presets wrote every word as asked, with the warm window light, the wet cobblestones and the guests inside. Turbo needs 1.9 seconds instead of 17.3, which suits trying many prompts; Dev is the full, non-distilled model with CFG and Bria's official recipe, and the model to train LoRAs on.
NEW — ONE PORTRAIT, THREE NEW EDIT PRESETS
A portrait made with Fibo 2 Turbo Text To Image, attached to the prompt as Image 1 and edited with each new edit preset and its default instruction (seed 261023):
Both Fibo 2.0 edits redrew the photo as an anime illustration and kept the pose, the clothes, the coffee cup and the flower shop. MOSS Image Edit replaced the street with a sunny beach and kept the same face, hair, clothes and pose, in 11.1 seconds.
NEW PRESETS — MOSS IMAGE EDIT: EDIT YOUR PHOTOS WITH ONE SENTENCE
"MOSS Image Edit - Advanced Image Edit - 261022"
"MOSS Image Edit - Turbo 8 Steps - 261022"
Both presets use MOSS-Image SFT INT8 ConvRot with ChenkinLab's Moss-Image-Edit LoRA, which the MossImage extension's new Image edit mode adds by itself together with the adapter's reference branch. Attach up to 4 images: image 1 is edited and the others are references, named image 1, image 2 (or @image1, @image2) in the instruction. Say what changes and what stays, for example "Change the background to a sunny beach, preserving the person's identity and pose."
Advanced Image Edit: the official edit recipe: 30 steps, CFG 4, shift 5 and an empty negative prompt. 11.1 seconds at 1024×1024 on an RTX 5090.
Turbo 8 Steps: the community 8-step DMD LoRA at 0.7 with 8 steps and CFG 1: 2.4 seconds. Turbo is close to the 30-step preset in many edits, but for our busy street portrait it kept the street on all five seeds we tried, while the 30-step preset gave a beach on all five. When Turbo ignores a big change such as a new background, use the 30-step preset.
The result has image 1's shape at up to 1024 pixels on the long side (MOSS Image Edit Reference Size; 1536 also works).
Colors, clothes, expressions, lighting, seasons, backgrounds and style edit best. Placing an object from a second image works; copying its exact look is weaker.
Where to find it: Presets tab → type MOSS Image Edit → click a preset card → Apply, then add the photo with Add Image / Video / Audio.
The preset sets MOSS Image Mode to Image edit (attached images). The new MOSS_Image_Edit_LoRA.safetensors (0.28 GB) is part of the updated MOSS-Image Core Bundle.
NEW PRESETS — FASTH3 OMNI: VIDEO WITH AUDIO FROM YOUR REFERENCES IN 8 STEPS
"FastH3 Omni - References to Video 8 Steps - 261022"
"FastH3 Omni - Long Lip Sync - References 8 Steps - 261022"
"FastH3 Omni - Text to Image with Optional References - 261022"
"FastH3 Omni - Audio Only with Optional References - 261022"
FastH3 Omni by FastVideo and Nuva Lab is MiniMax H3's references-to-video-and-audio model distilled to 8 steps. One model makes video with synchronized 32 kHz stereo audio from your prompt plus up to 9 images, 3 videos and 3 audio files, with no turbo LoRA: 8 steps, CFG 1. The presets load it in INT8 ConvRot with the INT8 ConvRot Qwen3-VL 32B text and vision encoder. The References to Video and Long Lip Sync presets also run the sparse attention FastH3 was trained with (VSA): sampling was about 1.4× faster for 5 seconds and 1.9× for 10 seconds of 1344×768 video on an RTX 5090, with the same quality in our tests.
References to Video: 1344×768, 124 frames (5.2 seconds) with audio: 53 to 57 seconds in our runs with one reference image (one run took 108 seconds; we could not find why). It also works as text to video without references.
Long Lip Sync: put the complete soundtrack in Init Audio and a portrait beside the prompt as @image1; one 24 FPS video follows the whole audio in overlapping 8-step windows at 832×1248. About 3.5 minutes for 15 seconds of speech (3.32 minutes in our SwarmUI test).
Text to Image with Optional References: stills from text, or with up to 9 reference images: 9.8 seconds at 1344×768 for a new prompt with one reference image.
Audio Only with Optional References: 32 kHz stereo audio returned as a lossless FLAC: 8 seconds of audio in 7.2 seconds. Put spoken words inside <d>[English] ... </d>.
Where to find it: first Quick Tools → Reset Params to Default, so no MiniMax H3 turbo LoRA from an earlier preset stays selected. Then Presets tab → type FastH3 Omni → click a preset card → Apply, add your images, videos or audio with Add References and mention them in the prompt as @image1, @video1, @audio1.
The prompt describes the woman from @image1 in front of a flower shop: she smiles, raises her coffee cup, greets the viewer with a spoken line in <d>, laughs and takes a sip. Nine frames from the start to the end of the clip:
Her face, curly hair, freckles, earrings and clothes stay those of the portrait in every frame, and the clip follows the prompt from the greeting to the sip. Video and its 32 kHz stereo soundtrack came out of one 8-step run.
NEW — MODEL DOWNLOADER: FASTH3 OMNI AND FIBO 2.0 CORE BUNDLES
FastH3 Omni Core Bundle (49.78 GB, 9 files): FastH3 Omni INT8 ConvRot (20.69 GB), the MiniMax H3 Qwen3-VL 32B INT8 ConvRot text and vision encoder (25.28 GB), the MiniMax H3 video VAE (2.62 GB) and audio VAE (0.56 GB), plus the face models the presets' face refinement and mouth pass use. MiniMax H3 users already have the audio VAE and the face models, and the INT8 ConvRot encoder if they chose it. The video VAE now comes from Comfy-Org, the file SwarmUI itself downloads: if an earlier downloader saved Kijai's copy at the same path, it is replaced once (2.62 GB).
Fibo 2.0 Core Bundle (20.50 GB, 4 files): Fibo 2.0 Dev and Edit Turbo in INT8 ConvRot (7.83 GB each), the Qwen3-VL 4B INT8 ConvRot text encoder (4.52 GB, saved as qwen3vl_4b_int8_convrot.safetensors so it never replaces the qwen3vl_4b.safetensors other models use) and the shared FLUX 2 VAE (0.31 GB).
MOSS-Image Core Bundle (21.91 GB, 7 files): now also downloads MOSS_Image_Edit_LoRA.safetensors (0.28 GB) for the MOSS Image Edit presets.
Where to find it: run Windows_Start_Download_Models_App.bat. The FastH3 Omni Core Bundle and the Fibo 2.0 Core Bundle are the first two bundles of the SwarmUI Bundles tab.
Optional, under Video Generation Models → FastH3 Omni Video Models: FastH3 Omni in BF16 (41.20 GB), FP8 (21.46 GB) and NVFP4 QAD (12.83 GB, for RTX 50 series and other Blackwell GPUs). Under Image Generation Models → Fibo 2.0 Models: both models in BF16 (15.32 GB each), the BF16 text encoder (8.27 GB, saved as qwen3vl_4b.safetensors, the name the FP8 encoder of FLUX 3 Klein and Krea 2 also uses, so keep one precision) and Comfy-Org's Fibo 2.0 turbo LoRA (1.85 GB), which lets Dev and your Dev LoRAs run in 4 steps; the presets do not select it.
NEW — THE DOWNLOADER KEEPS MODELS WHOSE HEADER SWARMUI REWROTE
The first time SwarmUI uses a model, it writes the model hash into the file's header (SwarmUI's ImageMetadataIncludeModelHash setting, on by default). The tensor data stays the same, but the file gets a little larger or its bytes change, so V183 failed its size or SHA-256 check and downloaded the whole model again. V184 reads the original header from Hugging Face, checks your file's tensor data behind it, and keeps the file when every tensor byte matches. A checked file is remembered, so the next run skips it at once.
The log says "[VERIFIED] ... the tensor data matches the Hugging Face file; only the header metadata differs" and "[SKIP] ... already complete". On our test PC, 59 model files with 581 GB in total were a little larger than their Hugging Face files, the sign of SwarmUI's header rewrite. The Fibo 2.0 Core Bundle's four files were checked and kept in 25 seconds instead of downloading 20.50 GB again. Files whose tensor data differs are still downloaded again, as before.
FIXED — SWARMUI NO LONGER DOWNLOADS THE QWEN IMAGE 2.1 TEXT ENCODER AGAIN
Thank you to the supporter who reported this: after the Qwen Image 2.1 Core Bundle, SwarmUI said the text encoder was missing and downloaded it again. The downloader saved the INT8 ConvRot Qwen3-VL 8B encoder as qwen3vl_8b.safetensors, but SwarmUI's native Qwen Image 2.1 support loads qwen3vl_8b_int8_convrot.safetensors; qwen3vl_8b.safetensors is the name SwarmUI uses for its Ideogram 4 encoder. The folder was right: the downloader puts text encoders in the folder SwarmUI's own settings name (Models/text_encoders or Models/clip).
V184 keeps the original names of all three Qwen Image 2.1 text encoders (INT8 ConvRot, BF16 and W4A8). The two Qwen Image 2.1 Unified presets and our QwenImage21 extension 1.0.2 select qwen3vl_8b_int8_convrot, so the presets and plain Qwen Image 2.1 generations share one 8.71 GB file. In our test the downloader got it in 1 min 28 s (1 min 46 s with the SHA-256 check), and SwarmUI then generated with it without downloading anything.
Downloaded with an older version? Older downloaders saved every Qwen Image 2.1 text encoder, and Ideogram 4's too, as qwen3vl_8b.safetensors. Check that file's size in SwarmUI's text encoder folder (Properties shows the bytes; a file SwarmUI has used is about 64 KB larger):
9,350,798,360 bytes: the Qwen Image 2.1 INT8 ConvRot encoder. Rename it to qwen3vl_8b_int8_convrot.safetensors. If SwarmUI already downloaded that file itself, the old one is a spare 8.71 GB copy.
17,534,334,616 bytes: the Qwen Image 2.1 BF16 encoder, rename it to qwen3vl_8b_bf16.safetensors. 6,312,105,364 bytes: the W4A8 encoder, rename it to qwen3vl_8b_w4a8.safetensors.
10,588,637,512 or 6,305,221,764 bytes: Ideogram 4's own FP8 or NVFP4 encoder: keep it.
After a rename, Ideogram 4 gets its own encoder from the Ideogram 4 Core Bundle or SwarmUI's automatic download. The Qwen Image 2.1 Unified presets do not download models, so they need qwen3vl_8b_int8_convrot.safetensors: rename the old file, or run the Qwen Image 2.1 Core Bundle again (it keeps the model and the VAE after checking them).
UPDATED — REDESIGNED, MUCH FASTER MODEL DOWNLOADER
V183 refreshed its status panels through a timer every half second, and after a search or Expand All every refresh blocked the browser for 0.2 to 0.35 seconds. V184 updates only what changed, and search and the precision filter no longer redraw the whole page. Measured in Chrome on our PC with the same catalog, V183 → V184:
Search while typing: results shown about 0.44 seconds after the last key → already shown 0.12 seconds after it.
Clear Search: 673 ms → 10 ms. Expand All: 648 ms → 194 ms.
Precision filter "Display only GGUF models": 851 ms → 11 ms.
Idle page after these steps: busy 65% of the time → 0%.
New look: bundle and model counts in the header, a light / dark switch that your browser remembers, model and bundle cards with an exact size on every file, colored buttons, a colored Activity Log and Active Transfers cards with a progress bar, size, speed and ETA (size and speed also while a file is being checked).
Search shows up to 50 result cards (was 15) with the category, the size or the bundle total; click a card to queue it.
Exact sizes: files below 0.1 GB show in MB, so a 4.8 MB file no longer reads "Size: 0 GB" and counts as unknown in the totals.
The status refresh pauses while the browser tab is hidden; downloads continue. The app's download, queue and cancel logic is unchanged; the only change to file checks is the header check above.
NEW AND UPDATED PROMPT GUIDES FOR LLMS
New: Fibo_2.0_Enchance_Prompt_Feed_For_LLMs.txt. Paste it into ChatGPT, Claude, Gemini or a local LLM, describe the image you want or the edit (attach the images to the LLM chat for an edit), and it returns a JSON caption in Bria's format. Its first part explains how the SwarmUI and ComfyUI presets use the prompt: where it goes, image labels, resolutions and the negative prompt. It is based on Bria's model card and the caption schemas of Bria's own captioning library, plus our tests.
MOSS_Image_Enchance_Prompt_Feed_For_LLMs.txt: new Part 7 for the MOSS Image Edit presets: one short plain-text instruction, the change first and then what stays, and how to name several images.
MiniMax_H3_Enchance_Prompt_Feed_For_LLMs.txt: new FastH3 Omni section: write FastH3 prompts exactly like MiniMax H3 references-to-video prompts.
UPDATED — BUNDLE REPORT AND FLUX 2 PRESET DESCRIPTIONS
Which_Bundles_Downloads_Which_Preset_Models.html is regenerated for V83: all 84 presets, and every new preset is fully covered by its core bundle. The report generator now knows FastH3 Omni's automatic text encoder and VAEs, the edit LoRA that MOSS Image edit mode loads, Qwen Image 2.1's own text encoder and VAE, and shows small files in MB.
The four FLUX 2 8-step presets told you to add --no-cache to the backend arguments for out-of-RAM errors. ComfyUI's argument is --cache-none; their descriptions are corrected.
TESTED — REAL GENERATIONS IN SWARMUI
Every result shown in this post is a real SwarmUI run on 23 October on an RTX 5090 (SwarmUI d74f8ce4, ComfyUI 6cbc3862) after Windows_Install_or_Update_SwarmUI.bat had installed the new extensions and nodes, with the V83 presets imported by Windows_Preset_Delete_Import.bat and applied by real clicks: all four Fibo 2.0 presets, both MOSS Image Edit presets and three FastH3 Omni presets, with seed 261023. The input portrait and four of the five MOSS seed checks ran earlier the same day, before that update. The FastH3 Omni Long Lip Sync preset was tested separately in SwarmUI with a portrait and 15 seconds of speech. The downloader ran from its own bat file: it kept the Fibo 2.0, MOSS-Image, Ideogram 4 and MiniMax H3 files that SwarmUI had rewritten, saved the Qwen Image 2.1 encoder under its own name, and the Qwen Image 2.1 Unified preset then generated without any download in SwarmUI.
CUSTOM NODE PACKS WE DEVELOP OR MAINTAIN UPDATED
SECoursesAudioTools
New Fibo 2.0 nodes for our SwarmUI and ComfyUI Fibo 2.0 presets: prompt and image encoding with ComfyUI's native Fibo 2.0 text encoder (JSON captions encoded like Bria's pipeline, edits keep the first image's size from 1 to 4 MP), Bria's size-dependent shift, and masked edits with ComfyUI's native Mask Editor.
ComfyUI now puts the Fibo 2.0 negative prompt into Bria's chat template itself (22 October), so our nodes no longer wrap it a second time. With the same seed, the four Fibo 2.0 presets give pixel-identical images to before.
New MOSS-Image Edit References node: the reference branch of ChenkinLab's Moss-Image-Edit adapter on ComfyUI's native MOSS-Image model, used by the MOSS Image Edit presets in both apps.
H3 Long Lip Sync without a soundtrack now says that the soundtrack is needed instead of failing with a TypeError.
ComfyUI-TeaCache
The MiniMax H3 Speed Optimizer has a new VSA mode for FastH3 models: ComfyUI's native Video Sparse Attention, used the way FastH3 was trained. Its output is bit-identical to ComfyUI's own VSA node, and sampling is 1.37× (5 seconds) and 1.87× (10 seconds) faster than dense attention at 1344×768 on an RTX 5090.
The default face-pass prompt no longer asks for "beard strands": on a FastH3 Omni image of a woman, the face pass had drawn a beard. Saved workflows keep their own text.
SWARMUI PREMIUM EXTENSIONS UPDATED
Fibo2 1.0.0 (new): Fibo 2 Advanced encoding like Bria's pipeline, Bria's size-dependent shift, automatic edit size (1 to 4 MP), masked edits from Init Image plus mask, the INT8 text encoder as default and prompt batches. It acts only on Fibo 2.0 models. Windows_Install_or_Update_SwarmUI.bat installs it.
MossImage: new Image edit (attached images) mode: it adds the Moss-Image-Edit LoRA and the reference branch, gives the result image 1's reference size, turns @image1 into "image 1" and keeps the attached images out of SwarmUI's text encoder.
MiniMaxH3References 1.21.0: the FastH3 sparse attention choice vsa; when SwarmUI's NVFP4 Qwen3-VL 32B encoder is missing and the INT8 ConvRot encoder of the FastH3 Omni Core Bundle is installed, MiniMax H3 models use it instead of downloading another 14.61 GB file; Audio Only works with FastH3 Omni without a routing warning.
QwenImage21 1.0.2: the Qwen Unified Text Encoder default is qwen3vl_8b_int8_convrot, the file SwarmUI's native Qwen Image 2.1 support loads.
HOW TO UPDATE
Download SwarmUI_Model_Downloader_v184.zip from this post and extract it over your existing downloader folder.
Run Windows_Install_or_Update_SwarmUI.bat. It updates SwarmUI to the version with native Fibo 2.0 support (added on 21 October), installs the new Fibo2 premium extension, MossImage with Image edit mode, MiniMaxH3References 1.21.0 and QwenImage21 1.0.2, and updates SECoursesAudioTools and ComfyUI-TeaCache.
Update the ComfyUI backend to its latest version. Fibo 2.0 needs ComfyUI from 21 October or newer and FastH3 Omni from 18 October or newer. With the updated SECoursesAudioTools of step 2, Fibo 2 Dev text to image encodes its negative prompt exactly like Bria's pipeline only on ComfyUI from late 22 October or newer. If SwarmUI uses our ComfyUI installation, update it with our ComfyUI installer V144 (Windows_Install_Or_Update_ComfyUI.bat). If SwarmUI uses its own ComfyUI, open Server → Backends, edit the ComfyUI Self-Starting backend and set AutoUpdate to Always Update for one start. You can switch it back afterwards. Our ComfyUI installer V144 also brings the same Fibo 2.0, MOSS Image Edit and FastH3 Omni presets and the new prompt guides to ComfyUI.
Restart SwarmUI and its ComfyUI backend so the new extensions and nodes register.
With SwarmUI running, run Windows_Preset_Delete_Import.bat to import Amazing_SwarmUI_Presets_v83.json. It backs up your current presets, thumbnails included, to utilities\presets_backups, deletes them all (your own presets too) and imports the new pack.
Run Windows_Start_Download_Models_App.bat and download the FastH3 Omni Core Bundle, the Fibo 2.0 Core Bundle and the MOSS-Image Core Bundle (the files you already have are checked and kept). Qwen Image 2.1 users: see the text encoder fix above.
ComfyUI Installer and Presets: https://www.patreon.com/posts/105023709
18 October 2026 Update V183
This is a MOSS-Image update. V183 adds six MOSS-Image presets and our new MossImage premium extension to SwarmUI's new native support for OpenMOSS's MOSS-Image: the official quality recipe, Fast 25 Steps and Turbo 8 Steps presets, two Base model presets, visual layout boxes, JSON layout prompts and prompt batches from .txt files. MOSS-Image renders the text you quote in your prompt, such as shop signs, poster titles and labels. The model downloader gets the MOSS-Image Core Bundle (21.64 GB), the zip adds a MOSS-Image prompt-writing guide for LLMs, and AvatarForever gets live previews and new tested lip motion settings.
Amazing SwarmUI Presets moves to V81 with 74 presets: 6 new MOSS-Image presets, an updated AvatarForever preset and new artwork for the three FLUX 3 Klein presets; the other 64 presets are unchanged. The new presets need the new MOSS-Image Core Bundle (21.64 GB) of the model downloader.
All 74 presets as they appear in the SwarmUI Presets tab; the six MOSS-Image presets are in the last two rows:
The three FLUX 3 Klein presets just before them have new artwork as well (an otter, a flamingo and a meerkat), so every preset has its own animal again. The six new MOSS-Image presets on their own:
Like every other preset, each MOSS-Image preset has its own artwork that shows its task: Text To Image 50 Steps, Fast 25 Steps, Turbo 8 Steps, Layout Boxes, and the Base Model Text To Image and Layout Boxes presets.
NEW PRESETS — MOSS-IMAGE: QUALITY, FAST, TURBO, LAYOUT BOXES AND BASE MODEL
"MOSS Image - Advanced Text To Image - 261017"
"MOSS Image - Fast 25 Steps - 261017"
"MOSS Image - Turbo 8 Steps - 261017"
"MOSS Image - Advanced Layout Boxes - 261017"
"MOSS Image Base - Advanced Text To Image - 261018"
"MOSS Image Base - Advanced Layout Boxes - 261018"
All six load MOSS-Image in INT8 ConvRot with SwarmUI's own Qwen3.5 9B INT8 ConvRot text encoder and the FLUX 2 VAE, set 1024×1024 and turn on our MossImage extension. MOSS-Image is a 5B single-stream image model that reads plain language, tags or JSON layouts, with prompts up to about 4K tokens. Measured in SwarmUI on an RTX 5090 at 1024×1024, for a new prompt with the models already loaded:
Advanced Text To Image: the official quality recipe (50 Euler steps, normal scheduler, CFG 3, shift 5, negative-branch skip of layer 22 and the official negative tags) plus PyTorch attention and the SFT de-band fix. 12.9 seconds per image.
Fast 25 Steps: 25 ER-SDE steps with Score Rescaling 0.95 and layers 22 and 24 skipped from 10% to 70% of sampling: 6.9 seconds, nearly twice as fast. Large lettering held up in our tests; small text can get an occasional misspelling.
Turbo 8 Steps: the community 8-step DMD LoRA (lodestones, Apache-2.0) at 0.7 with 8 Euler steps and CFG 1: 1.8 seconds per image. The LoRA is a work in progress, so its images differ more from the official model and can show stray small text.
Advanced Layout Boxes: the official recipe with visual layout boxes, ready with three example boxes.
Base - Advanced Text To Image and Base - Advanced Layout Boxes: the raw Base model with the official 50-step recipe (13.0 seconds for Base Text To Image). Base gives more varied images and is the model that LoRAs trained on Base expect; SFT gives clearer text grouping and layouts. With the same seed, 80 steps gave no visible gain over 50 in our tests, so the Base presets keep 50. There is no Base Turbo preset, because the 8-step LoRA was trained on SFT.
Where to find it: Presets tab → type MOSS Image in the filter → click a preset card → Apply. Its settings fill the new MOSS-Image group of the parameters.
The MOSS-Image group and the preset settings in detail:
Seed: -1 gives a new image every time; a fixed seed repeats the same image (pixel-identical in our tests).
Mode: Text / JSON prompt or Layout boxes.
Negative Skip Layers (22): the official skip-layer guidance; the negative branch runs without these layers. Skip Start / End limit it to part of sampling. It needs CFG above 1.
Attention (PyTorch): with PyTorch attention the INT8 ConvRot files gave images close to the official BF16 outputs. Setting it here also protects the BF16 files, which gave black images with a --use-sage-attention backend in our RTX 3090 tests.
Score Rescaling (1 is off): the Fast preset uses 0.95; lower values give more detail, higher values smoother images.
SFT De-Band (on): MOSS-Image SFT draws faint vertical color stripes on skies, fog, walls and backdrops. The fix loads MATLOWAI's de-band LoRA (Apache-2.0, in the MOSS-Image Core Bundle) for SFT models only; Base models never get it.
Your usual SwarmUI controls keep working: resolution, batch size, variation seed and LoRAs. 1 to 2 megapixel resolutions are recommended.
The extension acts only on MOSS-Image models. Apply any other preset and its own workflow runs unchanged.
NEW — ONE PROMPT, FOUR PRESETS: QUALITY, SPEED AND THE BASE MODEL
The same prompt and seed (261018) through four of the new presets. The prompt asks for a bookshop on a rainy evening with a sign that reads "THE LAST CHAPTER", a chalkboard that says "New arrivals / Hot tea inside", a sleeping orange cat in the window and string lights:
All four rendered the sign and the chalkboard text as written, with the sleeping cat in the window. Fast needs about half the time of the 50-step preset, and Turbo returns an image in 1.8 seconds, which suits trying many prompts. The 50-step presets are the most faithful; Base gives its own, more varied take on the same prompt.
NEW — VISUAL LAYOUT BOXES: PUT TEXT AND OBJECTS WHERE YOU WANT THEM
Choose Layout boxes mode, draw boxes on a canvas that has the shape of your output image and describe each box. The main prompt becomes the overall image caption. The extension builds the layout with ComfyUI's native MOSS-Image layout nodes, so the model places each box's content where you drew it.
Draw a box with the mouse and drag it to move it; select a box and drag its lower-right corner to resize it.
Each box is an Object, Character or Text / lettering box. Text boxes render their exact text, line breaks included, and every edge can be typed in percent.
Add, delete and reorder up to 32 boxes, or open the JSON.
Boxes guide the composition; they are not inpainting masks.
Where to find it: apply the MOSS Image - Advanced Layout Boxes preset or its Base version. The canvas is in the MOSS-Image group.
The preset's three boxes put the title "EXPLORE THE ALPS" at the top, the mountain lake illustration in the middle and the subtitle "Summer 2026 | Lakes, trails and fresh air" at the bottom, as drawn, in 13.5 seconds of generation time.
NEW — PROMPT SYNTAX AND JSON LAYOUT PROMPTS
MOSS-Image reads the prompt with Qwen3.5 as plain language, tags or JSON. Long, specific prompts work best; put the text that must appear in the image in quotes and use real line breaks to control its lines.
{red|blue} picks one at random, following the seed.
[red:blue:0.5] switches from one word to the other halfway through sampling, and [red|blue] alternates them from step to step.
(word:1.2) weights are removed, because Qwen3.5 would read them as literal text; describe emphasis in words instead. Times and ratios such as (10:30) or (4:3) stay.
A complete JSON layout passes through unchanged, and pixel drafts from the official MOSS-Image prompt enhancer are converted to the model's own box format at your output size. SwarmUI's own prompt tags keep working.
NEW — PROMPT BATCHES FROM .TXT FILES OR A WHOLE FOLDER
Generate a series with the same settings. Add prompt adds a prompt you type; Import .txt and Import folder load text files, one complete prompt per file, JSON layouts included. Generate prompt batch runs your main prompt first and then every extra prompt; Stop after current ends the batch after the current prompt finishes. The normal Generate button still generates only the main prompt.
Where to find it: MOSS-Image group → MOSS Image Prompt List, in every MOSS-Image preset.
Three imported .txt files plus the main prompt gave four images with the Turbo 8 Steps preset in 7.4 seconds, from click to finish: the bookshop, a "COASTAL ESCAPE" travel poster, a "MORNING RITUAL" coffee cup and a sleeping fox. A batch takes up to 63 extra prompts. If a preset is only selected and not applied, the batch asks you to apply it first, so the preset's own prompt cannot replace your batch prompts.
NEW — MODEL DOWNLOADER: MOSS-IMAGE CORE BUNDLE (21.64 GB)
One click downloads everything the six presets use:
moss_image_sft_int8_convrot.safetensors (5.04 GB): the quality fine-tuned SFT model, the preset default.
moss_image_base_int8_convrot.safetensors (5.04 GB): the raw Base model for the Base presets, LoRA training and variety.
Qwen3.5 9B INT8 ConvRot text encoder (9.08 GB), saved as qwen3.5_9b_int8_convrot.safetensors, the file and name SwarmUI itself downloads for MOSS-Image, so SwarmUI reuses it.
FLUX 2 VAE (0.31 GB), saved as VAE/Flux/flux2-vae.safetensors, the same file FLUX 2 Klein and FLUX 3 Klein use.
MOSS_Image_SFT_DeBand_L37-41.safetensors (1.05 GB, MATLOWAI, Apache-2.0): the SFT banding fix that the presets apply automatically.
MOSS_Image_SFT_8_Steps_DMD_LoRA.safetensors (1.11 GB, lodestones, Apache-2.0): the 8-step LoRA of the Turbo preset, saved from lodestones' latest release file, which can change upstream.
Where to find it: run Windows_Start_Download_Models_App.bat. The MOSS-Image Core Bundle is the first bundle of the SwarmUI Bundles tab.
Optional, under Image Generation Models → MOSS-Image Models: the original BF16 SFT and Base models (9.47 GB each; BF16 Base is the recommended precision for LoRA training and fine-tuning), a smaller Qwen3.5 9B W4A8 GPTQ text encoder for low RAM and VRAM (6.84 GB; select it in the Qwen Model parameter), the FLUX 2 small-decoder VAE that decodes faster with less VRAM (0.23 GB; select it as the VAE) and an extra banding fix for stubborn smooth skies (1.05 GB).
Why INT8 ConvRot with PyTorch attention: on an RTX 3090 the INT8 ConvRot files run about twice as fast as BF16, and with PyTorch attention their images were close to the official BF16 outputs. The INT8 files' own INT8 attention is about 20% faster, but it changed the images noticeably in our tests, so the presets use PyTorch attention. Our ComfyUI graph with these settings reproduced official MOSS-Image Space examples from the same seed, with the same composition, subject and details.
UPDATED — AVATARFOREVER: LIVE PREVIEWS, TESTED LIP MOTION SETTINGS AND A HISTORY FIX
Live previews: the AvatarForever sampler now shows a preview frame after every denoising step, starting at step 1, and the finished video after export; in SwarmUI the previews appear in the normal generation preview. They use taeltx2_3.safetensors (0.02 GB) when you have it, now part of the LTX 2.3 Avatar Forever Core Bundle (44.53 GB), and a fast latent preview otherwise. Previews do not change the video: with previews on and off, the same 452-frame MP4 came out byte for byte in our test.
Tested lip motion settings: the AvatarForever Unified Image + Audio or Text + Audio preset now renders an exact 512×768 portrait canvas with 16 latent frames per chunk (was 4) and a prompt that asks for expressive mouth movement. In a six-second comparison with the same portrait, song, prompt and seed, 4-frame chunks kept the lips closed and 16-frame chunks gave visible mouth movement, which held through the complete 18-second song. At 1024×1024 the mouth stayed mostly closed in the same test even with 16-frame chunks, so the preset keeps 512×768. Larger chunks use more memory; check a short sample when you change the image, the audio or the resolution.
History fix (AvatarForever 1.0.3): an interrupted video export could leave an incomplete MP4 that kept SwarmUI's History waiting, and with several browser tabs open even finished images stopped loading. SwarmUI now skips the metadata of such files; complete files and generation settings are unchanged.
Where to find it: Presets tab → type AvatarForever → click the preset card → Apply, then attach a speech or song to the prompt and, if you like, a portrait as Init Image.
NEW — MOSS-IMAGE PROMPT GUIDE FOR LLMS
New file: Automatic_Prompt_Generate_Enhance_To_Video_And_Image_Models_LLM_Instructions/MOSS_Image_Enchance_Prompt_Feed_For_LLMs.txt. Give it to ChatGPT, Claude, Gemini or a local LLM together with your idea, and it writes MOSS-Image prompts: detailed captions, exact text for signs and posters, and layout drafts with boxes that the presets convert to MOSS-Image's own coordinates at your output size. Its first part explains how the SwarmUI and ComfyUI presets use the prompt (recipe, prompt syntax, layout mode and resolutions); it is based on the official MOSS-Image user guide and its prompt enhancer.
UPDATED — BUNDLE REPORT AND DOWNLOADER APP
Which_Bundles_Downloads_Which_Preset_Models.html is regenerated for V81: all 74 presets, and every MOSS-Image preset is fully covered by the MOSS-Image Core Bundle. The report generator now knows SwarmUI's default MOSS-Image text encoder and VAE, the de-band LoRA that the MossImage extension loads by itself for SFT models, and AvatarForever's live-preview file.
Windows_Start_Download_Models_App.bat now installs the downloader's packages in pip's isolated mode, so your user pip configuration and PIP_ option variables no longer affect its install.
The downloader app shows v183.
HOW TO UPDATE
Download SwarmUI_Model_Downloader_v183.zip from this post and extract it over your existing downloader folder.
Run Windows_Install_or_Update_SwarmUI.bat. It updates SwarmUI to the version with native MOSS-Image support (added on 17 October), installs the new MossImage premium extension and AvatarForever 1.0.3, and updates SECoursesAudioTools.
Update the ComfyUI backend: MOSS-Image needs ComfyUI from 16 October or newer. If SwarmUI uses our ComfyUI installation, update it with our ComfyUI installer V143 (Windows_Install_Or_Update_ComfyUI.bat). If SwarmUI uses its own ComfyUI, open Server → Backends, edit the ComfyUI Self-Starting backend and set AutoUpdate to Always Update for one start. You can switch it back afterwards. Our ComfyUI installer V143 also brings the same six MOSS-Image presets and the MOSS-Image prompt guide to ComfyUI.
Restart SwarmUI and its ComfyUI backend so the new extension and nodes register.
With SwarmUI running, run Windows_Preset_Delete_Import.bat to import Amazing_SwarmUI_Presets_v81.json. It backs up your current presets, thumbnails included, to utilities\presets_backups, deletes them all (your own presets too) and imports the new pack.
Run Windows_Start_Download_Models_App.bat and download the MOSS-Image Core Bundle. AvatarForever users: also download taeltx2_3.safetensors (0.02 GB) for the live previews; it is in the LTX 2.3 Avatar Forever Core Bundle and under Video Generation Models → LTX 2.3 Video Models.
ComfyUI Installer and Presets: https://www.patreon.com/posts/105023709
16 October 2026 Update V182
This is a FLUX 3 Klein update. V182 brings Black Forest Labs' new FLUX 3 Klein image model to SwarmUI with three advanced presets and our new FLUX 3 Klein premium extension: text to image, image editing with up to 10 reference images, visual layout boxes, ComfyUI-style prompt syntax and prompt batches from .txt files. The model downloader gets the FLUX 3 Klein Core Bundle with our HQ INT8 ConvRot model (10.85 GB in total), and the zip adds a FLUX 3 Klein prompt-writing guide for LLMs.
Amazing SwarmUI Presets moves to V80 with 68 presets: 3 new FLUX 3 Klein presets, and the other 65 presets are unchanged. The new presets need the new FLUX 3 Klein Core Bundle (10.85 GB) of the model downloader.
All 68 presets as they appear in the SwarmUI Presets tab; the three FLUX 3 Klein presets are in the last row:
Like every other preset, each FLUX 3 Klein preset has its own artwork that shows its task: Text To Image, Image Edit or Layout Boxes.
NEW PRESETS — FLUX 3 KLEIN: TEXT TO IMAGE, IMAGE EDIT AND LAYOUT BOXES
"FLUX 3 Klein - Advanced Text To Image - 261015"
"FLUX 3 Klein - Advanced Image Edit - 261015"
"FLUX 3 Klein - Advanced Layout Boxes - 261015"
All three load our HQ INT8 ConvRot FLUX 3 Klein model with SwarmUI's own Qwen3-VL 4B text encoder and the FLUX 2 VAE, and set 1024×1024, 16 steps, Euler/simple and CFG 1 (CFG 1 ignores the negative prompt). Each preset turns on our FLUX 3 Klein extension in its own mode: Text / optional references, Reference edit or Layout / optional references. Measured in SwarmUI on an RTX 5090 at 1024×1024: about 2 seconds of generation time for an image and 3.9 seconds for an edit with one reference image.
Where to find it: Presets tab → type FLUX 3 Klein in the filter → click a preset card → Apply. Its settings fill the new FLUX.3 Klein group of the parameters.
The FLUX.3 Klein group and the preset settings in detail:
Seed: -1 gives a new image every time; a fixed seed repeats the same image (pixel-identical in our tests).
Mode: text to image, reference edit or layout boxes. Reference images are optional in the text and layout modes.
Reference Megapixels: the size of each attached image, from 0.1 to 4 megapixels (1 is about 1024×1024), independent of the output size.
Your usual SwarmUI controls keep working: resolution, batch size, variation seed and compatible FLUX 3 Klein LoRAs.
The extension acts only on FLUX 3 Klein models. Apply any other preset and its own workflow runs unchanged.
NEW — EDIT IMAGES WITH UP TO 10 REFERENCE IMAGES (@image1, @image2)
Attach images to the prompt with Add Image / Video / Audio and cite them in your prompt: @image1 is the first attachment, @image2 the second, and so on (@img1, @picture1 and @pic1 work too). Describe the complete result: what changes and what stays. The extension turns these names into FLUX 3 Klein's own reference labels.
Up to 10 images. Each keeps its aspect ratio, and transparent areas are filled with white.
Reference edit needs at least one image. If the prompt cites an image that is not attached, SwarmUI shows a clear error; video and audio attachments are refused with a message.
Where to find it: apply the FLUX 3 Klein - Advanced Image Edit preset, attach your photo above the prompt and describe the result.
Here the white mug of the attached photo became cobalt blue in 3.9 seconds of generation time, while the croissant, the window, the light and the camera angle stayed. To take one object out of several references, name the image it comes from and say which other objects should be absent: in our test with this coffee photo as @image1 and a red teapot as @image2, asking for the vessel from @image2 gave the teapot alone.
NEW — VISUAL LAYOUT BOXES: PUT EVERY OBJECT WHERE YOU WANT IT
Draw boxes on a canvas that has the shape of your output image, describe the object of each box, and cite the boxes in your prompt as <obj_1>, <obj_2>. SwarmUI turns them into FLUX 3 Klein's native layout table, so the model places each object in its box.
Draw a box with the mouse and drag it to move it; select a box and drag its lower-right corner to resize it.
Each box is an Object or Text / lettering with the exact text to render, and its edges can be typed in percent.
Add, delete and reorder up to 32 boxes, or open the JSON. Reference images still work in this mode.
Boxes guide the composition; they are not inpainting masks.
Where to find it: apply the FLUX 3 Klein - Advanced Layout Boxes preset. The canvas is in the FLUX.3 Klein group.
The preset's two boxes put the red teapot on the left and the blue cup on the right, as drawn, in 2.0 seconds of generation time.
NEW — COMFYUI-STYLE PROMPT SYNTAX
The FLUX 3 Klein presets understand the prompt syntax of our ComfyUI presets, next to SwarmUI's own prompt tags:
(ceramic:1.1) weights a phrase.
[red:blue:0.5] switches from one word to the other halfway through sampling.
[red|blue] alternates the words from step to step.
<break> splits the prompt into separately encoded parts.
{red|blue} picks one at random, following the seed.
SwarmUI tags such as <random:red|blue>, <fromto[0.5]:red|blue> and <alternate:red|blue> keep working.
A complete native caption with its JSON layout table, as the new prompt guide writes it, works in the text and reference edit modes.
NEW — PROMPT BATCHES FROM .TXT FILES OR A WHOLE FOLDER
Generate a series with the same settings and reference images. Add prompt adds a prompt you type; Import .txt and Import folder load text files, one complete prompt per file, in natural file-name order. Generate prompt batch runs your main prompt first and then every extra prompt; Stop after current ends the batch after the current prompt finishes. The normal Generate button still generates only the main prompt.
Where to find it: FLUX.3 Klein group → FLUX Klein Prompt List, in every FLUX 3 Klein preset.
Three imported .txt files plus the main prompt gave four images in 8 seconds, from click to finish. A batch takes up to 63 extra prompts.
NEW — MODEL DOWNLOADER: FLUX 3 KLEIN CORE BUNDLE (10.85 GB)
One click downloads everything the three presets use:
FLUX-3-Image-Klein-INT8-ConvRot-HQ.safetensors (5.66 GB): our HQ INT8 ConvRot version of the distilled FLUX 3 Klein, with selected layers kept in BF16.
Qwen3-VL 4B FP8 text encoder (4.88 GB), saved as qwen3vl_4b.safetensors, the file name SwarmUI itself uses for FLUX 3 Klein.
FLUX 2 VAE (0.31 GB), saved as VAE/Flux/flux2-vae.safetensors, the same file FLUX 2 Klein uses: if you already have it, it is verified and skipped.
Where to find it: run Windows_Start_Download_Models_App.bat. The FLUX 3 Klein Core Bundle is the first bundle of the SwarmUI Bundles tab.
Optional, under Image Generation Models → FLUX 3 Klein Models: the original distilled BF16 model (10.34 GB), the undistilled Base BF16 checkpoint for fine-tuning and LoRA training (10.32 GB; accept its license on Black Forest Labs' gated Hugging Face page and enter your Hugging Face token in the downloader; it uses its own 50-step, guidance 3.5 settings), and a Qwen3-VL 4B BF16 text encoder (8.27 GB). The BF16 encoder has the same qwen3vl_4b.safetensors name as the FP8 one, so download only one of them.
Why HQ INT8: in our 40-state test against the original BF16 model (BF16 text encoder, RTX 3090), the denoiser output of our HQ INT8 ConvRot model differed by 2.14% on average, against 3.05% for the official INT8 ConvRot file with its default settings and 2.27% with matched attention; with matched attention, one held-out edit case favored the official file (3.84% against 3.58%; 4.73% with its default settings). The transformer peaked at about 6 GiB of VRAM instead of 10.5 GiB and ran each step about twice as fast as BF16. These are numerical-fidelity measurements, not image-quality percentages.
UPDATED — TEXT ENCODERS GO WHERE YOUR SWARMUI LOOKS FOR THEM
SwarmUI keeps text encoders in the folders of its SDClipFolder setting: text_encoders;clip on installs from late November 2025 on, and only clip on older installs, because SwarmUI never changes a setting it has saved. The downloader now reads this setting from your SwarmUI (Data/Settings.fds next to the Models folder you download into):
A text encoder you already have in one of those folders is verified where it is, instead of being downloaded a second time.
A new text encoder goes into the first folder, where SwarmUI saves its own downloads too.
If the downloader cannot find your SwarmUI settings, it uses clip, which SwarmUI scans by default on every install.
The ComfyUI and Forge folders are unchanged. We tested this on new, old and custom SwarmUI folder settings.
NEW — FLUX 3 KLEIN PROMPT GUIDE FOR LLMS
New file: Automatic_Prompt_Generate_Enhance_To_Video_And_Image_Models_LLM_Instructions/FLUX_3_Klein_Enchance_Prompt_Feed_For_LLMs.txt. Give it to ChatGPT, Claude, Gemini or a local LLM together with your idea, and it writes FLUX 3 Klein prompts: text-to-image captions, edits that say what stays and what changes, multi-reference prompts with @image1 and @image2, exact text for posters and signs, and layout boxes. Its first part explains the SwarmUI presets (modes, @image numbering, the box editor, prompt syntax and batches); the rest is the same research-based guide, version 1.5, as in our ComfyUI installer, with original examples.
UPDATED — BUNDLE REPORT, PRESET BACKUPS AND DOWNLOADER APP
Which_Bundles_Downloads_Which_Preset_Models.html is regenerated for V80: all 68 presets, and the three FLUX 3 Klein presets are fully covered by the FLUX 3 Klein Core Bundle. The report generator now also knows SwarmUI's default FLUX 3 Klein text encoder and VAE.
Preset backups keep their thumbnails: Windows_Preset_Delete_Import.bat now saves every preset's thumbnail inside the backup file, so a backup holds everything needed to bring your own presets back with their images.
The downloader app shows v182 and uses Gradio 6.30.0. Windows_Start_Download_Models_App.bat updates Gradio automatically, and the RunPod / SimplePod and Massed Compute instructions install it.
TESTED — REAL GENERATIONS IN CHROME
The extension and the presets were tested on 15 October with SwarmUI 967f005c and ComfyUI 20b46065 on an RTX 5090 in a real Chrome: text to image, scene-preserving edits, two-reference selection, visual layout (drawing, moving and resizing boxes), advanced prompt syntax, prompt batches, .txt file and folder import, Stop after current, and 23 workflow checks (reference numbering, the 10-image limit, duplicate images, layout tables, reference sizing and model gating). Every generated result shown in this post is a real run on the HQ INT8 model. The optional BF16 models and FLUX 3 Klein LoRAs were not generated separately.
CUSTOM NODE PACKS WE DEVELOP OR MAINTAIN UPDATED
SECoursesAudioTools
LoRA and DoRA files trained with SECourses Musubi Trainer now load completely. ComfyUI silently skipped some of their layers: the MLP halves of Qwen Image 2.1, the q, k and v parts of Z-Image's fused attention, and every HiDream-O1 layer. The node pack now maps the trainer's layer names to them, DoRA scales included. SwarmUI gets this too: Windows_Install_or_Update_SwarmUI.bat installs this node pack for SwarmUI's self-starting ComfyUI backend.
New FLUX.3 Gallery Prompt and FLUX.3 Optional Gallery References nodes connect our reference gallery to FLUX 3 Klein in our ComfyUI presets: @image1 names, seeded {red|blue} choices and up to 10 references with their own size.
SWARMUI PREMIUM EXTENSIONS UPDATED
FluxThreeKlein 1.0.0 (new): FLUX 3 Klein advanced prompting, image references, visual layout boxes and prompt batches on SwarmUI's native FLUX 3 workflow. Windows_Install_or_Update_SwarmUI.bat installs it.
HOW TO UPDATE
Download SwarmUI_Model_Downloader_v182.zip from this post and extract it over your existing downloader folder.
Run Windows_Install_or_Update_SwarmUI.bat. It updates SwarmUI to the version with native FLUX 3 Klein support (added on 15 October), installs the new FluxThreeKlein premium extension and updates SECoursesAudioTools.
Update the ComfyUI backend: FLUX 3 Klein needs ComfyUI from 15 October or newer (tested with 20b46065). If SwarmUI uses our ComfyUI installation, update it with our ComfyUI installer V141 (Windows_Install_Or_Update_ComfyUI.bat). If SwarmUI uses its own ComfyUI, open Server → Backends, edit the ComfyUI Self-Starting backend and set AutoUpdate to Always Update for one start. You can switch it back afterwards.
Restart SwarmUI and its ComfyUI backend so the new extension and nodes register.
With SwarmUI running, run Windows_Preset_Delete_Import.bat to import Amazing_SwarmUI_Presets_v80.json. It backs up your current presets, thumbnails included, to utilities\presets_backups, deletes them all (your own presets too) and imports the new pack.
Run Windows_Start_Download_Models_App.bat and download the FLUX 3 Klein Core Bundle.
ComfyUI Installer and Presets: https://www.patreon.com/posts/105023709
10 October 2026 V180
This is a MiniMax H3 RefMods, Long Lip Sync and presets update. V180 adds MiniMax H3 RefMods (save a character, a scene or a voice once and reuse it in every H3 preset), the new RefMod Builder preset, two MiniMax H3 Long Lip Sync presets (one complete soundtrack → one continuous video), new artwork for every preset, MiniMax H3 Optimizations on ComfyUI's native SOL attention, model downloader bundles that now hold every model their presets use, and the matching updates of our SwarmUI premium extensions and custom nodes.
Amazing SwarmUI Presets moves to V79 with 65 presets: 3 new presets, the 4 Wan 2.1 presets removed, and all 62 other presets updated. No new model downloads are needed for the SwarmUI presets. The new presets use the models of the existing MiniMax-H3 Core Bundle.
All 65 presets as they appear in the SwarmUI Presets tab, with their new thumbnails:
Every preset now has its own artwork: a different animal or mythical creature for each preset, with the model family or task and the mode or step count on the thumbnail, so you can find the right preset at a glance.
Exact V178 → V179 zip audit:
7 changed files: Amazing_SwarmUI_Presets_v77.json was replaced by Amazing_SwarmUI_Presets_v79.json, the downloader app shows the new version, update_swarmui.py now moves SwarmUI's ComfyUI-GGUF and ComfyUI-LTXVideo node packs to our fixed forks, and the model catalog, its size data, the bundle report generator and the regenerated Which_Bundles_Downloads_Which_Preset_Models.html carry the bundle updates described below.
The other 35 files are identical.
Everything else for SwarmUI arrives through Windows_Install_or_Update_SwarmUI.bat: our premium extensions, FoleyExtension, SECoursesAudioTools, ComfyUI-TeaCache and the ComfyUI-GGUF and ComfyUI-LTXVideo fork switch. Our ComfyUI installer V139 brings the same fork fixes plus ComfyUI-WanVideoWrapper.
NEW — MINIMAX H3 REFMODS: SAVE A CHARACTER, SCENE OR VOICE ONCE, REUSE IT IN EVERY H3 PRESET
A RefMod is a small .safetensors file that stores ready-encoded MiniMax H3 references: photos, a video with its soundtrack, audio, or a mix. Build it once, then cite it in any MiniMax H3 prompt as @refmod1, @refmod2 and so on. You no longer attach the same media and VAE-encode it again for every generation.
Works in 11 MiniMax H3 presets: text to image, text to video, image to video, first and last frame, references to video, audio only, and both new Long Lip Sync presets.
No attachment needed: RefMods alone are enough for MiniMax H3 References, and normal prompt attachments still work next to them.
Every row has its own controls: enable, strength from 0 to 1 (how much reference detail is kept) and a component filter: all, visual or audio. Use only the voice of a video bundle, or only its look.
The right model is picked for you: active RefMods switch an FL2VA model to the matching Ref2VA model, and the FL2V Turbo 4-step and 8-step LoRAs to their Ref2V versions.
No fixed number of files: add as many rows as your VRAM and context allow. The picker shows the extra reference tokens.
Fully optional: empty, disabled and zero-strength rows change nothing, and RefMods are off in every preset until you turn them on.
Compatible: reads MiniMaxH3Mod v4 and legacy single-reference files, v5 bundles and Fantastic files, and new bundles open in those projects too. Files live in the ComfyUI backend's models/refmods folder (SwarmUI's Models/refmods is found as well); Import files uploads them from your computer and never overwrites an existing file.
Where to find it: apply any MiniMax H3 preset, open the MiniMax H3 RefMods group (between Resolution and Sampling) and tick MiniMax H3 RefMods.
Each row picks a file, its strength and which parts of the bundle to use:
Here @refmod1 is a character-and-voice bundle at full strength, and @refmod2 adds only the voice of a video bundle at strength 0.8. Validated on an RTX 5090: all 11 RefMods-enabled MiniMax H3 presets generated in SwarmUI with active RefMods, and in our parity check, disabled and zero-strength rows built exactly the same SwarmUI workflow as no RefMods. RefMods add reference conditioning; they do not guarantee an exact identity or voice clone.
NEW PRESET — MINIMAX H3 REFMOD BUILDER: BUILD YOUR OWN REFMODS INSIDE SWARMUI
"MiniMax H3 RefMod Builder - Images Video Audio - 261009"
Apply the preset first (it clears the prompt, Init Image and Init Audio), then attach photos of one character, a video, audio or a mix above the prompt, set the RefMod name and press Generate. Only the H3 video and audio VAEs run: no diffusion sampling, no text encoder and no training. On an RTX 5090 a bundle took a few seconds in our tests (4.6 seconds of generation time in SwarmUI). The file is saved in models/refmods and a source preview returns to SwarmUI; press Refresh in the RefMod picker to use it right away.
One character stacks several photos into one visual reference; separate references keeps them as numbered pictures.
Videos keep their soundtracks, so one bundle can carry both a look and a voice.
Resolution 512 and compression none are the defaults. Compression 32, 16 or 8 pools the encoded references spatially to save tokens.
Existing files are never overwritten: new files get numbered names. One bundle can hold up to 256 references.
Where to find it: Presets tab → MiniMax H3 RefMod Builder. Its settings are in the MiniMax H3 RefMods group.
The builder settings in detail:
Here three photos of the same presenter become one "my_presenter" character RefMod.
NEW PRESETS — MINIMAX H3 LONG LIP SYNC: ONE COMPLETE SOUNDTRACK → ONE CONTINUOUS VIDEO
"MiniMax H3 Long Lip Sync - References - Turbo 4 - 261008"
"MiniMax H3 Long Lip Sync - First Frame - Turbo 4 - 261008"
Put your complete narration or song in Init Audio, add a portrait and press Generate. The video follows the whole recording automatically: there is no manual duration, and every word and pause keeps its original timing. The result is one 24 FPS H.264/AAC MP4 with your soundtrack.
Native latent continuation: the video is generated in overlapping native H3 windows that follow your soundtrack (default 243 frames, about 10.1 seconds) with a 39-frame overlap (about 1.6 seconds). Each window continues from the previous window's clean latents instead of re-encoding decoded pixels.
Two modes: References attaches the portrait beside the prompt as @image1 (Ref2VA); First Frame puts it in Init Image (FL2VA).
Tuned defaults: 832×1248, Turbo 4 steps and MiniMax H3 Low VRAM on, with MiniMax H3 Optimizations and First Block Cache off. Long mode always samples with Euler/simple at CFG 1, so the Sampler, Scheduler and CFG Scale settings are not used.
Video Face Inpainting must be off in long mode; SwarmUI shows a message if it is on.
Adjustable: H3 Long Window Frames 141, 192, 243, 294 or 345 and H3 Long Overlap Frames 39 or 90. Smaller windows need less memory.
The prompt token meter shows the estimate for one window, so a long soundtrack stays within budget.
RefMods work here too.
Where to find it: Presets tab → a Long Lip Sync preset. The new controls are in the Init Audio group.
The new Init Audio controls, here with a 2-minute narration:
Measured in SwarmUI end to end on an RTX 5090: a 30-second 512×768 First Frame video took 125.81 seconds, and a 60-second 832×1248 References video took 753.98 seconds (with MiniMax H3 Optimizations on), both covering the full audio. Long mode is experimental, so review every result: joins can show short soft frames or motion changes, and very long high-resolution clips can drift in framing.
UPDATED — THE 9 EXISTING MINIMAX H3 PRESETS ARE REFMODS-READY
The 9 existing MiniMax H3 presets now include the optional RefMods settings, explicitly switch Long Lip Sync, RefMods and the RefMod Builder off, and set MiniMax H3 Video Shift to 12, the value they already used. Applying one of them after a Long Lip Sync or RefMod Builder preset therefore turns those features off and restores Video Shift 12 (the First Frame preset uses 6). Their descriptions explain how to use @refmod1, @refmod2.
Updated: Text to Image with Optional References, References to Video, Text to Video, Beginning and Ending Image/Frame to Video, Image to Video, Audio Only with Optional References, Text to Video 8 Steps, Beginning and Ending Image/Frame to Video 8 Steps, and References to Video 8 Steps.
New: Long Lip Sync - References, Long Lip Sync - First Frame and the RefMod Builder.
REMOVED — THE 4 WAN 2.1 PRESETS
Removed from Amazing SwarmUI Presets: Wan 21 Text To Video Base Model, Wan 21 FusionX Text to Video, Wan 21 Image To Video Base Model and Wan 21 LightX2V LoRA Image to Video. Windows_Preset_Delete_Import.bat removes them when it imports V79.
The Wan 2.1 models and the Wan 2.1 Core Models Bundle stay in the model downloader, and the Wan 2.2 presets are unchanged.
UPDATED — NEW ARTWORK FOR EVERY PRESET AND TWO PRESET FIXES
New thumbnails for all 65 presets, shown in the first image of this post.
FLUX 2 - 8 Steps Fast Realism - Low RAM - 251230: Flux Guidance Scale 7 → 1. It now matches the standard Fast Realism preset; only the GGUF model and the GGUF text encoder differ.
FLUX Kontext Dev Edit Images: the description now explains how to use it: put the image in Init Image (or paste it into the prompt) and describe the change. SwarmUI detects the official Kontext file as plain Flux.1 Dev, which ignores your image, so the description also gives the one-time fix: the model's ☰ menu → Edit Metadata → Architecture: Flux.1 Kontext Dev → Save.
UPDATED — MINIMAX H3 OPTIMIZATIONS NOW USE COMFYUI'S NATIVE SOL ATTENTION
The MiniMax H3 speed option is now called MiniMax H3 Optimizations (it was MiniMax H3 4x Speed). Its parameter ID did not change, so saved presets and API calls keep working. Through our ComfyUI-TeaCache fork it now uses ComfyUI's own native H3 sparse attention (SOL), and SOL and caching have separate switches:
MiniMax H3 Speed Sparse Attention (SOL): auto benchmarks the complete attention path on your GPU and keeps SOL only when it is at least 5% faster; enabled skips the benchmark; disabled keeps normal attention.
MiniMax H3 First Block Cache (new switch): on by default, as before. Untick it to run SOL alone.
First Block Cache now works together with other native attention patches, such as SLA, instead of replacing them.
Where to find it: type MiniMax H3 into Filter parameters, tick Display Advanced Options? and tick MiniMax H3 Optimizations in Core Parameters. First Block Cache then appears in Advanced Sampling next to Speed Sparse Attention; both apply only while Optimizations is on.
The same controls in the parameter panel, numbered like the steps above:
Measured on an RTX 5090 (FL2VA INT8, Turbo 4 steps, 832×1248, 10-second narration, First Block Cache off): 110.9 s with dense attention and 99.2 s with the default protected SOL schedule (1.12×); at 512×768 the gain was 1.06×. In a matched run, the old bundled integration took 207.1 s against 116.0 s dense, so this update removes that slowdown. SOL and caching are approximations, not lossless.
FIXED — INIT AUDIO MATCH DURATION IN SWARMUI'S OWN MINIMAX H3 GRAPHS
With Init Audio and Init Audio Match Duration on (the default), the video length now follows your audio in text to video, image to video with a MiniMax H3 Video Model, and native reference generations. These graphs used to keep the Text2Video Frames or Video Frames length, so longer audio was cut and shorter audio was padded with silence. Now, for example, 2.5 to 3 seconds of audio gives 73 frames (3.04 seconds), tested at 22.05, 44.1 and 48 kHz.
FIXED — OPTIONS OF DISABLED FEATURES ARE NO LONGER SENT WITH EVERY GENERATION
The option groups of our extensions (AvatarForever, Qwen Unified, SVI Pro, LTX 2.3 Foley, LTX 2.5 Audio To Video, LTX 2.3 Licon MSR and the MiniMax H3 groups, including Video Face Inpainting) stayed visible and were sent with every generation, even while their feature was off. In one test, a plain Z-Image Turbo picture carried 85 unused parameters, and SwarmUI SHA-256 hashed 11 unrelated model files (47.1 GB) into its metadata. Now these options stay hidden while their feature is off and are no longer sent: the same picture has 0 unused parameters and is pixel-identical. With a feature on, its values are sent exactly as before.
FIXED — MINIMAX H3 AUDIO NODES WORK WITHOUT TORCHAUDIO
ComfyUI removed torchaudio on 21 September, and new Linux SwarmUI installs no longer have it. Our MiniMax H3 Init Audio node (FoleyExtension) and the H3 Long Lip Sync and H3 Streaming nodes (SECoursesAudioTools) now resample with ComfyUI's own resampler, so any sample rate works on a torchaudio-free install. The new resampling was bit-identical in all 88 comparisons, and a real MiniMax H3 Turbo 4 run with a 48 kHz soundtrack gave 120 of 120 identical frames and bit-identical audio.
UPDATED — MODEL DOWNLOADER: EVERY CORE BUNDLE NOW DOWNLOADS ALL MODELS ITS PRESETS USE
We checked all 65 SwarmUI presets and all 90 presets of our ComfyUI installer against the downloader: every model file a preset names, including bypassed nodes and optional switches, and every model our extensions and nodes load by default, is now downloaded by that preset's own core bundle. Files that a node downloads by itself, such as the RIFE interpolation model, are not part of the bundles.
MiniMax-H3 Core Bundle (67.55 → 69.91 GB) and MiniMax-H3 Low VRAM Bundle (51.85 → 54.21 GB): add codeformer.pth and the InsightFace det_10g.onnx detector for the CodeFormer mouth pass of the ComfyUI MiniMax H3 Lip Synch preset, the InsightFace recognition model (w600k_r50.onnx) that MiniMax H3 face refinement uses to tell faces apart and the landmark model (2d106det.onnx) of its landmark-align node, and the LightX2V Turbo-SLA 4-step LoRA that the ComfyUI H3 lip sync presets load when SLA sparse attention is on.
LTX 2.3 Avatar Forever Core Bundle (44.34 → 44.50 GB): adds w600k_r50.onnx and 2d106det.onnx next to det_10g.onnx. insightface downloads its buffalo_l pack only when that folder does not exist yet, so a folder with det_10g.onnx alone left the face refinement without its recognition and landmark models. The updated face refinement also repairs such a folder itself.
LTX 2.5 Video Core Bundle (64.64 → 64.82 GB): adds the same three InsightFace files, which the LTX 2.5 Audio To Video presets use for face detection (without them they fall back to a simpler detector).
New SDXL Core Bundle (6.46 GB): RealVisXL V5, the model of the ComfyUI SDXL preset, which no bundle downloaded before.
Which_Bundles_Downloads_Which_Preset_Models.html is regenerated for V79. The report now reads the model settings of our AvatarForever and Qwen Image 2.1 extensions and SwarmUI's default MiniMax H3 and LTX 2.5 text encoders and VAEs, so every preset shows its own core bundle with full coverage. Run Windows_Update_And_Open_Model_Report.bat to check your own model folders.
The MiniMax-H3 Low VRAM Bundle holds the W4A8 versions of the two main H3 models: with only that bundle, select them in place of the INT8 models the presets name.
HOW TO UPDATE
Download SwarmUI_Model_Downloader_v179.zip from this post and extract it over your existing downloader folder.
Run Windows_Install_or_Update_SwarmUI.bat. It updates SwarmUI, our premium extensions (MiniMax H3 References 1.19.1, AvatarForever 1.0.2), FoleyExtension, SECoursesAudioTools and our ComfyUI-TeaCache fork, and it moves an existing ComfyUI-GGUF (SwarmUI's DLNodes) and ComfyUI-LTXVideo (SwarmUI's own ComfyUI) to our fixed forks. If you install GGUF support in SwarmUI later, run this .bat once more.
Update the ComfyUI backend: native SOL needs comfy-kitchen 0.2.37 or newer, and Long Lip Sync was tested with ComfyUI 52f98af2 (v0.39.0 plus 24 commits, 8 October). If SwarmUI uses our ComfyUI installation, update it with our ComfyUI installer V139: Windows_Install_Or_Update_ComfyUI.bat, plus Windows_Custom_Nodes_Bundles_Installer.bat options 5 and 17 for ComfyUI-LTXVideo and ComfyUI-WanVideoWrapper. It updates ComfyUI, comfy-kitchen and our forks together; updating ComfyUI alone breaks the GGUF and LTX 2.3 Foley presets. If SwarmUI uses its own ComfyUI, open Server → Backends, edit the ComfyUI Self-Starting backend and set AutoUpdate to Always Update for one start (step 2 already moved its GGUF and LTXVideo node packs to our forks). You can switch it back afterwards.
Restart SwarmUI and its ComfyUI backend so the new nodes and parameters register.
With SwarmUI running, run Windows_Preset_Delete_Import.bat to import Amazing_SwarmUI_Presets_v79.json. It backs up your current presets to utilities\presets_backups, deletes them all (your own presets too) and imports the new pack.
Run Windows_Start_Download_Models_App.bat and download the MiniMax-H3 Core Bundle (or the Low VRAM Bundle). If you already have it, run it again: files that are already there and verified are skipped, so only the new files download.
ComfyUI Installer and Presets: https://www.patreon.com/posts/105023709
22 September 2026 V178
New Qwen Image 2.1 presets added into Amazing_SwarmUI_Presets_v77.json file
I have coded necessary extensions so that all features of the Qwen Image 2.1 is fully supported not like default SwarmUI installation
As usual run Install Update bat file to get necessary extensions automatically installed, import json file and download Qwen Image 2.1 core bundle
LLM instructions folder renamed into Automatic_Prompt_Generate_Enhance_To_Video_And_Image_Models_LLM_Instructions
New instructions Qwen_Image_2.1_Enchance_Prompt_Feed_For_LLMs.txt added to the folder
All of Our SwarmUI Presets with Core Bundles To Download - 20 September 2026
20 September 2026 V177
AvatarForever Long Talking Videos, Stable Video Infinity 2.0 Pro, Qwen Image 2.1 and 481 Models
All MiniMax H3 presets now uses Int8 ConvRot VAE (almost same quality and lower VRAM tested)
1) NEW: AvatarForever — image + audio or text + audio to a long avatar video
AvatarForever is now available directly inside SwarmUI. Attach speech or singing audio to the prompt, optionally add a portrait as Init Image, describe the performance, and generate a continuous H.264 MP4 with the supplied soundtrack.
Use our LLM instructions to write your prompts
What the unified AvatarForever workflow does
Image + audio or text + audio. Keep Use Image on for a portrait-driven video, or turn it off and describe the person entirely in the prompt.
The duration follows your recording. Duration Seconds = 0 uses the full audio. You can also select a starting point and cap the run at 10, 30, 60 or 300 seconds. Short audio is never looped or stretched.
Your soundtrack remains your soundtrack. It is re-encoded to AAC for the MP4; the workflow does not clone, regenerate or replace the voice.
Rolling long-video generation. The released four-step AvatarForever schedule generates chunks with bounded history, a retained first-chunk identity sink, relative positions and optional ForeverCache. Long clips do not require one enormous latent tensor.
Portrait-aware resolution. With Match Image Aspect enabled, Width × Height is treated as a pixel budget and fitted to the portrait on a 32-pixel grid. Turn it off for an exact canvas.
Optional mouth enhancement. CodeFormer restores only an aligned, feathered mouth region before the first video export. It defaults to fidelity 0.9 and blend 0.7. Turn it off to preserve the original decoded frames.
Normal Swarm controls still apply where they should: Main Model, Prompt, Seed, Width, Height, Video FPS and compatible LoRAs. Main Steps, CFG, negative prompt, refiner and unrelated video controls are bypassed by this dedicated sampler.
Download AvatarForever bundle to make it use right away
2) NEW: Stable Video Infinity 2.0 Pro — continuous Wan 2.2 long video
The new StableVideoInfinity premium extension and SESVIProVideo node add the official SVI 2.0 Pro cross-clip conditioning method to SwarmUI and ComfyUI using the native Wan 2.2 models and samplers.
Start from an image and generate several connected clips. Write one prompt per paragraph; blank lines advance the prompt schedule, and the final paragraph repeats if more clips are requested.
Default duration: 3 clips produce 233 frames, or 14.5625 seconds at 16 FPS. Increase SVI Clip Count for a longer result.
Fast mode: 4 steps, CFG 1 with the existing Seko high/low fast LoRAs. Quality mode: turn Fast Mode off and use 30 steps, CFG 4, Euler/simple, shift 5 and switch sigma 0.875.
Continue a video. Enable SVI Continue Video, select a source, and optionally append that source to the output. A separate Init Image can override the identity anchor.
Correct high/low switching. The two Wan models change at the actual sigma boundary. Official SVI PEFT LoRA keys are translated in memory for ComfyUI's native loader; files on disk are never converted or overwritten.
Bounded decoding. Only one decoded clip is kept in memory and the output is encoded incrementally. Tiled VAE is available to reduce decode VRAM.
The new preset is Stable Video Infinity 2 Pro - Unified Long Video - Wan 22 - 260919. The two official SVI Pro LoRAs, 1.14 GB each, were added to the existing Wan 2.2 Core 4 Steps Bundle. That bundle moves from 23 models / 104.16 GB to 25 models / 106.45 GB.
SVI 2.0 Pro produces silent video. Input audio is not copied. SVI-Talk, SVI-Dance and EverAnimate are separate models and are not switches in this workflow.
Each line becomes each section prompt. Use Stable_Video_Infinity_2_Pro_Enchance_Prompt_Feed_For_LLMs.txt for writing your prompt
Download Wan 2.2 Core Bundle to have all models
3) NEW: Qwen Image 2.1 models plus a unified ComfyUI workflow
The downloader now contains the complete Qwen Image 2.1 model family from Comfy-Org and a practical one-click bundle.
Qwen Image 2.1 Core Bundle — 16.10 GB / 3 files: INT8 ConvRot diffusion model (6.76 GB), INT8 ConvRot Qwen3-VL 8B text encoder (8.71 GB) and the Qwen Image 2.1 BF16 VAE (0.63 GB).
SwarmUI not added yet, as soon as adds, hopefully i will update zip file
6) Critical MiniMax H3 Speed Optimizer workflow fix
Newer ComfyUI frontend 1.53+ changed its save path. The old display-reordering script could save widgets_values in display order and load them in backend order, rotating one value on every reload, tab switch or undo. In a badly scrambled workflow, the string gpu could land in a numeric percentage field, silently turn the speedup on and fail the first folder-batch item plus every item after it.
Download Qwen Image 2.1 Core Bundle to have all the models
The updated ComfyUI-TeaCache fork now:
Translates values through the frontend's universal serialize/configure hooks instead of reordering the node's widgets
Recognizes backend order, named values and display order when loading older workflows
Validates numeric ranges and choices inside the backend node, including saved API prompts that bypass the frontend
Falls back to the normal path with a warning instead of killing the batch
Resets values only when no valid layout can be recovered, with the four-times speedup safely off
This was verified on real ComfyUI frontend 1.51.9 and 1.53.6, with both canvas and Vue nodes and the shipped presets.
How to update
Download SwarmUI_Model_Downloader_v176.zip from this post and extract it over your existing downloader folder.
Run Windows_Install_or_Update_SwarmUI.bat. This updates SwarmUI, our ComfyUI-TeaCache fork and the premium extension pack, including AvatarForever and StableVideoInfinity.
Update your ComfyUI backed - you have to use our ComfyUI installer
Restart SwarmUI and its ComfyUI backend so the new nodes and extension parameters are registered.
Run Windows_Start_Download_Models_App.bat. Choose the LTX 2.3 Avatar Forever Core Bundle, Qwen Image 2.1 Core Bundle, or the updated Wan 2.2 Core 4 Steps Bundle according to the workflow you want.
Run Windows_Preset_Delete_Import.bat to import Amazing_SwarmUI_Presets.json.
Thank you all for the support that makes this work possible.
16 September 2026 V175
Currently recommended backend args are like below to not have any VRAM issues
--disable-pinned-memory --use-sage-attention --enable-triton-backend --disable-comfy-compiler
New preset LTX25 Audio To Video - Optional Image - Core 2-Stage 8+3
The preset is ultra fast with 8 steps + 3 steps upscale and working amazing
Higher resolution keep identiy and speaking animation much better
This preset working amazing to generate fully lip snyched + animated avatar videos
You can see what kind of animation it generates in our latest video : https://youtu.be/KgUhPAOjagI
Use Windows_Preset_Delete_Import.bat when SwarmUI running or manually import via latst Amazing_SwarmUI_Presets.json file coming inside zip
This is a custom programmed preset and there isn't anything close to this preset online
With special programing it improves Identity anchors fight drift on long clips.
With an image, it re-injects that image as an LTX keyframe guide every Anchor Every Seconds (4 s) at Anchor Strength (0.4), mid-clip only, snapped to the quietest nearby moment so an anchor never lands on a spoken syllable — and with the mouth and jaw region excluded so lip motion stays free.
Mouth detection uses insightface when it is installed, otherwise the Haar cascade bundled with the extension.
End Anchor = full ends the clip back on your input pose.
To use this preset with image + audio
Use init image + add prompt audio
I improved the prompt bar significantly for this preset and future models
It works with all models right now not specific to LTX 2.5
Now you can upload image or video or audio much better into the prompt bar
See below image how it works
MiniMax H3 reference cards also got ✂ Trim: for a video the full data is kept and only the backend trim window changes (the popup preselects your current window; select the full range to clear it), audio references are re-sliced in the browser.
Bug fixes in the same round: MiniMax H3 image cards are now built through the core's own imagePromptAddImageData, so the title updater no longer throws on header-less cards; clicking a card header no longer inserts a token by accident; and the core's clear button keeps its own label while a non-H3 model is selected.
MiniMax H3 Audio Shift and MiniMax H3 Video Shift
Someone requested MiniMax H3 Audio Shift and MiniMax H3 Video Shift since some custom community MiniMax H3 LoRAs requiring custom values
I have added it into our custom extension
So just run Windows_Install_or_Update_SwarmUI.bat to get it
It is located under Advanced Sampling
MiniMax H3 768p Turbo LoRAs and the TaoMate streaming LoRA (added in V173/V174)
LightX2V released newer 768p-trained Turbo LoRAs that actually match the 1344x768 default our presets use, and Alibaba TaoLive released the TaoMate streaming generator. All four are one click away in the downloader:
MiniMax H3 FL2V Turbo 8-Step LoRA v1.0 768p (ComfyUI BF16) — 1.82 GB
MiniMax H3 FL2V Turbo 4-Step LoRA v1.2 768p (ComfyUI BF16) — 1.82 GB
MiniMax H3 Ref2V Turbo 8-Step LoRA v1.0 768p (ComfyUI BF16) — 1.82 GB
MiniMax H3 TaoMate 3-Step Streaming LoRA (Full-Rank ComfyUI BF16) — 1.16 GB
Presets are also updated to use these.
TaoMate only available in ComfyUI and its quality is lower than 4 steps LightX2V LoRA imo
But it can do true stream of long video like can generate 1 minute 2 minute 3 minute etc videos i tested
The model report finally finds ALL of your models (added in V174)
This was the most requested fix. If you kept models outside the downloader's base folder, the report used to call them "missing". Not anymore.
Windows_Update_And_Open_Model_Report.bat now:
Reads SwarmUI's own Data\Settings.fds and scans every Paths.ModelRoot entry you have configured, plus any absolute SD*Folder overrides. SwarmUI is auto-detected next to the downloader or above your model folder. A full, dependency-free FDS parser was written for this.
Asks you for extra folders before running. Type one or several (C:\ws.data\Models;D:\More_Models), paste a path with quotes or a trailing backslash, or point it at a whole SwarmUI install folder. Input is cleaned up and validated, and missing folders are reported instead of silently ignored.
Prints a real console summary: preset counts (complete / partial / missing), unique models present vs missing, best-effort missing download size, file counts per scanned folder, and the 15 largest missing models.
Keeps the window open until you press Enter, so you can actually read that summary.
De-duplicates overlapping roots, so a SwarmUI sub-folder nested inside your base path no longer counts the same file twice.
The HTML report lists every scan root and where it came from (yours / SwarmUI settings / in-repo), including which SwarmUI settings file was read.
Power users get new CLI flags too: --extra-path, --swarmui-dir, --no-swarmui-settings, plus the MODEL_REPORT_EXTRA_PATHS environment variable.
PROMPT FEED v3.1 (MiniMax H3)
MiniMax_H3_Enchance_Prompt_Feed_For_LLMs.txt updated and grew by roughly 470 lines. Paste it into your LLM and it writes correct H3 prompts for you. New in this edition:
A full new chapter on directing a PERFORMANCE: a cause-and-visible-change method (starting state → trigger → eye and face response → body and voice response → settle) with a table of concrete emotional turns and the audible behaviour that matches each one.
A bounded inventory of 25 community dialogue cues (<pause>, <breath>, <inhale>, <i>word</i>, <whisper>, <sighs>, <laughs>, <gasp>, <stutter>, <softer> and more) with strict rules for using them sparingly — and an explicit ban on ever adding them to an exact script or a locked soundtrack.
Correct handling of the locked-soundtrack path: prompt tags CANNOT insert a pause or a sigh into audio you supplied, so the guide now tells the LLM to choreograph the face around your actual recording instead.
Rules for the multi-frame continuation context (1, 5, 22, 39 or 56 previous frames), including budgeting words around the overlap that gets removed on merge.
Reference label normalization notes and a hard rule against inventing H3 fields like "emotion:" or <sadness=0.8>.
The updater itself
Update bat file now validates the newly bundled extension files (the SwarmLtx25AudioToVideo nodes and face cascade, the shared secourses_prompt_media assets) before installing, so a half-finished download cannot leave you with a broken extension folder.
ComfyUI-TeaCache (our fork, installed by the SwarmUI updater) — 4 updates
Fixed MiniMax H3 cache allocations with the AIMDO compiler — no more rogue allocations or AIMDO errors with allocation compilation enabled.
Major SOL attention upgrade. sparse_backend=auto now verifies and benchmarks native Comfy Kitchen INT8 SOL against the bundled kernel and picks the fastest one that passes the numerical gate. Adds sparse_extra_tokens (default 256) and sparse_dense_last_steps (default 1, protecting the final detail step). Text, reference and audio prefix KV blocks stay fully routed and their query outputs are recomputed in full precision.
Measured on an RTX 5090 with real H3 tensors (37,741 tokens, 56 heads, 1344x768 / 124 frames): 2.10x–2.63x faster attention with 22–37% lower sampled error than the previously bundled SOL path. Existing preset JSON needs no rewiring — the new widgets are appended after the originals. Requires comfy-kitchen>=0.2.33, which our installers already target.
Honest note: that is an attention-path measurement, not an end-to-end generation speed claim or a decoded video-quality rating.
Low VRAM + SOL no longer crashes. ComfyUI's H3 block-replace wrapper started passing an attention= override; the low-VRAM block forward rejected it, so enabling SOL together with save_vram failed with a TypeError. Fixed.
Merged upstream Evs91/ComfyUI-TeaCache main (LTXV embeddings_connector import).
torch.compile now works on INT8 ConvRot models. The optional torch.compile node routes comfy.ops.linear_input_act for TensorWiseINT8Layout weights through the opaque torch.ops.comfy_kitchen.int8_linear custom op, so Dynamo can trace the fused INT8 ConvRot MLP path instead of failing on the raw data-pointer call. Same kernel, same arguments — only the dispatch path changes.
Fixed: the new fast tiled decode returned the decoder's raw −1..1 output, so frames came out dark and clipped. It now applies the VAE's output transform on the GPU after the tile blend. Verified against VAEDecodeTiled on an 81-frame 768x512 LTX-2.5 latent with spatial and temporal tiling: float32 accumulation is bit-identical, float16 is within one 8-bit level.
How to update or fresh install:
Download latest SwarmUI_Model_Downloader.zip from this post (at the top) and extract it over your existing install and overwrite all older files.
Run Windows_Install_or_Update_SwarmUI.bat - this updates everything to latest.
Run Windows_Start_Download_Models_App.bat and grab the updated MiniMax-H3 Core or Low VRAM bundle (or just the four new LoRAs individually), and the LTX 2.5 Video Core Bundle if you want audio-to-video.
Run Windows_Preset_Delete_Import.bat to import latest Amazing_SwarmUI_Presets.json.
This has to be run while SwarmUI running.
You can also manually import.
Run Windows_Update_And_Open_Model_Report.bat to see exactly what you have and what you are missing - including the models you keep on your other drives.
Also get latest ComfyUI and update it to latest as well : https://www.patreon.com/SECourses/posts/comfyui-auto-2-105023709
28 August 2026 Update V172
Reference tokens are more robust now.
Variants such as @IMAGE1, @ image # 2, and <picture1> are recognized and normalized automatically.
Prompt-label handling and makes token estimates more accurate for multi-frame continuation and Image-to-Video workflows now.
Better Video Continuation
Init Video Continuation 1.3.1 now supports 1, 5, 22, 39, or 56 context frames.
One-frame mode preserves the previous behavior. Multi-frame mode uses MiniMax H3's native clip conditioning for smoother transitions.
Replayed context frames and matching audio are removed automatically during merging, preventing duplicated footage or sound.
Source audio is preserved and generated audio is appended at the continuation boundary.
Start Frame / Init Video now accepts an image or video. Uploaded videos can be continued and optionally merged with the newly generated segment.
Continuation no longer consumes a Ref2VA reference-image slot.
MiniMax H3 Fixes
Fixed text-only Audio Only workflows failing to recognize SwarmUI's H3 latent canvas.
Fixed pasted multiline wildcard text being collapsed.
WD14 Tagger Hardening
All 19 supported model repositories are now allowlisted and pinned to reviewed, immutable Hugging Face revisions.
Downloaded Taggerine inference code must pass SHA-256 verification before execution.
Added stricter validation for model IDs, thresholds, image types and sizes, model directories, and temporary output paths.
Runtime package installation was removed. Missing dependencies now produce clear installation instructions instead of modifying the ComfyUI environment during generation.
Added dedicated security-boundary tests.
Install from this repo manually https://github.com/FurkanGozukara/SwarmUI-WD14Tagger
Automatic Installers / Updaters Reliability
Aggressive Git updates now clean untracked leftovers after repository resets, improving recovery from blocked or inconsistent node updates.
Get latest zip file extract and overwrite all files
Import latest presets
Run Windows_Install_SwarmUI.bat to update latest
Get latest ComfyUI and update it to latest as well
More latent frames will increase vram usage a lot so be careful
This feature added to ComfyUI presets as well read changelogs on ComfyUI
22 August 2026 Update V170
This is a very big update : LTX 2.5 video generation presets + LTX 2.5 Video Core Bundle, new FLUX 2 Klein INT8 / INT4 ConvRot HQ models, MiniMax H3 Init Audio (make the video follow any soundtrack with lipsync), live prompt token meter, Face Inpainting upgrades, a big model downloader UI upgrade and a huge installer / updater overhaul so please read all
LTX 2.5 video generation presets finally added : Amazing_SwarmUI_Presets_v70.json brings 6 new presets : LTX25 Text To Video - Core 2-Stage 8+3, LTX25 Image To Video - Core 2-Stage 8+3, LTX25 First Last Frame - Core 2-Stage 8+3, LTX25 Audio Reference To Video - Core 2-Stage 8+3, LTX25 Text To Video - Dev HQ 30+6 and LTX25 Image To Video - Dev HQ 30+6 (all 260822)
Audio Reference To Video : add a Prompt Audio file and describe the matching visuals - the native LTX 2.5 audio branch makes the video follow your audio (lipsync, timing) - Image To Video and First Last Frame presets : add an Init Image (and a Video End Frame) and describe only the motion, camera and sound instead of re-describing the still
All 6 presets are two-stage : the first stage generates at 960x544, then SwarmUI's Refine / Upscale applies the official LTX 2.5 latent spatial upscaler x2 and refines to model-aligned 1080p (1920x1088) with synchronized audio - 97 frames at 24 FPS (about 4 seconds) and H264 MP4 output by default
Core presets use my LTX 2.5 Distilled INT8 ConvRot Premium model with 8 + 3 steps at CFG 1 and Euler Ancestral so they are fast - Dev HQ presets use my LTX 2.5 Dev INT8 ConvRot Premium model with 30 + 6 steps at CFG 3 and Euler for maximum prompt adherence and detail
Everything is exposed as normal SwarmUI parameters so you can change resolution, duration, steps and the refiner settings as you like - increase Text2Video Frames for longer clips
Import Amazing_SwarmUI_Presets_v70.json or run Windows_Preset_Delete_Import.bat to get them - the Which_Bundles_Downloads_Which_Preset_Models.html report is regenerated for the new presets as well
LTX 2.5 Video Core Bundle added to the model downloader : 1 click downloads everything the LTX 2.5 presets need - 10 files, 64.64 GB : Dev and Distilled INT8 ConvRot Premium transformers, Gemma 4 12B INT8 ConvRot v2 text encoder (SwarmUI's current default), Gemma 4 E2B INT8 ConvRot prompt enhancer, the convolutional and the diffusion video VAE, the audio VAE, the latent spatial upscaler x2, the pixel spatial upscaler x2 IC-LoRA and the motion track control IC-LoRA
This one bundle covers every model referenced by all of our LTX 2.5 presets in both SwarmUI and ComfyUI, so you do not need to hunt for a second bundle to make any of them run
The text encoder and both VAEs are saved with the exact LTX-2 sub folder and file names SwarmUI's own automatic downloader expects, so SwarmUI never downloads a second copy - in ComfyUI mode they go to models/text_encoders/LTX-2 and models/vae/LTX-2
No more gated downloads for LTX 2.5 : I have mirrored the remaining official Lightricks files into my own non-gated repository, this covers the official Dev and Distilled Comfy INT8 ConvRot transformers, the Distilled NVFP4 transformer (for RTX 5000 Blackwell), the pixel spatial upscaler x2 IC-LoRA and the latent spatial upscaler
These official quantizations stay optional alternatives - my own Premium INT8 ConvRot conversions remain the default of the bundles and presets because they are higher quality, and the Gemma 4 E2B INT8 ConvRot prompt enhancer is also listed separately
Model downloader UI upgrade : every model list and every bundle Includes list is now grouped by the folder the files are saved into : Main Models, Text Encoders, VAEs, LoRAs, ControlNets, Upscalers, Vision, Detection, LLM - biggest files first, and every group header shows the file count and total GB so you instantly see what goes where
New Active Transfers panel with a real progress bar per file (percent, downloaded / total, speed, ETA) above the Activity Log - the log is chronological now with a Follow newest log entry checkbox, it refreshes twice per second, and multiple open browser tabs no longer steal each other's updates
The URL Downloader tab now shows its progress in the same panel instead of looking frozen, and every download reports [STARTING] immediately instead of showing nothing while the file metadata is fetched
Console progress line reordered : percent, downloaded / total size, speed and ETA come first and the long file name goes last, so a narrow CMD window never cuts off the numbers - and when the output is not a real terminal (RunPod / Massed Compute logs) progress is still printed once per second instead of being silent
New LTX2.5_Enchance_Prompt_Feed_For_LLMs.txt inside the Prompt_Generate_LTX_MiniMax_And_Presets_How_To_Use folder : an official-aligned LTX 2.5 prompting guide for LLMs - give it to ChatGPT / Claude / Gemini, tell it what you want, and it writes properly formatted LTX 2.5 prompts (text / image / audio to video, multi-shot and IC-LoRA conventions are all covered)
MiniMaxH3References extension 1.13.2 : videos and audio files attached with SwarmUI's own native prompt uploader are now recognized as MiniMax H3 references too - same cards, same @video1 / @audio1 tokens with autocomplete, same Ref2VA workflow
All extensions and custom nodes automatically installed by our installers
FLUX 2 Klein INT8 and INT4 ConvRot HQ models : I have compiled high quality INT8 ConvRot and INT4 ConvRot versions of FLUX 2 Klein myself - Base 9B, Distilled 9B and Distilled 9B kv - 6 new models added to the model downloader app
FLUX 2 Klein Core Bundle and Complete Image Generation and Editing Bundle now download the INT8 ConvRot HQ versions - almost same quality as BF16 but massively faster, same as our Krea 2 and LTX 2.5 ConvRot models
With Amazing_SwarmUI_Presets_v70.json both FLUX 2 Klein Base and FLUX 2 Klein Distilled 8 Steps presets now use the INT8 ConvRot HQ models
MiniMax H3 FL2V Turbo 4-Step 768p speed LoRA updated to v1.1 : the model downloader now downloads minimax_h3_fl2v_turbo_4step_v1.1_768p_bf16.safetensors instead of the old v1.0 - this is the optional 16:9 aspect ratio speed up LoRA for 768p generations from LightX2V - MiniMax-H3 Core and Low VRAM bundles download the new version automatically
MiniMax H3 Init Audio : new Init Audio parameter group added - select or upload any audio file (or even a video file, its soundtrack is used) and MiniMax H3 generates the video to follow that soundtrack exactly : lipsync, action timing, ambience - and keeps it as the exact final audio track of the output
Works for Text To Video, Image To Video and References To Video presets and Init Audio Match Duration makes the video length automatically follow the audio length
Live token meter : a real time Tokens meter now appears above the prompt and shows estimated MiniMax H3 token usage (prompt + references + resolution + duration) against the 109k budget - you instantly see when your references or duration are too big before generating
Face Inpainting upgrades : new Face Inpaint Faces parameter selects which detected faces get refined : 1 = biggest face (default), 2 = second biggest, 1,3 or all - faces are ranked biggest to smallest and each selected face is refined in its own pass
New hallucination guard prevents pasting a wrong / neighbouring face over your subject and Face Inpaint Detector can no longer block generations
Installer / Updater overhaul : Windows_Install_SwarmUI.bat and Windows_Update_SwarmUI.bat are merged into a single Windows_Install_or_Update_SwarmUI.bat
Windows_Start_SwarmUI.bat is now ultra fast : it starts SwarmUI directly without updating or downloading anything and rebuilds only if the code actually changed
Update runs are much faster : unchanged extensions (Premium Extensions, Licon MSR, Foley) are skipped with a quick revision check instead of being re-downloaded on every run - FFmpeg and Cloudflared downloads are also skipped when already latest
Missing Git is now auto installed (winget on Windows, apt-get on Linux) instead of failing with cryptic WinError 2 errors
SwarmUI is now installed as a full git clone and older shallow installs are auto completed - fixes the scary Tag list empty?! warning of SwarmUI's own update check
.NET 10 SDK now installs into SwarmUI/.dotnet folder, matching SwarmUI's own launcher search order and persisting on RunPod / SimplePod network volumes - DOTNET_ROOT is exported so in-app Update and Restart now rebuilds with the correct SDK and works properly
Linux / cloud : Cloudflared is now installed system-wide from Cloudflare's official APT repository and launching prints the public trycloudflare.com URL so you can open SwarmUI from your own computer's browser on RunPod / SimplePod / Massed Compute - new --no-cloudflared and --host options added
RunPod, SimplePod and Massed Compute instruction files rewritten with exact HOW TO OPEN SWARMUI steps
Windows_Preset_Delete_Import.bat made more robust : works from any folder and auto picks the correct Python
To update : get latest zip file, extract and overwrite all, then run Windows_Install_or_Update_SwarmUI.bat and import Amazing_SwarmUI_Presets_v70.json
Also update your ComfyUI backend to latest V127 and get the 8 new LTX 2.5 ComfyUI presets : https://www.patreon.com/SECourses/posts/comfyui-auto-2-105023709
16 August 2026 Update V168
MiniMax H3 bundle updated to include newest 8-steps LoRA and also necessary Yolo Face model for new automatic ComfyUI Face Inpainting feature in presets for MiniMax H3
SwarmUI presets are now also using new Ref2V LoRA for references presets, you can switch to FL2V LoRA and see which one performs better so easy from LoRA tab and compare
Automatic video face inpainting implemented
Get latest zip file, extract and overwrite all, restart PC, run Windows_Update_SwarmUI.bat to update
Also import latest Amazing_SwarmUI_Presets json file
You can see impact of Face Inpainting here and now even better working : https://www.reddit.com/r/comfyui/comments/1vpl5bb/secourses_comfyui_installer_and_ready_presets_now/
14 August 2026 Update V166
We have added LTX 2.5 video upscaler and MiniMax Music 3 Text to Music presets into ComfyUI : https://www.patreon.com/SECourses/posts/comfyui-auto-2-105023709
LTX 2.5 video upscaler preset is able to upscale any video, any resolution, any aspect ratio into 2x
It is a very serious generative upscale so adds lots of details
Therefore, try to upscale at 1 chunk to have better consistency
Read preset information carefully after loading into ComfyUI
Our SECourses Premium Upscaler Pro app also fully supports LTX 2.5 now : https://www.patreon.com/SECourses/posts/secourses-pro-150202809
It is standalone does not require ComfyUI and it auto downloads models when you first time use
To make these presets work out of the box, 2 new ComfyUI core bundles added to the downloader app as below
LTX 2.5 video generation presets and bundles will be added soon hopefully but its quality lower than MiniMax H3
Moreover, I have converted the LTX 2.5 Int8 ConvRot myself higher quality and more accurate than officially published ones
Someone detected one of the model is corrupt and broken - officially released ConvRot Int8 variant
I converted them from BF16 versions after big research and experimentation
So our model downloader app will download better quality Int8 ConvRot of LTX 2.5 models
Our Musubi Trainer convert to quant tab also now supporting LTX 2.5 Int8 ConvRot compile : https://www.patreon.com/SECourses/posts/secourses-musubi-137551634
MiniMax Music 3 is reportedly lower quality than ACESTEP XL 1.5
Our very advanced ACESTEP XL 1.5 Premium app below and it has full LoRA and DoRA training support
MiniMax Music 3 also missing necessary resources to be trainable
MiniMax Music 3 prompt templates shared here : https://github.com/MiniMax-AI/MiniMax-Music3/tree/main/skills/music-caption-rewriter/templates
To get updated model downloader app, as usual, get latest zip file, extract and overwrite all
12 August 2026 Update V165
The most professional speed up LoRA maker LightX2V just published new speed LoRAs for the MiniMax H3
Our MiniMax H3 core bundles now auto downloads both FL2VA Turbo 8-step v1.0 and FL2VA Turbo 4-step v1.0 768p
Older v0.1 removed from downloads and presets so you can delete it from LoRA downloads if you did download before
All presets now by default uses FL2VA Turbo 8-step v1.0 but you can switch it to FL2VA Turbo 4-step v1.0 768p
If you switch to FL2VA Turbo 4-step v1.0 768p you may be needed to overwrite Video shift value and set to 6, SwarmUI default sets to 12
Sadly LoRAs only made for non-reference model atm but works on both models until they publish specific new LoRA for reference model
Quick set megapixels feature implemented
Based on selected aspect ratio, it will set resolution according to your set
Moreover, now it will display visualization of the output based on aspect ratio and resolution
To update get the latest zip file, extract and overwrite all, run Windows_Update_SwarmUI.bat and also use Windows_Preset_Delete_Import.bat to get latest updated presets
Moreover get latest ComfyUI backend zip file and also update it through installer bat file : https://www.patreon.com/SECourses/posts/comfyui-auto-2-105023709
10 August 2026 Update V162
ComfyUI infinite video with MiniMax H3 tutorial published : https://youtu.be/1580ZDX-60Q
Hopefully a tutorial for SwarmUI coming soon
Completely Rebuilt Download Engine
Exact byte-range resume after cancellation, connection loss, or application restart.
Previously downloaded ranges are preserved instead of downloaded again.
One sparse staging file replaces multiple temporary chunks and the expensive final merge.
Significantly lower temporary disk-space requirements.
Models are atomically installed only after size and SHA-256 verification.
Corrupt or incomplete downloads can never replace a finished model.
Faster Hugging Face Downloads
Native Hugging Face Xet is now recommended and enabled automatically when available.
Xet automatically scales between 8 and 32 streams according to available system RAM.
Improved recovery from stalled transfers, rate limits, server errors, and expired signed URLs.
Removed the previous forced approximately 84.5 GiB Xet cache.
New downloads default to no chunk cache, with an optional bounded reuse mode.
Improved Download Controls
Direct URL downloads now appear in the active-download counter.
Cancel and Cancel All now work with direct URL downloads.
Completed ranges remain available after cancellation.
Prevented overlapping standalone URL downloads.
Small and large direct downloads now use the same atomic download system.
Improved URL validation and rejection of HTML pages masquerading as model files.
Safer Hugging Face and CivitAI domain detection.
Catalog And Interface Improvements
Replaced the unavailable LTX 2.3 x2 spatial upscaler v1.0 with the official v1.1 long-video hotfix : ltx-2.3-spatial-upscaler-x2-1.1.safetensors
Removed duplicate physical models from search and LoRA displays while keeping aliases searchable.
Improved automatic folder detection for filenames containing hyphens and underscores.
Improved mobile and narrow-screen layout.
Updater now detects and repairs incomplete FoleyExtension installations, including optional-image components.
MiniMax H3 Improvements
Expanded the MiniMax H3 prompt-enhancement guide to v2.3.
Added stable reference-roster numbering rules for multi-scene and folder-batch prompts.
Added clearer image, video, audio, and paired-soundtrack reference rules.
Corrected first-and-last-frame preset instructions to use Image To Video > Video End Image.
For updating please also update ComfyUI to latest as well and also this one
Use Windows_Update_SwarmUI.bat and import latest presets file
9 August 2026 Update V160
Model downloader app updated and Int4 ConvRot MiniMax H3 models bundle added as Low VRAM
This bundle is great for 12 GB and below GPUs
At the presets both ComfyUI and SwarmUI, just replace Int8 Model with below Int4 variants
This format Int4 ConvRot models right out of the box working with our ComfyUI (use our installers) backend and installers for SwarmUI
We are using official code of Lightricks not Kijai implementation therefore our LoRA works better
However, Kijai had a VRAM optimization and i just implemented it
You will see it like below in presets, MAX savings reduces VRAM more than 40% but may reduce quality, exact saves 15%+ but same quality
Make sure to use our ComfyUI installer and update it for all features to run via our bat file
Remember MiniMax H3 specific features will only appear when you have selected MiniMax H3 architecture model in your model selection
If you set reference image size to max like this, it improves quality and accuracy but may use more memory, default is match
The importance of Max option is very significant
Video clip references are processed differently and below explains how they processed when used as a reference
8 August 2026 Update v159
This is a very major upgrade with so many new amazing stuff so please read carefully
Famous Lightricks released 4 steps LoRA for MiniMax H3
I have fully and accurately implemented it into our installers
Currently not working with SwarmUI or ComfyUI as default but works with our installers due to myself developed premium nodes and extensions
Make sure to install update your ComfyUI and SwarmUI users must also use our ComfyUI backend
This LoRA added into our model downloader MiniMax H3 core bundle please download
I have made new presets that uses this LoRA and does 8 steps so now you can use them
It can go as low as 4 steps but I recommend 8 steps
New updated presets are as below
Audio Only preset is super fast and generates audio directly
Audio generation can be used to see if your prompt and duration matching before generating video since it is like real 2x time speed
So 1 minute audio generation takes like 30 seconds on RTX 5090 therefore use audio generation preset to see if your prompt and duration matches - for speaking having videos
So quickly iterate, see if your audio accurate, then generate full video
Moreover, I recommend you to use 8 steps and 0.4 megapixel resolution to quickly generate your videos, verify they are accurate, then move full generation like 1344x768px and 20 steps high quality
This works great
Add references and write prompt field improved so that now you can add videos and audios with trim as you wish as below (trimming is optional)
For @video1's soundtrack, type <Audio 1>; @audio1 is the first standalone audio file
According to the above rule, MiniMax_H3_Enchance_Prompt_Feed_For_LLMs.txt file improved and literally working amazing
It is located inside Prompt_Generate_LTX_MiniMax_And_Presets_How_To_Use folder inside zip file
Trim interface supports both Audio and Video trimming
ComfyUI presets have amazing batch folder processing read changelogs : https://www.patreon.com/SECourses/posts/comfyui-auto-2-105023709
For SwarmUI, you can use Wildcards as a batch processing
To process them with order not randomly, enable Display Advanced Options
Then find Swarm Internal and change Wildcard Seed Behaviour to index
Set regular seed 0 so it will start from first prompt in Wildcard and how many generations you make, it will make with order
For everything to work, make sure that you use our ComfyUI installer as a backend and it is updated : https://www.patreon.com/SECourses/posts/comfyui-auto-2-105023709
Then get latest SwarmUI zip file, extract and overwrite all, then run Windows_Update_SwarmUI.bat to update
Both install and update installs necessary extensions automatically
To update presets, i recommend use Windows_Preset_Delete_Import.bat
You can also manually import latest Amazing_SwarmUI_Presets json file and overwrite all
6 August 2026 V158 - Update 2
We have significantly improved our add references extension - auto installed with our installer
Now when you add or remove an attachment, it will not modify and alter your written prompt
Now you can drag and drop attachments to change their position like @image1 into @image3 etc
Now even if you have reference in the prompt but not in attachment, it will not throw error and at the back end it will remove @image1 etc and prompt will just run fine
Zip file is same just run windows update bat file
Also I published a speed up quality comparison video : https://www.reddit.com/r/comfyui/comments/1vh5wd8/i_have_implemented_solattn_crossstep_cache_from/
6 August 2026 V158
I have fully and accurately implemented https://nvlabs.github.io/Sana/Sol-Engine/H3-OnDevice/ into our ComfyUI installation via our always installed and myself developed custom nodes
Deploy MiniMax-H3 with Sol Engine achieving up to 4.52× acceleration on on-device GPUs
Their claimed 4.52x speed acceleration happens at 50 steps and default attention meanwhile I measured with 20 steps and against Sage Attention 2.8.3 latest (already speeds up like 20%+)
Moreover, I did set more conservative but almost same quality 0.08 thus it skips lesser steps
Use our ComfyUI backend installer and update it so this will work : https://www.patreon.com/SECourses/posts/105023709
Then updated our SwarmUI extensions to fully support it
Existing presets have a minor fix and they will work
When you select a MiniMax H3 architecture model, the parameters will appear like this - the backend has to be loaded so wait a little bit for it to appear
Default values quality wise conservative so almost same quality and brings 1.39x speed compared to Sage Attention 2 speeds at 1344x768 with 362 frames
It used only 20 GB vram for 1344x768 px at 362 frames - 15 seconds but it can use even lower depending on your GPU VRAM
The speed up is not enabled by default so check checkbox to enable it
The first run will be slow due to kernel compile but consquent runs will be faster and won't recompile
In logs first 4 steps are same, then step speed increases and then you start to see jumps in steps like below
For updating get latest zip file, overwrite all files, and then run Windows_Update_SwarmUI.bat - fresh install will also get this
Make sure that you have updated your ComfyUI backend with our ComfyUI installer
5 August 2026 V157
We have officially moved enhanced SwarmUI with this update with the extensions I developed - all auto installed - please read carefully
MiniMax H3 is now fully supported with our Enhanced SwarmUI and ComfyUI installers and model downloaders
Can generate up to 30 seconds and 2K native resolution just fine
In SwarmUI model downloader - download MiniMax H3 Core bundle
All of our presets both ComfyUI and SwarmUI, uses this bundle therefore downloading this will let you run presets right away
Even if you have low VRAM GPU still this model working perfect
Make sure to have at least 100 GB virtual RAM set
Current GGUF releases are all almost same size as Int8 ConvRot therefore they are useless
If Q4 proper one gets released I plan to add that into MiniMax-H3 Video Models section inside Video Generation Models tab
5 perfectly working new amazing MiniMax 3 presets with latest Amazing_SwarmUI_Presets_v62.json
MiniMax H3 Beginning and Ending Image/Frame to Video - 260804
MiniMax H3 Image to Video - 260804
MiniMax H3 References to Video - 260804
MiniMax H3 Text to Image with Optional References - 260804
MiniMax H3 Text to Video - 260804
I recommend you to use Windows_Preset_Delete_Import.bat but you can always manually import too
Now recommended to add --disable-pinned-memory into your ComfyUI backend arguments until its bugs fixed to not have any OOM error
I have improved our special installers and extensions therefore you can use prompt input field to add 3 audio, 3 video and 9 image references to use with MiniMax H3 in SwarmUI easily
They are referenced and named as @image1 @image2 @video1 @video2 @audio1 @audio2 and so on
The special upload will be allowed only when you select MiniMax H3 model
It auto detects and enables based on selected model's architecture
Official SwarmUI still doesn't have this feature but we have with our installers
We still always install official latest version SwarmUI but this comes from our own developed extension
New folder : Prompt_Generate_LTX_MiniMax_And_Presets_How_To_Use
Our system uses references as @image1 @image2 @video1 @video2 @audio1 @audio2 and so on
Up to 9 images, 3 audios and 3 videos
<audio 1> means the audio of the first input <video 1>
MiniMax_H3_Enchance_Prompt_Feed_For_LLMs.txt file is modified to instruct LLM to write prompts according to our system (as mentioned above) so you will get perfectly ready copy to paste prompts
New quick video extend feature
Works with all of the init image to video models including MiniMax H3
Apply Image to Video preset
Then into init image field, instead of an image, input a video
Then enable the Continue Init Video From Last Frame checkbox
It will get last frame of your input video, generate image to video, and then finally merge videos and generate final output
Installers and update bat file made much more robust both for Windows and Linux
Get latest zip file, overwrite all, and run Windows_Update_SwarmUI.bat
19 July 2026 V154
This is a very big update with so many important new features, changes and improvements so please read all thank you
Seven New Presets
Make sure to import and overwrite Amazing_SwarmUI_Presets_v59.json or use Windows_Preset_Delete_Import.bat while SwarmUI running
Krea 2
Krea 2 Raw/Base Slow - 260711 uses 52 steps, Euler, Simple, and CFG 4.5. It is the slower Raw/Base path intended mainly for fine-tuning, post-training, and LoRA work.
Krea 2 Turbo Fast - 260711 uses 8 steps, Euler, Beta, and CFG 1 for fast final inference.
Krea 2 Turbo Image Edit - 260716 adds denoise-based image editing with a default creativity value of 0.65. Add the source as Init Image and describe the requested change.
The new Krea 2 Core Bundle contains the Raw/Base INT8 HQ model, Turbo INT8 HQ model, Qwen3-VL 4B text encoder, and Qwen Image VAE. Its cached total is approximately 34.83 GiB.
The catalog also adds individually selectable Krea 2 Turbo Q8 GGUF and NVFP4 High Quality variants.
Automatically installed ComfyUI-QuantOps updated (With our ComfyUI Installer for SwarmUI Backend)
Now it is only used if native ComfyUI is not supporting the loaded model
I have compared the new famous Int8 ConvRot of Krea 2 and the speed difference is like 100%, I plan to update all models to Int8 ConvRot
New updated ComfyUI-QuantOps supports Krea 2 GGUF as well
When you click and see full size of above image and analyze results you will see that:
Int8 ConvRot is 96.2% similar to BF16 meanwhile GGUF Q8 is only 90.0% and FP8 Scaled is 82.2% and NVFP4 is 63.7%
Moreover, Int8 ConvRot generates the output in 3.05 seconds, making it 1.82× faster than BF16, which takes 5.56 seconds.
NVFP4 takes 3.8 seconds and is 1.46× faster than BF16, whereas GGUF Q8 takes 6.06 seconds and is approximately 8.3% slower than BF16.
So Int8 ConvRot generated with our Musubi Trainer app at high quality is almost 100% faster and almost same quality as BF16
High quality generation takes few hours on RTX 5090
Updated SwarmUI downloader has amazing quality Int8 ConvRot which reaches almost BF16 quality : https://www.patreon.com/SECourses/posts/swarmui-auto-and-114517862
Krea 2 Core bundle downloads this model and uses it in SwarmUI preset
ComfyUI preset also uses that model default
You can generate Int8 ConvRot models with our updated Musubi Trainer app : https://www.patreon.com/SECourses/posts/secourses-musubi-137551634
LTX 2.3 Foley Video-to-Audio
The new LTX23 Foley Video To Audio preset generates synchronized audio for a silent or muted video while keeping the source video unchanged.
Put the video in Init Image, then describe the visible sound source, action, material, and timing. The preset uses the LTX 2.3 Foley V2A LoRA with 30 steps, audio CFG 6, STG scale 1, modality scale 3, and a maximum of 169 frames at 24 FPS.
The Foley LoRA is approximately 0.21 GiB and is now part of the LTX 2.3 Video Core Bundle. The updated premium installer also installs the maintained FoleyExtension node package automatically.
This model uses Dev version of LTX 2.3 not Turbo, therefore, now Dev version is included in the LTX 2.3 core bundle
For clean Foley generation, a useful prompt ending is: No speech is present. No music is present.
Download LTX 2.3 Core Bundle and in ComfyUI, install bundle 100
LTX 2.3 Licon MSR V2
The new LTX23 Licon MSR V2 Multi Subject Reference preset supports multiple subject, object, texture, or viewpoint references plus a required background image.
Add two to five images to Prompt Images in this exact order:
One to four subject, object, texture, or viewpoint references (so first 1-4 images are subjects)
The required background image last (last image is always background)
Therefore, you need minimum 2 input images into the prompt field
Do not use Init Image for this workflow. Identify each numbered reference and its role in the positive prompt, then describe the new action, scene, camera, and lighting.
The preset uses the official eight-step distilled sigma schedule, Euler Ancestral sampling, a 65-frame internal reference sequence, and the LTX 2.3 Licon MSR V2 IC-LoRA. The new LoRA is approximately 0.61 GiB and is also included in the LTX 2.3 Video Core Bundle.
A dedicated guide is included with reference ordering, recommended settings, installation paths, and licensing details.
For this to work, we have coded custom extensions and all is automatically installer with SwarmUI installer or Updater files
Download LTX 2.3 Core Bundle and in ComfyUI, install Bundle 100 for ComfyUI backend
It is insanely powerful and fast I mean look below example
LTX 2.3 Presets Updates
LTX 2.0 model presets removed since LTX 2.3 is better in everyway
LTX 2.3 presets now using myself compiled Int8 ConvRot HQ models since they are literally 100% faster
Model downloader LTX 2.3 core bundle is now downloading these new models
You can generate Int8 ConvRot models with our updated Musubi Trainer app : https://www.patreon.com/SECourses/posts/secourses-musubi-137551634
So Int8 ConvRot HQ is 100% faster than FP8 Quant Scaled and 50% faster than BF16 on RTX 5090
The quality is also excellent almost same as BF16
Phantom-Wan Character Reference T2V
Two Phantom-Wan presets are included:
Phantom Wan 14B Character Reference T2V Fast
Phantom Wan 14B Character Reference T2V Quality
Add character references to Prompt Images, not Init Image, then describe every subject, the scene, action, camera, and lighting. The reference images condition identity while generation begins from an empty video latent, so this is true reference-conditioned text-to-video rather than first-frame image-to-video.
One to four references are the officially recommended path. The integration can accept up to six for experimentation.
Download Phantom Wan 14B Character Reference T2V Bundle and in ComfyUI, install Bundle 100 for ComfyUI backend
Bundle 100 for ComfyUI Backend
Inside ComfyUI zip file : Windows_Custom_Nodes_Bundles_Installer.bat
Unified Fast Robust Model Downloader Improvements
Run Windows_Start_Download_Models_App.bat to start Downloader
The previous hf_transfer checkbox has been replaced with two clearer controls:
Custom Downloader Threads: 1 to 32, with 16 as the recommended default
Hugging Face Xet Backend: optional and disabled by default
The custom thread setting applies to individual catalog downloads, bulk queues, bundles, search results, snapshots, and direct URL downloads. When Xet is enabled, the custom thread setting is ignored. If Xet fails, v150 falls back to the custom resumable backend unless the operation was cancelled.
The custom backend now has much stronger recovery behavior:
Known-size partial downloads are preserved when cancelled.
Parallel downloads save their range layout so later attempts can resume safely.
Legacy v145 16-part downloads can be migrated and resumed.
Every retry re-reads the saved byte count to prevent duplicate ranges and corruption.
Six attempts use bounded exponential backoff.
HTTP 429, 500, 502, 503, and 504 responses receive controlled retries.
Separate connection and idle-read timeouts recover stalled transfers sooner.
Progress output reports when it is waiting for the server and uses smoother speed estimates.
Free space is checked before downloading and before merging parallel parts, with a safety reserve.
Disk-full failures preserve useful partial data.
Parallel parts are merged to a temporary file and atomically activated.
Hugging Face main file URLs can use SHA verification and the optional Xet path.
SHA-256 verification and the verified-file cache remain in place. The goal is simple: fewer restarts from zero, clearer logs, and safer recovery on large model downloads.
Now you can set Hugging Face token for faster or private repo downloads
Now you can set HF Xet Download - even faster optional
Now you can set number of download threads
Now you can set Parallel File Downloads count
This is extremely useful on high bandwith systems to download even faster than Hugging Face single file download limits like reaching 1 GB per second
Installation and Update Reliability
Windows install and update scripts no longer require one exact .NET SDK patch version. The new helper searches common system, user, registry, PATH, and environment locations for any usable stable .NET 10.x SDK while ignoring preview-only builds.
If installation is needed, it tries WinGet first and then falls back to the latest official Microsoft installer for x64, ARM64, or x86. The fallback download is checked against Microsoft's published release hash before it runs, and installer progress and log locations are shown instead of failing silently.
The premium-extension installer now validates the required integrations before changing the active setup. It installs or updates:
FoleyExtension nodes plus the managed LTX 2.3 Foley SwarmUI integration
Phantom character-reference helper nodes for the Comfy backend
The managed Licon MSR SwarmUI integration
The official Licon MSR Comfy node at the tested pinned revision
The new helper-node installers use staged directories, required-file checks, backup/rollback paths, bounded Git operations, and stale compiled-extension cleanup. Windows, RunPod, and Massed Compute update instructions now stop with an explicit error if this stage fails instead of continuing into a partial setup.
Updated Preset Model Report
A new file, Windows_Update_And_Open_Model_Report.bat, regenerates and opens the HTML report that shows which bundles and model files cover every SwarmUI preset.
The report generator now handles Krea 2, Phantom-Wan, Ideogram 4, negative-model parameters, saved-filename aliases, and LTX 2.3 Dev versus Distilled connector selection more accurately. The packaged report has been refreshed against Amazing_SwarmUI_Presets_v55.json.
How to Upgrade
Existing installation
Close SwarmUI and the model downloader.
Back up custom presets and any source changes you made inside the SwarmUI Git checkout. The updater resets tracked checkout files before pulling upstream changes.
Extract the latest zip file over your existing downloader package folder and allow release files to be replaced. Do not delete your SwarmUI or model directories.
Run Windows_Update_SwarmUI.bat.
Start the model downloader and download only the new bundles or individual files you need.
Import Amazing_SwarmUI_Presets_v55.json.
Windows_Preset_Delete_Import.bat can automate the preset replacement while SwarmUI is running. It first backs up the current presets, then deletes all active presets and imports the latest pack. Read its confirmation carefully if you maintain custom presets.
Users who already have the rest of LTX 2.3 can download only the new Foley and Licon LoRAs instead of downloading the full LTX bundle again.
Run Windows_Update_And_Open_Model_Report.bat afterward to compare the new preset requirements with the models currently on disk.
Fresh installation
Extract latest zip file into a clean folder.
Run Windows_Install_SwarmUI.bat.
Start the model downloader with Windows_Start_Download_Models_App.bat.
Download the bundle or individual models needed by your chosen presets.
Start SwarmUI and import latest Amazing_SwarmUI_Presets_.json
RunPod and Massed Compute users should follow the updated instruction files included in the package.
2 July 2026 V145
SwarmUI going to require .NET SDK 10 therefore, both installer and bat file will auto install .NET SDK 10 if you are missing
It will ask permission to install so click yes / ok
RunPod, SimplePod, Massed Compute installers also updated to auto install, no need confirm there
Ideogram 4 models added into Image Generation Models tab in model downloader app
For JSON prompt generation for Ideogram 4 model, we have ultimate new app : https://www.patreon.com/SECourses/posts/162527725
Ideogram 4 Core Bundle added into SwarmUI Bundles tab in model downloader app
With Amazing_SwarmUI_Presets_v51.json 3 new presets added as below
Ideogram 4 Highest Quality - 260629.json
Ideogram 4 Balanced - 260629.json
Ideogram 4 Turbo - 260629.json
All 3 preset uses Ideogram 4 unconditional model as well for high quality but you can disable if you want
I think when you disable it becomes more non-realistic
Ideogram 4 Turbo generates amazing images at 2048x2048 and really fast speed
Use Windows_Preset_Delete_Import.bat to fresh update
If you have custom presets they will be deleted be careful
You can also import Amazing_SwarmUI_Presets_v51.json and just import new ones
Updated new all presets
26 April 2026 V142
New LoRA LTX2.3_Crisp_Enhance.safetensors (0.66 GB) added to the LTX 2.3 Core bundle in our model downloader app
All SwarmUI LTX 2.3 presets updated to use this new LoRA which improves quality significantly
Get latest zip file and get v50 preset file
Which_Bundles_Downloads_Which_Preset_Models.html updated
23 April 2026 V141
LTX 2.3 Video Core Bundle updated and the following new LoRAs added
LTX23_Anime-To-Real_v1.safetensors (0.61 GB)
LTX23_Edit-Anything_v1.safetensors (1.22 GB)
LTX23_Real-To-Anime_v1.safetensors (0.61 GB)
Default model replaced to LTX 2.3 v1.1 : LTX2.3-Distilled-v1.1-FP8-Quant-Scaled.safetensors
I have converted it to Quant Scaled very high quality
LTX 2.3 presets updated to use v1.1 model as default
New preset LTX23 Video To Video 8 Steps added
This preset takes video as input instead of image from same place
It automatically recognize it is video or image don't worry
By using above LoRAs or even without LoRA, you can change existing videos
Play prompt and Init Image Creativity level (default 0.5) to get your desired results
Example simple prompt : Convert the video into a anime style
New updated presets are provided with Amazing_SwarmUI_Presets_v49.json
I recommend to use Windows_Preset_Delete_Import.bat file
Our model downloader latest statistics are as below
22 April 2026 V140
New precision filters added
New FLUX 2 Klein Models (11 models - Total: 105.17 GB) and ERNIE Image Models (7 models - Total: 59.45 GB) added into Image Generation Models
New bundles added
FLUX 2 Klein Core Bundle (Total: 34.74 GB, 5 models)
ERNIE Image Core Bundle (Total: 23.12 GB, 4 models)
I have compiled myself the following models for you via our SECourses Musubi Tuner model quantizer so you get them first time in the entire community - https://www.patreon.com/posts/137551634
ernie-image-turbo-Quant-FP8-Scaled.safetensors (7.81 GB)
ernie-image-Quant-FP8-Scaled.safetensors (7.81 GB)
FLUX-2-Klein-Base-9b-Quant-FP8.safetensors (8.79 GB)
FLUX-2-Klein-Distilled-Quant-FP8-Scaled-9b-kv.safetensors (8.79 GB)
FLUX-2-Klein-Distilled-9b-Quant-FP8-Scaled.safetensors (8.79 GB)
Quant FP8 Models are all tested and excellent quality
After extensive testing on 4x RTX PRO 6000 GPUs having Massed Compute machine, new following presets added to the our arsenal with Amazing_SwarmUI_Presets_v48.json
ERNIE Image Base - 260422
ERNIE Image Turbo - 260422
FLUX 2 Klein Base - 260422
FLUX 2 Klein Distilled 8 Steps - 260422
You can use Windows_Preset_Delete_Import.bat or manually import new presets
For presets to work, download FLUX 2 Klein Core Bundle (Total: 34.74 GB, 5 models) and ERNIE Image Core Bundle (Total: 23.12 GB, 4 models)
Tutorials tab on the downloader app updated
Get latest zip file and overwrite older files to have updated files
30 March 2026 V138
I have compiled and added LTX2.3-Dev-FP8-Quant-Scaled.safetensors (27.14 GB) to model list
SwarmUI presets updated according to the latest SwarmUI version so you won't get warnings, i recommend overwrite all others via Windows_Preset_Delete_Import.bat
Gradio version of the app upgraded now even better and faster
16 March 2026 V137
LTX 2.3 models bundle updated for ComfyUI presets - use ComfyUI latest zip file to get them
LTX2.3_Enchance_Prompt_Feed_For_LLMs.txt significantly improved
Works great for Text to Video and Image to Video prompt generation
Upload it to your favorite LLM e.g. ChatGPT, provide input image or whatever you want and describe and tell LLM to generate / improve prompt
13 March 2026 V135
Model downloader app upgraded to Gradio 6.9.0 and working blazing fast now
New LTX 2.3 Video Models (26 models - Total: 275.22 GB) added
Some model sizes were wrongly displayed and it is fixed
Which_Bundles_Downloads_Which_Preset_Models.html file updated
New LTX 2.3 Video Core Bundle (Total: 67.70 GB, 15 models) added please download this
I have myself compiled the very best FP8 Quant Scaled LTX 2.3 model
At the moment GGUF of LTX 2.3 not working in SwarmUI at the moment since ComfyUI didn't add native support yet, thus i recommend using FP8 Quant Scaled should work amazing if you have RAM memory
We still have GGUF models in downloader app and I plan to update ComfyUI with presets for GGUF
Our ComfyUI already supports GGUF of LTX 2.3
New 2 presets added with Amazing_SwarmUI_Presets_v46.json
LTX23 Image To Video 8 Steps - 260310
LTX23 Text To Video 8 Steps - 260310
No LoRAs used at the moment but I may update in future if there be nice LoRAs
Atm working amazing even without any LoRAs
Download latest zip file and overwrite all files in install folder and use newest model downloader app and import new presets
28 February 2026 V133
Gradio version upgraded to 6.8.0 now downloader app runs faster
RIFE frame interpolation repo was outdated, i forked and fixed it in SwarmUI installers
There was a bug when downloading models from CivitAI and now it is made more robust and error fixed
14 February 2026 V131
This update was for our Ultimate Unified AI Models downloader app
Massive revamp and improvement made to the User Interface (UI) and User Experience (UX)
We have at least 10x improved the app performance and responsiveness of the UI
I also have reported a bug to Gradio team to even further improve performance hopefully
URL downloader significantly improved and made more robust
Tutorials section fully updated
Check the below screenshots to see and please try every feature and tab and let me know what else you want
For updating as usual download the zip file and overwrite all of the previous files in your installation folder
3 February 2026 V130
Model downloader app interface significantly modernized and improved
CMD download logs and messages improved
LTX2_Enchance_Prompt_Feed_For_LLMs.txt added into zip file which you can upload to https://aistudio.google.com/prompts/new_chat and write your prompt and that is it. You can also upload image + this file and prompt to enhance your prompt for free
Following new presets added into our amazing SwarmUI presets with Amazing_SwarmUI_Presets_v45.json - so update your presets
LTX2 Image To Video 8 Steps
Image to video works great and upscale normally not supported
For this preset to make also LTX 2 upscale work, I have developed our first SwarmUI premium extension and now it is auto installed with SwarmUI installers and also update bat files
Also for upscale, you need to set your base resolution half and apply upscale like 480x480 and 2x upscaled to 960x960
I am using distilled FP8 Scaled model directly thus it is 8 steps
LTX2 Text To Video 8 Steps
Working amazing and you can use it with LTX2 Video Upscale preset
To use with LTX2 Video Upscale preset, reduce your base resolution to half like 480x480 instead of 960x960
Z Image Base preset added and working amazing new Z Image base model
The preset is 40 steps but you can do lesser if you want
Also i recommend you to combine it with Upscale Images 2X preset to get very high res and quality images
Currently Z Image doen't work accurately with --use-sage-attention so you can replace attention with --use-pytorch-cross-attention
For the model downloader app the following changes made
Z Image Turbo Core Bundle renamed into Z Image Core Bundle
For Z Image presets you need this bundle
Z Image BF16 model added into Z Image Core Bundle - this is new today published base model and works great
Z Image Turbo Models renamed into Z-Image Models
Z Image BF16 (11.46 GB) and Z Image Quant FP8 Scaled (5.86 GB) models added
I have compiled myself the Z Image Quant FP8 Scaled and its quality almost same
I have used our SECourses Musubi Trainer app with highest quality Quant FP8 Scaled Maker
LTX 2 Video Models added to the downloader app
LTX 2 Video Core Bundle added to the app
For LTX 2 presets you need this bundle
If you get OOM with LTX 2 presets, add the following arguments into your ComfyUI backend and test
--lowvram or --novram
Also you can try --disable-smart-memory
All these can be also combined with --cache-none : lowest RAM
LTX 2 Video Models also has LTX-2-19b-Dev-GGUF_Q4_K_M (12.46 GB) and gemma_3_12B_it_fp4_mixed.safetensors (8.80 GB) for low VRAM + low RAM machines
Complete Image Generation and Editing Bundle now includes new Z Image BF16 model as well
I am working on a tutorial video to explain all hopefully
And hopefully I will update our ComfyUI installer as well to add there too
21 January 2026 V127
MassedCompute_SwarmUI_Install_Instructions-Optional.txt added
This is optional since we have SwarmUI already in Massed Compute as shown in tutorials
You can use this to install SwarmUI in Linux machines
I tested and it works perfect
15 January 2026 V126
I have updated the NVFP4 quantizer app and implemented Mixed NVFP4 for FLUX models
With this change, new FLUX SRPO Mixed NVFP4 model added into app which has amazing quality : https://imgsli.com/NDQyNjk5
Mixed NVFP4 speed is same and size is slightly bigger than full NVFP4
14 January 2026 V125
I have developed a NVFP4 quantizer app and published here : https://www.patreon.com/posts/148217625
With using it I have generated FLUX SRPO NVFP4 and it is amazing for low VRAM GPUs and ultra fast
FLUX SRPO NVFP4 added into the downloader app into FLUX Models section
We are the first one to compile and publish it
Hopefully I will work on Qwen 2512 later
13 January 2026 V124
New model FLUX 1 Dev Kontext Quant FP8 Scaled added to the downloader app
I compiled it myself and amazing quality
ComfyUI is now supporting NVFP4 LoRAs not perfect and not all models but i believe it will hopefully get better, I tested on FLUX 2, FLUX 1 and Z Image Turbo
FLUX 2 failed, FLUX 1 i think worked nice, Z Image Turbo quality dropped
RunPod template link updated and now we fully support SimplePod which is much faster and cheaper than RunPod
SIMPLEPOD CHEAPER AND FASTER THAN RUNPOD
Now we fully support SimplePod as well please use this link to register : https://simplepod.ai/ref?user=secourses
SimplePod is faster and cheaper than RunPod and works exactly same
E.g. RTX 5090 on RunPod is 0.89 USD per hour, on SimplePod it is 0.45$ per hour,
RTX PRO 6000 on RunPod is 1.84 USD per hour and on SimplePod it is 0.79 USD per hour
Please use this template on SimplePod : https://dash.simplepod.ai/account/explore/100/ref-secourses/
For permanent storage, generate it from Storage tab with any name and size you want and when selecting template with above link, click Edit and Use, select Persistence Volume and change mount point to /workspace
Up-to-date SimplePod tutorial starting from 21:51 : https://youtu.be/yOj9PYq3XYM?si=Z86wZZLBeYzWo1Qo&t=1311
As usual follow Massed_Compute_Instructions_READ.txt and RunPod_SimplePod_Instructions_READ.txt to install and use and watch the tutorials
10 January 2026 V123
Full tutorial published : https://youtu.be/yOj9PYq3XYM
We have added NVFP4 models both into their respected Model category and as a bundle
FLUX 2 Dev NVFP4 (21.21 GB)
FLUX Dev Kontext NVFP4 (8.56 GB)
FLUX Dev NVFP4 (8.56 GB)
Z Image Turbo NVFP4 (4.20 GB)
NVFP4 models uses same presets as FP8 Scaled or BF16
Sadly, currently, LoRAs not working with NVFP4
I always recommend FP8 Scaled over GGUF models if your RAM is sufficient to do Block Swapping / VRAM Streaming
Use --disable-smart-memory args in your ComfyUI backend if you are getting stuck or OOM - this reduces VRAM usage but a little bit slower
To use NVFP4 models, you have to install latest V73 or above ComfyUI read its announcement : https://www.patreon.com/posts/105023709
I have compiled Quant FP8 Scaled of FLUX 1 Dev and added to downloads
Moreover, for lower VRAM machines, Z Image Turbo now has the following text encoder models
Qwen 3 4B Text Encoder FP8 Mixed (For Z Image Turbo) (5.25 GB)
Qwen 3 4B Text Encoder FP4 Mixed (For Z Image Turbo) (3.24 GB)
They will be saved with SwarmUI expected file name so you don't need to modify presets
Don't forget to overwrite all previous files when updating, including utilities folder super important
Hopefully I am producing a new tutorial for NVFP4, quality comparison and LTX 2 presets coming soon
The Speed difference of NVFP4 is massive as below
7 January 2026 V120
SwarmUI tag error (visual not important) and Tecache import (ComfyUI broken it and i fixed the custom node myself) errors fixed
3 January 2026 V119
New Which_Bundles_Downloads_Which_Preset_Models.html added to the zip file
You can use that file to see which bundles in Windows_Start_Download_Models_App.bat downloads the necessary models for that preset
It covers all of the presets we have e.g. 2 examples shown below
1 January 2025 V118
I have spent literally like 1 day to compile new Qwen Image 2512 model into best quality Quant FP8 Scaled
I had to do more than 10 compiles, comparisons and find best workflow to compile
Each quality compile takes like 2 hours on RTX 5090
Our compiled model is significantly better than ComfyUI published raw FP8
You can compare and see yourself if you wish :)
This new model added into our downloader app as Qwen_Image_2512_Quant_FP8_Scaled.safetensors
Older Qwen Image removed from Qwen Image Core Bundle and Complete Image Generation and Editing Bundle and replaced with newer Qwen_Image_2512_Quant_FP8_Scaled.safetensors
With Amazing_SwarmUI_Presets_v43.json the following changes made - it took me literally like 1000 image generation and massive amount of analysis
Qwen Image 50 Steps Official Slow removed from presets since it is not useful or better than anything
Qwen Image Edit Plus 50 Steps removed from presets since it is not useful or better than anything
Qwen Image Realism Fast removed from presets since it is not useful or better than anything
Qwen Image UHD UlRealism 4+4 Steps - was typo duplicate removed
Those still exists in Amazing_SwarmUI_Presets_v41 file which is in older_presets folder
Use Qwen Image Core Bundle to get newer needed models
Qwen Image 8 Realism Fast upgraded to Qwen Image 8 Realism Fast 260101
Now uses Qwen Image 2512 and new turbo LoRA, upgrade difference is massive (same duration ultra fast) : https://imgsli.com/NDM3OTIz
Qwen Image 8 Steps Ultra Fast upgraded to Qwen Image 8 Steps Ultra Fast 260101
Now uses Qwen Image 2512 and new turbo LoRA - for non realistic images can be used
Qwen Image Edit Plus UHD Realism - 4+4 Steps upgraded into Qwen Image Edit 2511 UHD Realism - 4+4 Steps - 260101:
Now uses Qwen Image Edit 2511 and new turbo LoRA, upgrade difference is massive (same duration ultra fast) : https://imgsli.com/NDM3OTUx
Qwen Image 2512 UHD High Realism 4+4 Steps - 260101 added to the presets
Qwen Image UHD Realism Tier - 4+4 Steps upgraded to Qwen Image 2512 UHD Realism - 4+4 Steps - 260101
Now uses Qwen Image 2512 and new turbo LoRA, upgrade difference is massive (same duration ultra fast) : https://imgsli.com/NDM3OTUz
Both Qwen Images Stylized UHD Tier 1 and Qwen Images Stylized UHD Tier 2 are valid and up-to-date for Qwen Image Base or Qwen Image 2512 or Qwen Image Edit 2509 or Qwen Image Edit 2511.
Now I am adding last updated dates to titles and also preset descriptions so you can keep and not overwrite if you wish.
Hopefully I will convert all presets into ComfyUI workflows and publish them seperately in our ComfyUI installer.
31 December 2025 V116
After doing massive research on new FLUX 2 Turbo LoRA I have come up with the very best preset
The research literally took over 1000 image generation and lots of comparison
Recommended bundle is FLUX 2 Core Bundle but if your RAM becomes not enough there are 2 options
My models are default set 2048x2048 thus generated images are 4 megapixel and this model works best at this resolution
4 Megapixel image generation takes like 40 seconds on RTX 5090 with FP8 Scaled model
1st: Modify your backend and add this as args --cache-none
This really improves RAM management - VRAM is handled by ComfyUI so the RAM is the real issue for this model
Because it caches text encoder and completey deloads text encoder model - Mistral it is massive
2nd: Download and use FLUX 2 Low RAM Bundle
This bundle uses much lesser RAM and VRAM since it is Q4 GGUF both Text Encoder and the Model but lower quality and slower than FP8 Scaled
4 New presets added as below
FLUX 2 - 8 Steps Fast HQ - 251230, FLUX 2 - 8 Steps Fast HQ - Low RAM - 251230, FLUX 2 - 8 Steps Fast Realism - 251230, FLUX 2 - 8 Steps Fast Realism - Low RAM - 251230
Since requested now I will put date to new presets e.g. 251230 = 2025, month 12, day 30
Some download bugs fixed and downloader app improved
28 December 2025 V115
Z-Image-Turbo-Fun-Controlnet-Union-2.1-8steps (6.25 GB) added to the downloads
I believe it is a little bit better than previous Z-Image-Turbo-Fun-Controlnet-Union
To learn how to use ControlNet here : https://youtu.be/ezD6QO14kRc
24 December 2025 V114
Our downloader app now has Qwen_Image_Edit_2511_Quant_Scaled_FP8
I have made this model myself by using the special ComfyUI-QuantOps repo
It took like 2 hours on RTX 5090 and AMD 9950X CPU
This is the ultimate quality of quantization right now, half space, almost full BF16 quality, ultra fast
For this to work, we have to install this ComfyUI-QuantOps repo as a custom node into ComfyUI
Thus get the latest ComfyUI zip file, extract into your ComfyUI folder, overwrite and run Windows_Install_Or_Update_ComfyUI.bat file : https://www.patreon.com/posts/105023709
Moreover, Complete Image Generation and Editing Bundle and
Qwen Image Core Bundle now downloads Qwen Image 2511 instead of older 2509Futhermore, older Qwen Image 25-09 preset removed and new Qwen Image Edit - 2511 - 12 Steps: added
For this preset to work you need to have downloaded Qwen_Image_Edit_2511_Quant_Scaled_FP8 base model or BF16 version and Qwen-Image-FP8-Lightning-4steps-V1.0-fp32 LoRA - yes this LoRA is working best among all, i tested all
I recommend you to download Qwen Image Core Bundle
Qwen Image Edit 2511 Quality is significntly better than 25-09, you can see how 25-11 fixed image coloring issue here : https://imgsli.com/NDM2MzE3/0/1
It can do higher resolution as well so you can change base resolution of model to 1920x1920 from 1328x1328 for higher quality. 2560x2560 works even better compare for your use cases.
Qwen Image Edit 25-11 BF16 added to Qwen Image Edit models bundle
New tutorial Qwen Image Edit 25-09 vs 25-11 data shared with prompts and metadata having images and source images : Qwen_Image_Edit_2509_vs_2511.zip
Moreover now we dont enable Smart Image Prompt Resizing and preset will disable it auto for you so it is 1-click to use as usual
18 December 2025 V111
Wan 2.2 Core 8 Steps Bundle renamed into Wan 2.2 Core 4 Steps Bundle since now presets are using 4-steps with newest speed up LoRAs and models
2 New models added to following bundles (latest Wan 2.2 Text To Video speed up LoRAs)
Wan2.2_T2V_High_Noise_Lightx2v_4steps_LoRA_1217.safetensors (0.57 GB)
Wan2.2_T2V_Low_Noise_Lightx2v_4steps_LoRA_1217.safetensors (0.57 GB)
Bundles : Wan 2.2 Core 4 Steps Bundle and Wan 2.2 LoRAs and Complete Image Generation and Editing Bundle
Wan 2.2 presets updated to generate 16 FPS instead of 24 or 20 and 81 frames as default (5 seconds)
Moreover, we have enabled RIFE 2x FPS since our installers now auto installs RIFE thus you will get 5 seconds 32 FPS videos
Read the new presets descriptions and use Amazing_SwarmUI_Presets_v38.json
11 December 2025 V109
Model downloader app installers upgraded to Gradio 6.1.0
There was a massive bug that caused massive slowness for Downloader App and after I have reported they fixed the issue with Gradio 6.1.0 and now Model Downloader app is 100x more faster and responsive than before
Just extract zip file and overwrite older files and start model downloader as usual with Windows_Start_Download_Models_App.bat - it will auto upgrade and use latest version
3 December 2025 V108
Z-Image-Turbo-Fun-Controlnet-Union model added to the downloader
Included in Z Image Turbo Core Bundle, Complete Image Generation and Editing Bundle and Z Image Turbo Models
I did set its metadata accurately so it will work right away out of the box
SwarmUI and ComfyUI uses different folders so check ComfyUI checkbox if you need
Read its description for how to use but I am working on a tutorial to show Z Image Turbo models training in AI Toolkit + ControlNet + how to use it overal hopefully soon
Windows_Install_SwarmUI.bat and Windows_Update_SwarmUI.bat now includes
ControlNet Preprocessor node of SwarmUI so when you run Update or make a fresh install it will auto get installed - this has so many different ControlNet processorsAlternatively on the interface you can use install button
This is included in RunPod installer as well
For MassedCompute SECourses image i emailed them to auto include hopefully
To be able to use newest ControlNet run Windows_Install_Or_Update_ComfyUI.bat first in your ComfyUI install and then use Windows_Update_SwarmUI.bat and then download the new model
1 December 2025 V107
Now when there is a SHA256 cache mismatch it will get latest SHA256 and fix the issue automatically and if needed it will redownload model
Wan2.2-I2V-A14B-Moe-Distill-Lightx2v-High_fp8_scaled.safetensors updated with better newer version - just redownload
2 New Wan 2.2 Text to Image Lightning LoRAs added
They really improving quality really cool improvement for free
They are added under Wan 2.2 LoRAs and Wan 2.2 Core 8 Steps Bundle
Wan2.2-T2V-A14B-4steps-lora-rank64-Seko-V2.0-Low.safetensors (1.14 GB)
Wan2.2-T2V-A14B-4steps-lora-rank64-Seko-V2.0-High.safetensors (1.14 GB)
So download them and then import new Amazing_SwarmUI_Presets_v36.json either manually or use Windows_Preset_Delete_Import.bat (deletes all and makes clean preset install recommended)
Then use Wan 22 Text To Video 8 Steps for these new LoRAs having preset
Currently this is the very best Lightning (4-steps) Wan 2.2 Text to Video workflow until a better one arrives hopefully
Also Z-Image Turbo model with Ostris AI Toolkit best workflow research is almost done will publish hopefully tomorrow
28 November 2025 V105
FLUX 2 presets added into Amazing_SwarmUI_Presets_v35.json
Z Image Turbo presets added into Amazing_SwarmUI_Presets_v35.json
Downloader app completey revamped and extremely improved
The initial loading of the app will take a while but after that you will see it is amazing
The model downloader app version increased to Gradio 6.0.1 with latest features
FLUX 2 models download added into downloader app as FLUX 2 Models and FLUX 2 Core Bundle
FLUX 2 models added into Complete Image Generation and Editing Bundle as well
Z Image Turbo models download added into downloader app as Z Image Turbo Models and Z Image Turbo Core Bundle
I also set the default resolution of the models as 1536x1536 pixels
I have added Z Image FP8 Convert into our SECourses Premium Musubi Tuner app and thus added FP8 Scaled version of the model as well into our downloader app
First time done by me so only SECourses followers getting it
FP8 Scaled version quality almost same as BF16 and GGUF Q8 but much faster than GGUF Q8 with only 6 GB size so lower VRAM GPUs can use it
You can see full size screenshot of the app here
SDXL Realism preset updated / upgraded
18 November 2025 V101
Wan 2.2 T2V High Noise 14B BF16 (26.61 GB) and Wan 2.2 T2V Low Noise 14B BF16 (26.61 GB) added to the Wan models list
BF16 works higher quality than FP16 in most cases
3 New Qwen Image Edit LoRAs added
Check their description to see example prompts
Qwen-Image-Edit-2509-Multiple-Angles-LoRA.safetensors (0.22 GB)
Qwen-Image-Edit-2509-Relight-LoRA.safetensors (0.22 GB)
Qwen-Image-Edit-2509-Fusion-LoRA.safetensors (0.22 GB)
BiRefNet HR (High Resolution) Background Remover https://www.patreon.com/posts/121679760
14 November 2025 V99
New SwarmUI bundle : Trained Models Min Requirements (Total: 31.13 GB, 11 models)
This model downloads all automatically downloaded models by SwarmUI for FLUX, Qwen, Wan, HiDream models including 2 LoRAs we use for Qwen
Super useful to use on RunPod and Massed Compute or at fresh installs
10 November 2025 V97
2 New Qwen Image models Realism LoRAs added to the model downloader
They are included in both Image Generation and Editing Bundle and Qwen Image Core Bundle
You can find them individually under Qwen Image Models
- Qwen_LoRA_Amateur_Photo_v1.safetensors and - Qwen_LoRA_Skin_Fix_v2.safetensors
Presets updated and 4 new presets added
Inpainting Preset - Extremely useful to inpaint your existing images and fix errors, improve face etc
Follow this tutorial video to understand how it works : Link will be added hopefully
Qwen Image UHD UlRealism 4+4 Steps : This is ultra realistic but works only for real life like images, so if you are generating an image with riding a Dragon won't work
You can increase/decrease selected LoRA weight according to your needs
Qwen Image UHD High Realism 4+4 Steps : This is a very balanced LoRA that brings both realism + keeps amazing capability of the model
Currently LoRA is set to 0.6 weight for self trained models
If you use on raw Qwen Model, make weight like 1.2 makes really good realism improvement
You can increase/decrease selected LoRA weight according to your needs
Outpainting Preset FLUX Dev Fill : This preset used FLUX Dev Fill model to outpaint images as shown in tutorial
FLUX DEV Fill model is now included in Image Generation and Editing Bundle since it is useful to fix mistakes or like remove text on existing images
Follow this tutorial video to understand how it works : Link will be added hopefully
30 October 2025 V96
Expired download token refreshed
New 2 presets added
Qwen Images Stylized UHD Tier 1 and Qwen Images Stylized UHD Tier 2
They are useful when you generate stylized images, style images after training
22 October 2025 V92
Presets and bundle downloads updated for Qwen Image Realism
Now on your self trained Qwen Image Models realism is next level here below few examples
Import latest v29 preset file
You can use below presets for such realism
Qwen Image UHD Realism Tier 1 - 8+8 Steps
Qwen Image UHD Realism Tier 2 - 4+4 Steps
Qwen Image Edit Plus UHD Realism - 4+4 Steps
This is for training your subject on Qwen Imaged Edit Plus 2509 model with pure black control images - our Musubi Tuner already have this feature to generate such black control images
17 October 2025 V90
Missing LoRA added to bundles
2 LoRAs were downloaded into diffusion_models folder inaccurately and fixed this issue
Few downloaded model names fixed to match presets
Qwen_Image_FP8_Scaled.safetensors download error fixed
There is now 2 presets
Wan 22 Image To Video 4 Steps HQ
Wan 22 Image To Video 8 Steps UHQ
I prefer 8 steps but 4 is also great
15 October 2025 V86
Image Generation and Editing Bundle and Qwen Image Core Bundle updated
Qwen Image Edit Plus 2509 is now FP8_Scaled instead of GGUF Q8
Advantage is up to 1.5x more speed on modern GPUs, same quality, lesser VRAM
I did a converter script to convert this model into scaled FP8 and it took huge time
Scaled FP8 is many times more quality than just base FP8
Qwen Image Model updated to Qwen Image FP8 Scaled from GGUF Q8
New Wan 2.2 Image to Video base models added to the Wan 2.2 Core 8 Steps Bundle
I have converted these new models into FP8 Scaled myself
Their names are Wan2.2-I2V-A14B-Moe-Distill-Lightx2v-Low_fp8_scaled and Wan2.2-I2V-A14B-Moe-Distill-Lightx2v-High_fp8_scaled
These models are working without LoRAs with just 4 steps - and quality is amazing
They were published today :)
According to these new models the following presets updated
Qwen Image 50 Steps Official Slow
Qwen Image 8 Realism Fast
Qwen Image 8 Steps Ultra Fast
Qwen Image Edit Plus 12 Steps
Qwen Image Edit Plus 50 Steps
Qwen Image Realism Fast
New preset Wan 22 Image To Video 4 Steps HQ added and I recommend this for image to video generation now - best one and only 4 steps ultra fast
So download these updated bundles, don't worry it won't redownload existing models just new ones
Run Download Qwen Image Core Bundle and Download Wan 2.2 Core 8 Steps Bundle to new models
13 October 2025 V84
This is a major update for presets
With ComfyUI installer v56 now we have the missing new Samplers and Schedulers inside SwarmUI such as beta57, bong_tangent, res_2s : https://www.patreon.com/posts/105023709
So make sure to get latest ComfyUI installer zip file and run installer 1 time to get this update
I have updated images of all presets so that it will be way easier to recognize presets
New preset Qwen Image 8 Realism Fast added and it is really really much more realistic
I will hopefully update Qwen LoRA training post and add realistic results soon : https://www.patreon.com/posts/137551634
Of course it is realistic for other LoRAs or no LoRAs too
I did a massive research on Wan 2.2 image generation and we completey remade Wan 22 Image Realism preset
Now it is really really realistic look at this example : https://huggingface.co/MonsterMMORPG/Generative-AI/resolve/main/Wan_2_2.png
This also had our 2x upscale preset applied
For new Wan 2.2 and Qwen Image Realism presets to work you have to update your ComfyUI installation as mentioned above
For this update just import Amazing_SwarmUI_Presets_v24.json and overwrite older ones
26 September 2026 V82
Amazing ULR Downloader implemented into the app that lets you download from CivitAI and Hugging Face into relative or absolute path with custom file name support
It uses 16 connections and you can blazing fast download from even CivitAI
25 September 2026 V81
Qwen Image Lightning 8steps-V2 LoRA added to the downloader
This is a significant quality improvement compared to V1 and presets updated to use this LoRA
The bundles are also updated to download this new LoRA
FLUX SRPO models added to the downloader
Both BF16 and GGUF are available
FLUX SRPO is extremely realistic model
FLUX Bundle will now download FLUX SRPO BF16 model as well
I have uploaded a metadata fixed, so it will be auto set for you
Fixed metadata of FLUX Kontext Dev model as well with re-upload
Always check metadata of models in SwarmUI and verify accurate
FLUX Dev Models preset info and thumbnail image updated
FLUX Krea Dev Official removed since it is identical of FLUX Dev Models preset
FLUX Dev Models preset supports all FLUX Dev, FLUX Krea and FLUX SRPO
Just change base model
FLUX Kontext Dev Edit Images preset info, thumbnail image and settings updated
Qwen Image Edit Plus models added to the downloader and older Qwen Image Edit model replaced with Plus model in bundle (Qwen Image Edit 2509)
Presets are also updated to use this model
I have personally fixed metadata of the BF16 and FP8 version of the model
For GGUF again make sure that SwarmUI sees as Qwen Image Edit Plus the architecture
I added a json file for bundle downloaded GGUF Q8 so it will auto recognize it
Qwen Image Edit plus supports up to 3 input images so no more image sticthing is necessary and this model has huge quality
We are using Qwen Image Lightning 8steps-V2 LoRA with this model as well
If any better speed up LoRA arrives I will hopefully update
GGUF models are way slower compared to BF16 models on many hardware so if your VRAM is sufficient, prefer BF16 like on RTX A6000 GPUs
If your VRAM is not sufficient still you can compare speed of BF16 vs GGUF and see which one works best on your PC
ComfyUI auto does block swapping so you can even run BF16 on 24 GB GPUs
Make sure you have lots of RAM otherwise block swap will fail and your PC will freeze or use lower quant Like GGUF_8 or GGUF_5 etc
Under Other Models (e.g. Yolo Face Segment, Image Upscaling) section, Download All Auto Yolo Masking/Segment Models updated
Now it has Yolo V12 Large model as well : yolov12l-face.pt
Previously this was not working but after I opened an issue on SwarmUI and we did debug, it is fixed and now working
Face segment bundle is included in Qwen Image Core Bundle, FLUX Models Bundle and HiDream-I1 Dev Bundle
Example usage : photo of a man <segment:yolo-yolov12l-face.pt> a man
Model downloader completeyt refactored for future easier development
Some of the files moved inside utilities folder and content of zip file simplified
Now we are not using Hugging Face downloader anymore and we are using much more robust uGet style 16 connection downloader that I developed
The advantage of this approach is that, it is extremely robust and works perfect even on low speed connections
It has 100% resume capability even if at stops at 99.99% percent
It auto checks SHA256 to verify accuracy of the download so the model is never corrupted
Sometimes you may get errors like on this on SwarmUI : https://github.com/mcmonkeyprojects/SwarmUI/issues/1065
On that case close SwarmUI, run Manually_Rebuild_SwarmUI_On_Errors.bat and restart SwarmUI
This will SwarmUI to compile necessary libraries on next run again
Use Windows_Preset_Delete_Import.bat to delete your all previous presets and import new updated ones : Amazing_SwarmUI_Presets_v22.json
Hopefully I will look for newer Wan 2.2 and Wan 2.1 speed up LoRAs to update presets even better quality
I recommend using 12 Steps Qwen Image Edit Plus and 8 Steps Qwen Image presets rather than 50 steps ones
Tutorials
Main ComfyUI + SwarmUI setup tutorial (4 May 2025) : https://youtu.be/fTzlQ0tjxj0
Wan 2.1 Tutorial shows how to use LoRA as well (19 May 2025) : https://youtu.be/XNcn845UXdw
ComfyUI + SwarmUI RunPod master tutorial (10 June 2025) : https://youtu.be/R02kPf9Y3_w
SwarmUI Wan 2.1 FusionX + Image Upscale tutorial (18 June 2025) : https://youtu.be/Xbn93GRQKsQ
SwarmUI FLUX Kontext (Ultimate image editing with just prompts) tutorial (27 June 2025) : https://youtu.be/adF9X9E0Chs
MultiTalk (image + audio = full animated lip synched video) main tutorial (ComfyUI only) (10 July 2025) : https://youtu.be/8cMIwS9qo4M
MultiTalk new workflows tutorial (ComfyUI only) (12 July 2025) : https://youtu.be/wgCtUeog41g
Wan 2.2 & FLUX Krea Full Tutorial - Automated Install - Ready Perfect Presets - SwarmUI with ComfyUI (2 August 2025) : https://youtu.be/8MvvuX4YPeo
Qwen Image Dominates Text-to-Image: 700+ Tests Reveal Why It's Better Than FLUX - Presets Published (7 August 2025) : https://youtu.be/R6h02YY6gUs
Wan 2.2, FLUX & Qwen Image Upgraded: Ultimate Tutorial for Open Source SOTA Image & Video Gen Models (19 August 2025) : https://youtu.be/3BFDcO2Ysu4
Qwen Image Edit Full Tutorial: 26 Different Demo Cases, Prompts & Images, Pwns FLUX Kontext Dev (23 August 2025) : https://youtu.be/gLCMhbsICEQ
Windows Requirements
Python 3.10, FFmpeg, CUDA 12.8, cuDNN 9.7 or above, C++ tools and Git
If it doesn't work make sure to below tutorial and install everything exactly as shown in this below tutorial
Requirements are important to have CUDA, C++ Tools, MSVC
Massed Compute (Recommend Cloud) :
Please register via this link : https://vm.massedcompute.com/signup?linkId=lp_034338&sourceId=secourses&tenantId=massed-compute
Use our coupon SECourses
Our coupon works on all GPUs now
H100 has amazing price and speed but you can use like RTX A6000 ADA as well
Full details here : https://www.patreon.com/posts/26671823
Then select our image SECourses from Creator dropdown
Then follow Massed_Compute_Instructions_READ.txt
Same as my any other Massed Compute installer script
Example tutorial for learn how to install and use Massed Compute
(Starts at 12:58) : https://youtu.be/KW-MHmoNcqo?si=G1WbG-Qw4ujWvOtG&t=778
RunPod (Cloud):
Please register via this link : https://runpod.io?ref=1aka98lq
Then follow Runpod_Instructions_READ.txt
Same as my any other RunPod installer script
Use the template written in Runpod_Instructions_READ.txt file
Example tutorial for learn how to install and use RunPod
(starts at 22:03) : https://youtu.be/KW-MHmoNcqo?si=QN8X8Sjn13ZYu-EU&t=1323
Total Models size you can download is over 3200 GB atm
OLDER VERSION HISTORY
12 September 2025 V80
The unified AI models downloader now supports newest Stable Diffusion WebUI Forge - Classic : https://www.patreon.com/posts/138680643
The unified AI models downloader now supports newest Stable Diffusion WebUI Forge - NEO : https://www.patreon.com/posts/138694680
Just check the checkbox for these 2 apps:
Forge WebUI / Automatic1111 Folder Structure
Example path : E:\Forge_Neo_v1\sd-webui-forge-classic\models
27 August 2025 V79
Demo images zip file upgraded to V3 since we published Nano Banana free image editing tutorial and compared it with Qwen Image Edit model for every case
Full tutorial link : https://youtu.be/qPUreQxB8zQ
A LoRA selection error of Qwen Image Edit model preset fixed with Amazing_SwarmUI_Presets_v21.json
22 August 2025 V77
Image upscale preset unified into single one
First use your regular preset then plus apply it
Prompt file modified to FLUX_Kontext_Qwen_Edit_Prompts.txt
Contains 26 unique cases and total 40 prompts
Demo images and results added, results have metadata so you can see and replicate
If Sage Attention on vs off, produces different results
Hugging Face download errors fixed
20 August 2025 V75
Use latest zip file extract and overwrite
Don't forget to edit metadata of the Qwen Image Edit model and set architecture of the model as Qwen Image Edit since it is not auto recognized yet
I have added 2 new presets called as Qwen Image Edit 50 Steps and Qwen Image Fast 12 Steps
Qwen Image Fast 12 Steps takes 30 seconds on RTX 5090
Qwen Image Edit 50 Steps takes 128 seconds on RTX 5090
Here a comparison of org image vs 12 steps vs 50 steps : https://imgsli.com/NDA4MTU1
Prompt is : change hair color to blue
The presets are using Qwen Image Edit model GGUF Q8 but you can use lower GGUF too
Our downloader has BF16, FP8, GGUF Q8, Q6, Q5, Q4
First update your ComfyUI and SwarmUI to latest version as shown in this tutorial and import presets and apply them : https://youtu.be/3BFDcO2Ysu4
This model doesn't work with input image but rather it takes input image from prompt like in FLUX Kontext : https://youtu.be/adF9X9E0Chs
This model doesn't support inpainting
Hopefully I will make a new tutorial for this but even ComfyUI official workflow is so bad atm - our preset way better
So we need some more time for this model to become better
Our Windows_Start_Download_Models_App.bat now have Qwen Image Edit model and it is also included in SwarmUI Qwen Image core bundle
SwarmUI update file will now auto install RIFE frame interpolation and Teacache
So just download bundle to download this model - it won't redownload existing models
19 August 2025 V72
Use latest zip file extract and overwrite
Full how to use tutorial published : https://youtu.be/3BFDcO2Ysu4
New following presets added and now we have Windows_Preset_Delete_Import.bat file
Execute this file while SwarmUI running, it will backup your existing presets into presets_backups folder, then delete all your presets and import newest presets
The added presets are
Qwen Image 8 Steps Ultra Fast 2x Upscale : Really upscales amazingly Qwen Image generations and fast too
You can increase steps count to 20 if not good enough
Wan 22 High Quality I2V 20 Steps : Super high quality but slow
Wan 22 High Quality T2V 20 Steps: Super high quality but slow
Wan 22 Image To Video 8 Steps: Really decent quality and ultra fast
Wan 22 Text To Video 8 Steps: Really decent quality and ultra fast
Wan 22 Image Realism: Super high quality and fast
This preset generates static images unlike videos
Wan 2.2 Image Generation Test Grids
Preparing these very best presets literally took days and 100s of generations and comparisons
To not delete your older presets, use import and import v15 and overwrite existing, it will only overwrite same name presets
On RunPod and Massed Compute use regular import preset feature same as before
Also use model downloader file and use Wan 2.2 Core 8 Steps Bundle (Total: 70.41 GB, 11 models) to download new 4 LoRAs
It will download new lightx2v Wan2.2-Lightning LoRAs
Some bundle downloads were broken for some LoRAs and these errors fixed
Windows SwarmUI installer will now auto install RIFE Frame Interpolation and Teacache automatically
13 August 2025 V66
Download newest v66 zip file, extract and overwrite previous files as usual
New amazing Qwen Image preset added : Qwen Image 8 Steps Ultra Fast
Literally 6x faster than Qwen Image High Quality preset
To use this preset use latest models downloader and download Qwen Image Core Bundle (Total: 30.83 GB, 4 models) or Qwen Image Lightning 8steps V1.1 LoRA (Fast inference LoRA) (1.58 GB)
If you had downloaded older bundle before, it will just download new LoRA
Moreover, import new preset json file and overwrite : Amazing_SwarmUI_Presets_v12.json
All Preset images updated with Qwen Image 8 Steps Ultra Fast preset - so now all preset thumbnails have text to easier recognize
All presets will automatically select accurate models based on our model downloader
I have tested all presets and verified all works
Sage Attention is now working with Qwen Image again - make sure to update your ComfyUI and SwarmUI to the latest version
FLUX Dev Official 2x Latent Upscale preset works with FLUX Krea Dev model as well just change base model
I also tested all best schedulers for FLUX Dev model and updated presets for best one
You can see grid comparison here : click to download
I also have tested new Scheduler named as KL Optimal (Nvidia AYS) with all samplers and it produces some realistic and
I think our very best preset still overall better but it is a good one to test in some cases
You can see its comparison grid here : click to download
Hopefully I will update Wan 2.2 presets with new LoRAs after testing and verifying they are better
8 August 2025 V63
Now supports Automatic1111 Web UI and SD Forge Web UI model structure too
Just select Forge WebUI / Automatic1111 Folder Structure checkbox and give model path like below
Windows e.g. : E:\Forge_Installer_v10\stable-diffusion-webui-forge\models
Massed Compute e.g. : /home/Ubuntu/Downloads/Forge_Installer_v10/stable-diffusion-webui-forge/models/,
RunPod e.g. : /workspace/stable-diffusion-webui-forge/models
New tutorial published : Qwen Image Dominates Text-to-Image: 700+ Tests Reveal Why It's Better Than FLUX - Presets Published > https://youtu.be/R6h02YY6gUs
SD Forge Web UI Installers updated for Windows, RunPod and Massed Compute and now supports RTX 5000 series too and working perfect with more robust, easier and faster install : https://www.patreon.com/posts/118442039
The error we were getting when downloading upscale models and face restoration models fixed : exists and seems populated. Skipping snapshot download for
6 August 2025 why Qwen Image is the New King and the Research
You can download all the grid tests (700+ generations) i did for Qwen Image inference parameter research here : https://huggingface.co/BestModelsv2/test/resolve/main/qwen_image_inference_research_grids.zip
Put the grids into this folder SwarmUI\Output\local\Grids and restart SwarmUI then you will be able to load and see each grid, tested parameters and results in full quality
Moroever I have compared FLUX Dev Official vs FLUX Krea Dev Official vs Qwen Image Realism Fast vs Qwen Image High Quality presets and you can see full quality result grid image here : https://huggingface.co/MonsterMMORPG/Generative-AI/resolve/main/FLUX_vs_Krea_vs_Qwen.png
Hopefully I will fully research Qwen Image training as i did for FLUX and then we will have a new amazing training workflow :
6 August 2025 V62
I have generated 600 images to find out the very best presets of Qwen Image model and now 2 new presets added to our presets with Amazing_SwarmUI_Presets_v9 file inside zip file
We have Qwen Image High Quality which generates highest quality overal with maximum prompt following - default negative prompt also set which improves quality further
And there is Qwen Image Realism Fast which is 2x faster since cfg is set to 1
This preset may require more generation to get best result but definitely better realism
Moreover downloader app now have the following models to download with 1 click
Under Image Generation Models under Qwen Image Models
Qwen_Image_Q4_1 (11.96 GB) - Qwen_Image_Q5_1 (14.33 GB) - Qwen_Image_Q6_K (15.67 GB) - Qwen_Image_Q8_0 (20.27 GB) - Qwen_Image_FP8_e4m3f (19.03 GB) - Qwen_Image_BF16 (38.05 GB)
Just use SwarmUI Bundles and then Qwen Image Core Bundle (Total: 29.24 GB, 3 models) and the presets will work right away - uses Q8
Or manually select your model
Q4_1 quality is also amazing so if you are on low VRAM you can use it
Still all models should work if you have sufficient RAM since SwarmUI uses ComfyUI which does auto block swapping
Hopefully tomorrow I will share all grids so that you can download and analyze yourself locally with SwarmUI
Sage attention may cause black output so until fixed disable if you get
AllowGpuSpecificOptimizations inside Server > Server Configuration may cause black output so disable until it is fixed
Update your both ComfyUI and SwarmUI to the latest version
This is clearly the new King of models better than FLUX and hopefully I will cover training fully soon after full research
Image resolution has to be divisible to 16 at the moment so if you get error pay attention to that
2 August 2025 V61
Wan 2.2 + FLUX Krea full tutorial published > https://youtu.be/8MvvuX4YPeo
After carefully testing more, bad results yielding presets removed
Thus I recommend delete all previous presets and import new Amazing_SwarmUI_Presets_v8 - screenshot
I also removed the following bundles since they were not necessary anymore with recent LoRAs + base Wan models
Wan 2.1 FusionX FP16 Phantom Bundle
Wan 2.1 FusionX FP8 Phantom Bundle
Moreover, all presets will automatically select models downloaded from our downloader app bundles now so that you won't be needed to manually select presets
The bundles you should download for SwarmUI are as below
Wan 2.2 Core 8 Steps Bundle (Total: 65.84 GB, 7 models) - FP8 Scaled base models + LoRA
Base models can be replaced with other variants like GGUF Q8 or FP16 etc
Wan 2.1 Core Models Bundle (GGUF Q6_K + Best LoRAs) (Total: 36.89 GB, 10 models) - GGUF Q6 base models + LoRA
Base models can be replaced with other variants like GGUF Q8 or FP8 etc
FLUX Models Bundle (Total: 100.77 GB, 10 models)
Base models can be replaced with other variants like GGUF Q8 or FP8 etc
All models still remaining so just search the model name and download whichever you want
31 July 2025 V60
Amazing new official FLUX Krea DEV model published
I have added both FP16 and GGUF Q8, Q6, Q5 and Q4 models
Q8 almost same quality as FP16
It is way more realistic here first comparison test I made : https://www.reddit.com/r/SECourses/comments/1me4jkb/flux_krea_dev_is_really_realistic_improvement/
The download button is under Image Generation Models▼ FLUX Models▼ - at the very top
FLUX Models is now seperated into normal Models vs GGUF variants as 2 tabs
Search function significantly improved, just search krea and test
New Wan 2.2 text-to-video high quality preset added
Sadly none of the fast LoRAs of Wan 2.1 working for Wan 2.2 text-to-video greatly so it is best to not use lora and do at least 20 steps until a LoRA is published
Image to video works perfect with our Wan 2.2 bundle LoRA and preset and 8 steps
31 July 2025 V58
Both ComfyUI and SwarmUI zip files updated
Now they are more robust to ensure updates goes smooth
Make sure update your ComfyUI backend and SwarmUI
After doing literally 100s of grid Wan 2.2 generations I have prepared amazing 8 steps 2 presets
These presets are even selecting the accurate models and LoRA
First download new SwarmUI bundle : Wan 2.2 Core 8 Steps Bundle (Total: 65.84 GB, 7 models) - use Windows_Start_Download_Models_App.bat from latest zip file
Then import new preset Amazing_SwarmUI_Presets_v6.json
Then reset params to default and direct apply preset you want to use
For text to video model just type your prompt and also set your desired resolution and set your desired Text2Video Frames count - default 73 frames thus 3 seconds (24 fps), for 5 seconds video set 121 frames
For image to video select your input image, type your prompt, set target resolution according to image aspect ratio and set your Video Frames under Image to Video tab - default 73 frames thus 3 seconds (24 fps), for 5 seconds video set 121 frames
Thats all and hit generate and enjoy
I am recording a tutorial video for Wan 2.2 hopefully
Moreover we are using UniPC sampler and Eular Ancestral behaves significantly different you can compare both
I am also using --use-sage-attention which brings good speed up and working great
Also we have Video_Models_Prompt_Generate_Guide.txt file
You can use this file in free Gemini in Google Studio ai to generate amazing prompts, just upload file and write what you want to generate : https://aistudio.google.com/prompts/new_chat (free)
16 July 2025 V57
New 4-8 steps LoRA added for Wan 2.1 models
Wan 2.1 14B LightX2V CFG Step Distill LoRA V2 (T2V + I2V) (Rank 64) (0.69 GB)
This LoRA will be now downloaded with bundles
Works great for both image to video and text to video with Wan 2.1 Models
CausVid LoRA preset renamed to LightX2V
Do 4-10 steps, 8 is probably best quality / speed
Related posts
Wan 2.2 & FLUX Krea Full Tutorial: 1-Click Install & SwarmUI Presets
ComfyUI & SwarmUI on Massed Compute and RunPod: Cloud GPU Tutorial
Wan 2.2, FLUX Krea & Qwen Image Upgraded: Fast 8-Step Presets Tutorial
All posts in this topic: SwarmUI: 1-Click Installer, Model Downloader and Tutorials · New to SECourses? Start Here: Best and Most Popular SECourses Posts