Nvidia 50 Series (Blackwell) support thread: How to get ComfyUI running on your new 50 series GPU.

[comfyanonymous]

I will try keeping this post up to date as much as possible with the latest developments.

To get your nvidia 50 series GPU working with ComfyUI you need a pytorch that has been built against cuda 12.8

In the next few months there will likely be a lot of performance improvements landing in pytorch for these GPUs so I recommend coming back to this page and updating frequently.

Windows

The recommended download is the latest standalone portable package or desktop installer that you can download from the README

Manual Install

If you install stable pytorch make sure it is cu128.

pytorch nightly cu128 is available for Windows and Linux:

pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu128

You can also use the Nvidia Pytorch Docker Container as an alternative which might give more performance.

Link: https://catalog.ngc.nvidia.com/orgs/nvidia/containers/pytorch

Here's how to use it:

docker run -p 8188:8188 --gpus all -it --rm nvcr.io/nvidia/pytorch:25.01-py3

Inside the docker container:

git clone https://github.com/comfyanonymous/ComfyUI
cd ComfyUI
grep -v 'torchaudio\|torchvision' requirements.txt > temp_requirements.txt
pip install -r temp_requirements.txt
python main.py --listen

[mcmonkey4eva]

There are also Windows prebuilt wheels of pytorch on cuda 128 that nvidia gave w-e-w to publish https://huggingface.co/w-e-w/torch-2.6.0-cu128.nv

[comfyanonymous]

Yeah I see the package but it's still missing the cu128 torchvision, torchaudio and windows packages.

[mmartial]

It is. Thanks for clarifying the components we are waiting for.

I was trying to follow the guide by removing torchvision and torchaudio
and doing a pip3 install --pre torch --index-url https://download.pytorch.org/whl/nightly/cu128

spandrel too uses torchvision, which forces a downgrade on a Linux installation within a virtualenv.
https://github.com/chaiNNer-org/spandrel/blob/main/libs/spandrel/pyproject.toml#L36

EDIT: I will postpone testing until this is ready. Thanks for the details

[comfyanonymous]

I just noticed there are now nightly torchvision and torchaudio cu128 wheels for Linux. No windows packages yet though.

I tried them and they seem to work well so I have updated the post.

[mmartial]

Thank you 👍

With this news, I have made an updated release of https://github.com/mmartial/ComfyUI-Nvidia-Docker

For CUDA 12.8, I pre-install torch using pip3 install --pre torch torchaudio torchvision --index-url https://download.pytorch.org/whl/nightly/cu128 before installing ComfyUI from git and installing its requirements. I have tested it and can generate the bottle :)

No 50xx yet for me, so this test was done on a 3090 with driver 570 installed.

[diomsidney]

@comfyanonymous your ComfyUI app has to be one of the best all in one solutions. Great job creating a solution for Consumer Devices. @microsoft would be interested in purchasing your project. But congratulations are in order. It works a charm.

[edisonchan]

The Windows version said:
ModuleNotFoundError: No module named 'torchaudio'

[comfyanonymous]

This is because there was no torchaudio cu128 package provided.

The only thing this affects is the audio related nodes so it only affects you if you are using stable audio.

[edisonchan]

nf4 node does not work.

[ltdrdata]

Anything related to NF4 support should be resolved in the node pack repo that provides NF4 feature.

[edisonchan]

ImportError: tokenizers>=0.21,<0.22 is required for a normal functioning of this module, but found tokenizers==0.20.3.
Try: pip install transformers -U or pip install -e '.[dev]' if you're working with git main

How can I do to fix problem?

[adamreading]

Sorry - I have been banging my head against this for 6 hours - trying to run a cosmos workflow with 5090 - without SageAttention or TorchCompile - a 14 minute job on 4090 is taking 25 minutes. To turn them on I installed Triton and Sageattention - then I get this error afterwards whether I bypass the patch and compile or not:- KSampler
Command '['C:\Program Files (x86)\Microsoft Visual Studio\2022\BuildTools\VC\Tools\MSVC\14.42.34433\bin\Hostx64\x64\cl.EXE', 'C:\Users\JOANNA1\AppData\Local\Temp\tmpy8t3476k\cuda_utils.c', '/nologo', '/O2', '/LD', '/wd4819', '/IC:\Users\JoannaThompson\ComfyUI\ComfyUI_cu128_50XX\python_embeded\Lib\site-packages\triton\backends\nvidia\include', '/IC:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.8\include', '/IC:\Users\JOANNA1\AppData\Local\Temp\tmpy8t3476k', '/IC:\Users\JoannaThompson\ComfyUI\ComfyUI_cu128_50XX\python_embeded\Include', '/IC:\Program Files (x86)\Microsoft Visual Studio\2022\BuildTools\VC\Tools\MSVC\14.42.34433\include', '/IC:\Program Files (x86)\Windows Kits\10\Include\10.0.26100.0\shared', '/IC:\Program Files (x86)\Windows Kits\10\Include\10.0.26100.0\ucrt', '/IC:\Program Files (x86)\Windows Kits\10\Include\10.0.26100.0\um', '/FoC:\Users\JOANNA1\AppData\Local\Temp\tmpy8t3476k\cuda_utils.cp312-win_amd64.obj', '/link', '/LIBPATH:C:\Users\JoannaThompson\ComfyUI\ComfyUI_cu128_50XX\python_embeded\Lib\site-packages\triton\backends\nvidia\lib', '/LIBPATH:C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.8\lib\x64', '/LIBPATH:C:\Program Files (x86)\Microsoft Visual Studio\2022\BuildTools\VC\Tools\MSVC\14.42.34433\lib\x64', '/LIBPATH:C:\Program Files (x86)\Windows Kits\10\Lib\10.0.26100.0\ucrt\x64', '/LIBPATH:C:\Program Files (x86)\Windows Kits\10\Lib\10.0.26100.0\um\x64', '/LIBPATH:C:\Program Files (x86)\Microsoft Visual Studio\2022\BuildTools\VC\Tools\MSVC\14.42.34433\lib\x64', '/LIBPATH:C:\Program Files (x86)\Windows Kits\10\Lib\10.0.26100.0\ucrt\x64', '/LIBPATH:C:\Program Files (x86)\Windows Kits\10\Lib\10.0.26100.0\um\x64', '/LIBPATH:C:\Program Files (x86)\Microsoft Visual Studio\2022\BuildTools\VC\Tools\MSVC\14.42.34433\lib\x64', '/LIBPATH:C:\Program Files (x86)\Windows Kits\10\Lib\10.0.26100.0\ucrt\x64', '/LIBPATH:C:\Program Files (x86)\Windows Kits\10\Lib\10.0.26100.0\um\x64', '/LIBPATH:C:\Program Files (x86)\Microsoft Visual Studio\2022\BuildTools\VC\Tools\MSVC\14.42.34433\lib\x64', '/LIBPATH:C:\Program Files (x86)\Windows Kits\10\Lib\10.0.26100.0\ucrt\x64', '/LIBPATH:C:\Program Files (x86)\Windows Kits\10\Lib\10.0.26100.0\um\x64', '/LIBPATH:C:\Program Files (x86)\Microsoft Visual Studio\2022\BuildTools\VC\Tools\MSVC\14.42.34433\lib\x64', '/LIBPATH:C:\Program Files (x86)\Windows Kits\10\Lib\10.0.26100.0\ucrt\x64', '/LIBPATH:C:\Program Files (x86)\Windows Kits\10\Lib\10.0.26100.0\um\x64', '/LIBPATH:C:\Program Files (x86)\Microsoft Visual Studio\2022\BuildTools\VC\Tools\MSVC\14.42.34433\lib\x64', '/LIBPATH:C:\Program Files (x86)\Windows Kits\10\Lib\10.0.26100.0\ucrt\x64', '/LIBPATH:C:\Program Files (x86)\Windows Kits\10\Lib\10.0.26100.0\um\x64', 'cuda.lib', '/OUT:C:\Users\JOANNA1\AppData\Local\Temp\tmpy8t3476k\cuda_utils.cp312-win_amd64.pyd', '/IMPLIB:C:\Users\JOANNA1\AppData\Local\Temp\tmpy8t3476k\cuda_utils.cp312-win_amd64.lib', '/PDB:C:\Users\JOANNA1\AppData\Local\Temp\tmpy8t3476k\cuda_utils.cp312-win_amd64.pdb']' returned non-zero exit status 2.

I have run this through GPT and followed every instruction. Step 1: Verify CUDA Toolkit Installation
Run the following command in ComfyUI's embedded Python:

python
Copy
Edit
import torch
print(torch.version.cuda)
Expected Output: 12.8

If it does not match 12.8, you may need to reinstall PyTorch with the correct CUDA version.

✅ Step 2: Check NVIDIA Toolkit and Drivers
Ensure you have the latest NVIDIA GPU driver and CUDA toolkit installed.

Update your NVIDIA driver (Download latest)
Confirm CUDA installation with:
powershell
Copy
Edit
nvcc --version
Expected Output: release 12.8
If not, reinstall CUDA 12.8 from NVIDIA.

✅ Step 3: Fix Microsoft Visual Studio Build Tools
Open Developer Command Prompt for VS 2022

Run:

sh
Copy
Edit
cl.exe
If it says command not found, reinstall Microsoft C++ Build Tools:
Download: Visual Studio Build Tools 2022
Install:
C++ CMake tools for Windows
MSVC v143 - VS 2022 C++ x64/x86 build tools
Windows 10 SDK (latest version)
Restart the PC.

[yu-jingxiang]

your questiong in the same to me.

[rubentorresbonet]

anyone knows if other pytorch's cuda versions like 12.6 will work with blackwell?

[mpfcs162]

No, Blackwell GPUs require 12.7 or newer IIRC.
Edit: So 12.8 since that is the current version. It's a new architecture so I guess it makes sense that 12.6 and older versions are not supported. CUDA is not backwards compatible?

[fasfafefggwgrw]

When I use [ComfyUI package with a cuda 12.8 torch build], many custom_nodes are "IMPORT FAILED" including Manager, InstantID, ReActor ...

[Shad0w23333]

same, is is possible to build a PyTorch that works for 50s?

[mpfcs162]

Check Python version, reinstall dependencies, and look for incompatibilities in the logs - some nodes might need updates for CUDA 12.8. Fresh install or upgrade?

[dirtmonster1337]

i'm having the same issue, this package uses python 3.13 but ImpImporter was discontinued in these builds, anyone have a fix?

[kat3ri]

if you're running in runpod i made a template that installs everything, including the requirements for a lot of common custom nodes which should help avoid import errors. https://console.runpod.io/deploy?template=0tj9hbexwy&ref=xplhe9v9
takes a little time to install everything the first time but it will work. i turned off xformers since it was causing a lot of errors and not strictly needed for what i was doing.

[Shad0w23333]

I will try keeping this post up to date as much as possible with the latest developments.

To get your nvidia 50 series GPU working with ComfyUI you need a pytorch that has been built against cuda 12.8

In the next few months there will likely be a lot of performance improvements landing in pytorch for these GPUs so I recommend coming back to this page and updating frequently.

Windows

Windows users can download this standalone ComfyUI package with a cuda 12.8 torch build

Manual Install

At this moment Blackwell is not yet supported by stable pytorch.

pytorch nightly cu128 is available for Linux:

pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu128

You can also use the Nvidia Pytorch Docker Container as an alternative which might give more performance.

Is is possible to build a pytorch that works for 5090? If so how to do it?

[adamreading]

What do you mean? There is already one for CUDA 12.8: https://download.pytorch.org/whl/nightly/cu128/torch/ For Windows you can try the EA version found here: https://huggingface.co/w-e-w/torch-2.6.0-cu128.nv Or just use the standalone package linked above with a CUDA 12.8 torch build.

We spent 6 hours + yesterday trying to install comfyui portable around the special version at the top of the screen. It wont let me run Sageattention or Torchcompile in workflows - and because it’s missing TorchAudio - things I use every day like GripTape won’t load into custom nodes any more. Running the cosmos workflow without sage and compile - is twice as long as on a 4090 system (25 mins for 8 sec gen compared to 14 minutes on 4090) Although I was able to install SageAttention and Triton I don’t think they are working with this version of PyTorch

Have you checked the logs for errors or tried a clean environment to rule out conflicts?

Yes we went through the logs - they are in the earlier post above - we tried for 6 hours downloading different versions - Visual build tools, various Nvida toolkits - did everything GPT suggested - its a brand new out of the box machine - nothing previously installed except office.

[adamreading]

5090 Crash 10.txt
One of the complete logs

[mpfcs162]

Tried running Linux? https://download.pytorch.org/whl/nightly/cu128/torch/
To my knowledge there is no official Windows build of PyTorch that supports 12.8 yet.
Sorry I can't offer more help...

[adamreading]

Tried running Linux? https://download.pytorch.org/whl/nightly/cu128/torch/ To my knowledge there is no official Windows build of PyTorch that supports 12.8 yet. Sorry I can't offer more help...

Thanks for looking. The app I want to run (a comfyui-discord bot) requires portable and lm
Studio to work apparently. Plus I use the machine for office365 at the same time lol. Looks like I just have to be patient and work out what will run and what won’t. It ran an ltxv video generator in 36 seconds so some stuff is ok. Couldn’t load anything Pulid.

[hellonearthis]

I got it working with the https://huggingface.co/w-e-w/torch-2.6.0-cu128.nv files.
I turned them into ZIP files which I opened. I then went to the folder ComfyUI.venv\Lib\site-packages and deleted the corresponding folders and extracted the zip folders there.

One of the corresponding files was like ...+cu126.dist-info and I deleted it and plopped in the ...+cu128.nv.dist-info file from the zip.

I'm sure there are better ways to do the but it worked for me.

[adamreading]

Having given up on Portable for now - too many errors lol - I am using WSL - Ubuntu - everything else is set up and working in Comfyui - I have the latest pytorch nightly from pip install --pre torch torchvision torchaudio --index-url[ https://download.pytorch.org/whl/nightly/cu128] - but when I use Sageattention I get this:
loaded completely 29493.675 3667.902587890625 True
0%| | 0/80 [00:00<?, ?it/s]'sm_120' is not a recognized processor for this target (ignoring processor)
'sm_120' is not a recognized processor for this target (ignoring processor)
'sm_120' is not a recognized processor for this target (ignoring processor)
'sm_120' is not a recognized processor for this target (ignoring processor)
'sm_120' is not a recognized processor for this target (ignoring processor)
'sm_120' is not a recognized processor for this target (ignoring processor)
'sm_120' is not a recognized processor for this target (ignoring processor)
'sm_120' is not a recognized processor for this target (ignoring processor)
'sm_120' is not a recognized processor for this target (ignoring processor)
'sm_120' is not a recognized processor for this target (ignoring processor)
'sm_120' is not a recognized processor for this target (ignoring processor)
'sm_120' is not a recognized processor for this target (ignoring processor)
LLVM ERROR: Cannot select: intrinsic %llvm.nvvm.shfl.sync.bfly.i32
Aborted (core dumped)

[mpfcs162]

Wait, can u explain how to do that? Thank u

Do what?

[adamreading]

I managed to get it working in Windows yesterday. Had to manually install the torch-2.6.0+cu128.nv-cp312-cp312-win_amd64.whl and torchvision-0.20.0a0+cu128.nv-cp312-cp312-win_amd64.whl.

Originally had the wrong Python version installed. Easy fix but took me a long time (longer than I want to admit) to figure out...

There are some errors though (torchaudio missing), but I'm running SD 3.5 L to test on a 5090 and I get output. And fast too. But 32GB VRAM is not as much as I'd hoped and the impact compared to a 4090 is not that big - about ~30% faster which is expected I guess...

But I'd honestly recommend just running it on Linux atm.

Have you managed to get SageAttention working yet? (With Linux or Windows)

[mpfcs162]

No. I assume it won't be fully stable until we see an official release of PyTorch with CUDA 12.8 support. Or req some additional manual config. I just can't be bothered with that at this point. Been stuck in integration hell all day trying to troubleshoot some code for a client so haven't had much time to dig further into this topic. Sorry.

[merecesarchviz]

wrong phyton version!? witch version do you install to work? 3.12? I have the same errors that you and i have version 3.11

[Matriv-org]

Hi, Any update about it ?I had a working Sageattention2 and Torchcompile with 4090, but I've the same 'sm_120' is not a recognized processor for this target (ignoring processor) on 5090

SOLVED:
I've WAN2.1 working with torch compile + SageAttention2
In case you had installed Sageattention2 + Triton, you must update triton with
python_embeded\python.exe-m pip install -U --pre triton-windows
It works with torch 2.8.0.dev20250401+cu128, python3.12.9.

Here is the github: https://github.com/woct0rdho/triton-windows

[Teuflor]

im using comfyui over pinokio, is there any way to update my comfy to work with my new 5080?

i deleted the old files and changed in the torch.js on the patch above, but it wont start

[gachapengames]

Will using a docker allow for a working torchvision in windows?

[Six6pounder]

@gachapengames Yes, use nvcr.io/nvidia/pytorch:25.01-py3 as image and it works

[gachapengames]

Can someone write a little guide for getting a docker running torchvision, etc on window with the portable Blackwell comfyui release? Please write it for a normal user who has no programming knowledge.

There are basic instructions in the OP, but what even is a docker?

"docker run -p 8188:8188 --gpus all -it --rm nvcr.io/nvidia/pytorch:25.01-py3"

Where is this command supposed to be run? Is a docker something to be installed to system python or standalone folder?

How would this be installed on a fresh portable comfyui install?

With more Blackwell cards trickling out, there will most likely be more users needing help setting this up. Thank you.

[mmartial]

Please check the "Windows: WSL2 using podman" from https://github.com/mmartial/ComfyUI-Nvidia-Docker

With the caveat that you will need to adapt the podman run line if you want to use the proposed containers to be something like (adapted to attempt to reflect the above order):

podman run --userns=keep-id -p 8188:8188 --device nvidia.com/gpu=all --rm -it nvcr.io/nvidia/pytorch:25.01-py3

with the caveat that everything will be stored within the container since you have no mount (the -v from my original writeup)

so maybe:

podman run --userns=keep-id -p 8188:8188 --device nvidia.com/gpu=all --rm -it -v `pwd`/comfy:/comfy nvcr.io/nvidia/pytorch:25.01-py3

and then do the other steps listed above (pip install...) after cd /comfy so that the files persist when you exit the container

[gachapengames]

To anyone having trouble with the portable version for Blackwell gpus, do not update it! I noticed during updating that it uninstalled the cu128 and installed another version for example. Torchvision works with a fresh install without updating!

[mpfcs162]

So I've used both Comfy UI and Stability Matrix with SD Forge WebUI these past days since getting my 5090.

Each instance needs to have the correct version of Torch, and updating/installing will not ensure that (edit: if anyone reads this in the future there will eventually be an official torch release that supports CUDA 12.8 and in that case what is written here might no longer valid). Best way is to install Torch from the files above, making sure you use the correct version for your system and Python install. For SM, you need to install it to the package directly for example.

And please, do not run commands you find online unless you understand what they do. This applies to the below commands as well.

Navigate to the correct dir and type this to uninstall torch, torchvision and torchaudio:
python_embeded\python.exe -m pip uninstall -y torch torchvision torchaudio

Then, for ComfyUI at least, you can run this. Adjust the command as needed:
python_embeded\python.exe -m pip install H:\torch\torch-2.6.0+cu128.nv-cp310-cp310-win_amd64.whl H:\torch\torchvision-0.20.0a0+cu128.nv-cp310-cp310-win_amd64.whl

You can run this in your dir to check what is installed:
python_embeded\python.exe -m pip list

[mpfcs162]

I just want to add that asking chatGPT (and similar LLMs) about this might not be as helpful as you think.

A very common issue is coherence with these models, and they can easily spew out answers that look and seem correct, while in reality are not what you actually need. For example, if I ask it why I'm getting kernel error when attempting to run stock ComfyUI it will tell me to update using a pip command, but it will give me a link to a build that does not exists (cu127) since it believes that CUDA 12.7 is the version I need for my 5090 since that is the min. version of CUDA that Blackwell supports.

I can follow that trail together with chatGPT for hours and end up trying to compile my own version, uninstall/reinstall various extensions and packages and get absolutely nowhere. And chatGPT will always act and respond with 100% confidence no matter what, and I will never figure out the culprit unless I go to the repo or follow the trail myself to discover that it (cu127) does not exists.

Of course, LLMs are great for a number of reasons - but beware when trying to do stuff like this that the output might be complete nonsense.

[adamreading]

I just want to add that asking chatGPT (and similar LLMs) about this might not be as helpful as you think.

A very common issue is coherence with these models, and they can easily spew out answers that look and seem correct, while in reality are not what you actually need. For example, if I ask it why I'm getting kernel error when attempting to run stock ComfyUI it will tell me to update using a pip command, but it will give me a link to a build that does not exists (cu127) since it believes that CUDA 12.7 is the version I need for my 5090 since that is the min. version of CUDA that Blackwell supports.

I can follow that trail together with chatGPT for hours and end up trying to compile my own version, uninstall/reinstall various extensions and packages and get absolutely nowhere. And chatGPT will always act and respond with 100% confidence no matter what, and I will never figure out the culprit unless I go to the repo or follow the trail myself to discover that it (cu127) does not exists.

Of course, LLMs are great for a number of reasons - but beware when trying to do stuff like this that the output might be complete nonsense.

I concur from personal experience - however as someone who doesn't understand this stuff - and having had an incredible machine arrive a week ago, that I just couldn't get to work - and all my techie friends on discord were at a loss as well - ChatGPT was my only option. Since talking to LLMS is what i do a lot for work etc, I just kept everything inside one thread - so it remembered everything we had been doing - I used its commands to trouble shoot - but used my brain to realise when we were repeating or when it had forgotten about having to use 12.8 sm_120 etc and kept reminding it of these things - and with a little prompting it helped me through a maze of tests and trials I never would have got through on my own. In the end - I went back to the original Comfy Windows stand alone install thats at the top of the page. I learned to install custom nodes in new workflows - 1 at a time, restarting and checking the logs after every node - and trouble shooting anything that threw up one step at a time. And it's now pretty much running. It is slow however - without SageAttention or FlashAttention etc I think it struggles. And I am sure there will be more performance increases as more tools get compiled to use its abilities. I am old and have been around a while - so I realise that life at the bleeding edge is not always easy. I am still confused about what the difference is between Comfyui Stand Alone and Comfyui Portable - half my friends are saying go one way the other the other - but I think they are just two different names for the same install? As far as I can tell (having installed all of these this week - for a Windows machine the options are 1) Desktop Comfy that installs from an exe file - doesn't open in a browser - has a built in terminal in the bottom pull up in a tab next to the logs, but seemed restrictive to use (though I can't tell you why). 2) The Windows Stand Alone I am currently using - which, once you figure out where to run the pip commands etc is pretty straight forward, easy to back up directories - or reinstall by dragging the zip file over the top again 3) WSL - allowed me to use TorchAudio right away - feels like gets quicker support from the community because its in Linux - but got quite complicated very quickly - and installing an unrelated program in a docker in the same WSL killed the whole thing - and I lost everything and no way to recover my files... @mpfcs162 a case in point for your statement - ChatGPT told me to remove the WSL to clear the issue with YuE and gave me the command to run - which I did - and then it wiped everything. (My own fault)

[edisonchan]

So I've used both Comfy UI and Stability Matrix with SD Forge WebUI these past days since getting my 5090.

Then, for ComfyUI at least, you can run this. Adjust the command as needed: python_embeded\python.exe -m pip install H:\torch\torch-2.6.0+cu128.nv-cp310-cp310-win_amd64.whl H:\torch\torchvision-0.20.0a0+cu128.nv-cp310-cp310-win_amd64.whl

You can run this in your dir to check what is installed: python_embeded\python.exe -m pip list

python_embeded\python.exe --version
Python 3.12.8
PS E:\ComfyUI_cu128_50XX> python_embeded\python.exe -m pip install ".\update\torch-2.6.0+cu128.nv-cp310-cp310-win_amd64.whl" ".\update\torchvision-0.20.0a0+cu128.nv-cp310-cp310-win_amd64.whl"
ERROR: torch-2.6.0+cu128.nv-cp310-cp310-win_amd64.whl is not a supported wheel on this platform.

not work?

[adamreading]

So I've used both Comfy UI and Stability Matrix with SD Forge WebUI these past days since getting my 5090.
Then, for ComfyUI at least, you can run this. Adjust the command as needed: python_embeded\python.exe -m pip install H:\torch\torch-2.6.0+cu128.nv-cp310-cp310-win_amd64.whl H:\torch\torchvision-0.20.0a0+cu128.nv-cp310-cp310-win_amd64.whl
You can run this in your dir to check what is installed: python_embeded\python.exe -m pip list

python_embeded\python.exe --version Python 3.12.8 PS E:\ComfyUI_cu128_50XX> python_embeded\python.exe -m pip install ".\update\torch-2.6.0+cu128.nv-cp310-cp310-win_amd64.whl" ".\update\torchvision-0.20.0a0+cu128.nv-cp310-cp310-win_amd64.whl" ERROR: torch-2.6.0+cu128.nv-cp310-cp310-win_amd64.whl is not a supported wheel on this platform.

not work?

You are running python 3.12. You are trying to use the wheels for python 3.10… first go here and download the two 3.12 files to the same directory you are running the command in. https://huggingface.co/w-e-w/torch-2.6.0-cu128.nv Them adapt the command to use your version of the wheel you just downloaded.

[Tathiane123]

Excelente

[Tayyab-H]

Are there no windows pytorch options available for cuda 12.8 still? Any help would be greatly appreciated.

[adamreading]

Looks like you can import any model though - https://lmstudio.ai/docs/basics/import-model

[adamreading]

looks like 2gb of my gpu VRAM is there used all the time with nothing obviously running - then adding a 5gb Quen took it up to 7GB

[adamreading]

llava 1.5 used 12gb VRAM ran super fast

[Tayyab-H]

thanks for the help! Ill give LM studio a try! Its a shame it isnt working through code, id love to build stuff with it.

[pocketpoetry]

How did you get LMstudio to work on a 50 series card? mine will just refuse to load and give me this error, which I assume its cause it doesnt detect a CUDA "Failed to load LLM engine from path: C:\Users\user.lmstudio\extensions\backends\llama.cpp-win-x86_64-nvidia-cuda-avx2-1.22.0\llm_engine_cuda.node. Cannot find module 'C:\Users\user.lmstudio\extensions\backends\llama.cpp-win-x86_64-nvidia-cuda-avx2-1.22.0\llm_engine_cuda.node'
Require stack:

• C:\Users\user\AppData\Local\Programs\LM Studio\resources\app.webpack\lib\llmworker.js"

[yu-jingxiang]

Thank you, friend. My previous problem has been solved, My gguf and nunkaku can both run. But there is an old problem, [2025-06-10 06:31:08.647] [info] Initializing QuantizedFluxModel on device 0 [2025-06-10 06:31:08.681] [info] Loading weights from E:\models\diffusion_models\svdq-fp4-flux.1-dev\transformer_blocks.safetensors [2025-06-10 06:31:08.685] [warning] Failed to load safetensors using method MIO: CUDA error: operation not supported (at C:\Users\muyang\Desktop\nunchaku-dev\src\Serialization.cpp:130) [2025-06-10 06:31:18.948] [info] Done. Injecting quantized module [2025-06-10 06:31:19.296] [info] Set attention implementation to nunchaku-fp16 Loading configuration from E:\models\diffusion_models\svdq-fp4-flux.1-dev\comfy_config.json model_type FLUX Requested to load CLIPVisionModelProjection loaded completely 6937.49091796875 787.7150573730469 True CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cpu, dtype: torch.float16 clip missing: ['text_projection.weight'] It's always like this CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cpu, dtype: torch.float16 clip missing: ['text_projection.weight']， "clip missing: ['text_projection.weight']" What exactly does this mean It can also be generated in the end, why does it keep prompting this! Have you ever seen a similar problem? At 2025-05-15 22:08:22, "H.W.Prinz" ***@***.***> wrote: @a1chera did you get it fixed? i had the same issue, and it is not solvable from my point of view atm (already had some discussion with Dr. LT.Data on it) the Nighty portable packages come with python 3.13.2 that is the only problem you can install a bunch of nodes and stuff not taking care on it, but some, even essential as like Manager don't like 3.13..... at all. but if you're going for an actual download default released package, it will have python 3.12.10 pytorch2.7.0+cu12.8 all good with runs flawlessly have fun — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: ***@***.***>

[maoper11]

Safe versions:
PyTorch 2.7.0 + cu128
xformers==0.0.30
triton-windows==3.3.1.post19
Flash-Attention 2.7.4
SageAttention 2.1.1

https://youtu.be/9WTFauaLsn4

[oliverban]

Safe versions: PyTorch 2.7.0 + cu128 xformers==0.0.30 triton-windows==3.3.1.post19 Flash-Attention 2.7.4 SageAttention 2.1.1

https://youtu.be/9WTFauaLsn4

Already had these installed prior to installing my 5090. Have a 3090 since before. After some initial driver errors, which took a while to debug and fix, I fired off Comfy without any additional prep and the above works great! :) Thanks for pointing these out!

[repeatmlg]

hello anyone can help me ? i was broke my package comfy ui and cannot install custom node
the text is here

Restarting... [Legacy Mode]

Command: ['"D:\ComfyUI_cu128_50XX\ComfyUI_cu128_50XX\python_embeded\python.exe"', '"ComfyUI\main.py"']
D:\ComfyUI_cu128_50XX\ComfyUI_cu128_50XX\python_embeded\python.exe: can't open file 'D:\ComfyUI_cu128_50XX\ComfyUI_cu128_50XX\ComfyUI\custom_nodes\ComfyUI-Cloud\ComfyUI\main.py': [Errno 2] No such file or directory

D:\ComfyUI_cu128_50XX\ComfyUI_cu128_50XX>pause
Press any key to continue . . .

if i try to restart a note appears like this how i solved this eror ?

[repeatmlg]

hello anyone can help me ? i was broke my package comfy ui and cannot install custom node
the text is here

Restarting... [Legacy Mode]

Command: ['"D:\ComfyUI_cu128_50XX\ComfyUI_cu128_50XX\python_embeded\python.exe"', '"ComfyUI\main.py"']
D:\ComfyUI_cu128_50XX\ComfyUI_cu128_50XX\python_embeded\python.exe: can't open file 'D:\ComfyUI_cu128_50XX\ComfyUI_cu128_50XX\ComfyUI\custom_nodes\ComfyUI-Cloud\ComfyUI\main.py': [Errno 2] No such file or directory

D:\ComfyUI_cu128_50XX\ComfyUI_cu128_50XX>pause
Press any key to continue . . .

if i try to restart a note appears like this how i solved this eror ? @comfyanonymous thankyu for attention

[Rogala]

If git is installed in the system, go to the ComfyUI folder, open the console and update everything you deleted

git pull origin master

[Forajz]

I'm having trouble with my current ComfyUI setup because it's outdated. Looks like I’ll have to download a fresh copy, but I really want to avoid going through the nightmare of fixing dependencies, PyTorch versions, xformers, triton and all that again.

Does anyone have the latest version of ComfyUI that works smoothly on RTX 5090 or the new Blackwell series GPUs? I’d really appreciate if you could share it or point me to a reliable source.

Or should I somehow update my Comfy without destroying everything?

[hellonearthis]

I still don't have xformers or Triton working on my 5070 by i got SageAttention working. https://github.com/woct0rdho/SageAttention/releases
Versions of key software I run are cu128 torch2.7.1 python-cp311

[Rogala]

This thread has already described 20 times how to update ComfyUI, it takes about 20 minutes, installing all the additional nodes takes longer, so it is recommended to use the most standard ones to reduce this time.

[BradisonTech]

You really have to use Conda to manage this stuff. Installing ComfyUI is the easy part.

There are just too many different packages that have their own dependencies. I am running ComfyUI easily with my 5070 TI, and I even managed to compile SageAttention and FlashAttention on my Linux box. But this stuff is so bleeding edge that you have to be very specific about how stuff is installed, so you need to segregate the environments.

I have lost many hours to the cause, but it was worth it.

[omgitsgb]

I still don't have xformers or Triton working on my 5070 by i got SageAttention working. https://github.com/woct0rdho/SageAttention/releases Versions of key software I run are cu128 torch2.7.1 python-cp311

its pretty easy,
you have to upgrade your wheels, do the following:

update all of comfyUI (sometimes this messes up torchaudio, its ok though)

"C:\COMFY_UI_LOCATION\python_embeded\python.exe" -m pip install --upgrade pip setuptools
"C:\COMFY_UI_LOCATION\python_embeded\python.exe" -m pip install triton-windows==3.3.1.post19
"C:\COMFY_UI_LOCATION\python_embeded\python.exe" -m pip install --no-cache-dir torchaudio==2.7.1+cu128 --index-url https://download.pytorch.org/whl/cu128

hope this helps! :)

[YauhenLepshi]

This is not a plug‑and‑play solution, but I successfully enabled GPU acceleration on an NVIDIA RTX 5060 Ti (Blackwell) under Ubuntu/WSL2 using PyTorch nightly builds with CUDA 12.8. It requires manual setup and additional dependencies, but it proves that Blackwell GPUs already work with ComfyUI before official PyTorch support is released. Full details are documented here: https://medium.com/@v445683044/rtx-5000-comfyui-why-gpu-doesnt-work-and-how-to-fix-it-september-2025-0f6468bde81a

[mobcat40]

SageAttention working on RTX 5090 + PyTorch 2.11 For anyone with a 50-series card looking to add SageAttention for ~35% faster diffusion sampling: Prebuilt wheel + build instructions: https://github.com/mobcat40/sageattention-blackwell Existing wheels don't work with PyTorch 2.11 nightly (DLL errors). This wheel is compiled against PyTorch 2.11 + CUDA 12.8 for sm_120. Heads up for Qwen/Wan users: The --use-sage-attention flag causes black output. Use KJNodes "Patch Sage Attention" node with sageattn_qk_int8_pv_fp16_cuda backend instead.

[JuanpiAMVs]

You are my hero

[ssuukk]

I have another solution that works for me without any problems, for over a year now - just use uv for managing my ComfyUI installation instead of the default pip+poetry. Here's my setup if anyone wants to try it: https://github.com/ssuukk/comfy-uv, currently it runs on CUDA 13.0, but you can easily fall back to 12.8.

[mattbirk6]

I have an issue where my ComfyUI doesn't seem to be updating everything when I click on the update_comfyui_and_python_dependencies bat, the red errors go by so quickly I had to screen record the process then slow it down. But it says something to the affect that it cannot access the python embedded scripts folder. So I added that directory to the path in system variables tab. Now when my portable version boots up, I get "this app can't run can run on your PC" It seems to stop at Crystools in the loading procedure, but how do I know what is causing the error? Really appreciate any help, thank you.

[mattbirk6]

Okay I think my install is all kinds of screwed up. I can't get ComfyUI manager to update past 3.39. :(

[hellonearthis]

Gemini 3 pro is great at debugging and fixing comfyui installs

…

On Sat, 3 Jan 2026 at 08:55, mattbirk6 ***@***.***> wrote: Okay I think my install is all kinds of screwed up. I can't get ComfyUI manager to update past 3.39. :( — Reply to this email directly, view it on GitHub <#6643 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAA5XTIW5OIZEP2HIR5BKAD4E3EMPAVCNFSM6AAAAABWD7G2YGVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTKMZZGM2DKNA> . You are receiving this because you commented.Message ID: ***@***.*** com>

[mattbirk6]

Thank you very much. You were right. It's downright scary how helpful it was. I felt like I was conversing with one of the ComfyUI lead developers. Holy crap. You saved my sanity, thank you!

[Bonnie197]

回响科技招聘-tusiai
AI Infra工程师
职位描述

模型部署、推理服务框架优化、计算资源调度以及与硬件协同优化，确保模型在生产环境中的高效稳定运行。
深入理解大模型结构（如Transformer, U-Net）与计算图，综合运用模型蒸馏、模型量化、剪枝、算子融合、内核优化（如CUDA/Triton Kernel开发）及计算图优化等技术，提升模型推理速度和效率。
持续追踪、评估并引入业界前沿的开源AIGC技术。负责对这些开源技术进行二次开发、性能优化和适配，以快速集成到现有生产系统中，保持技术领先性。
职位要求

本科及以上学历，计算机相关专业，3年以上相关经验
精通 C++ 和 Python，具备扎实的编程、数据结构和算法基础
对Stable Diffusion等主流AIGC模型有深入的原理理解和丰富的优化实践经验。熟悉并有能力深入定制diffusers, ComfyUI等开源推理框架。
熟练掌握至少一种主流深度学习框架（如PyTorch, TensorFlow）
工作地址：上海普陀区国浩长风城南座6楼

投递链接：https://echotech.jobs.feishu.cn/qiandaozhaopin

[tylerMH]

I added one for folks on Windows / PyTorch 2.11 / Python 3.12 / Cuda 13.0.

For the one guy who is pulling out his hair. https://github.com/tylerMH/flash-attention-windows-5090/releases/tag/v2.8.3-cu130-cp312-torch2.11-win-rtx5090

[tedtninety]

I will try keeping this post up to date as much as possible with the latest developments.

To get your nvidia 50 series GPU working with ComfyUI you need a pytorch that has been built against cuda 12.8

In the next few months there will likely be a lot of performance improvements landing in pytorch for these GPUs so I recommend coming back to this page and updating frequently.

Windows

The recommended download is the latest standalone portable package or desktop installer that you can download from the README

Manual Install

If you install stable pytorch make sure it is cu128.

pytorch nightly cu128 is available for Windows and Linux:

pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu128
You can also use the Nvidia Pytorch Docker Container as an alternative which might give more performance.

Forgive if am missing something, may I ask what I may be missing out on with my current setup?
I thought I was close to the sweet spot.
5090 / Windows PyTorch 2.9.1/ Python 3.11.9/ CUDA 13.0
Thanks for all the info!

I will try keeping this post up to date as much as possible with the latest developments.

To get your nvidia 50 series GPU working with ComfyUI you need a pytorch that has been built against cuda 12.8

In the next few months there will likely be a lot of performance improvements landing in pytorch for these GPUs so I recommend coming back to this page and updating frequently.

Windows

The recommended download is the latest standalone portable package or desktop installer that you can download from the README

Manual Install

If you install stable pytorch make sure it is cu128.

pytorch nightly cu128 is available for Windows and Linux:

pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu128
You can also use the Nvidia Pytorch Docker Container as an alternative which might give more performance.

[rabbithill]

Forgive if am missing something, may I ask what I may be missing out on with my current setup? I thought I was close to the sweet spot. 5090 / Windows PyTorch 2.9.1/ Python 3.11.9/ CUDA 13.0 Thanks for all the info!

You're missing the part where this thread is a year old. You no longer need pre-release stuff.
pip3 install torch torchvision torchaudio xformers --index-url https://download.pytorch.org/whl/cu130
will get you torch 2.10 at cuda 13.0 which works perfectly fine with 5000 series cards, along with all the bells and whistles and badabing badaboom.
Pretty sure the standard install for comfy will have a workable torch version included by default these days.

For future venturers, you can find the latest torch install instructions on https://pytorch.org/

[tedtninety]

Roger that! Thanks a bunch.

[hiroki-abe-58]

Windows Native + cu130 + One-Click Setup (No WSL2, No Docker)

I spent 3 days getting ComfyUI fully working on Windows native with RTX 5090, and packaged the result into a reproducible one-click setup:

https://github.com/hiroki-abe-58/ComfyUI-Win-Blackwell

What it does

• setup.bat → ~20 min → fully working ComfyUI with Blackwell support
• PyTorch nightly cu130 (not cu128 — needed for NVFP4 2x speedup)
• No xformers (it force-downgrades PyTorch to stable — this was the hardest lesson)
• Uses triton-windows + torch.compile + SDPA instead
• 28 custom nodes individually verified (each tested for silent PyTorch downgrade)
• 5 I2V pipelines tested on 32GB VRAM (HunyuanVideo 1.5, Kandinsky 5.0, LTX-2, LongCat, etc.)
• Windows-native — no WSL2, no Docker, no NTFS conversion overhead

Included tools

• verify_env.py — checks sm_120, cu130, Triton, torch.compile in one shot
• fix_windows_compat.py — converts Linux workflow paths to Windows + replaces sageattn → sdpa
• update.bat — updates everything while preserving cu130 compatibility

The 5 rules I found (break any one and the env dies)

1. Use PyTorch nightly cu130 (stable has no sm_120 kernels)
2. Never install xformers (silently downgrades PyTorch)
3. Strip torch from all requirements.txt (custom nodes pull stable torch)
4. Verify PyTorch version after every custom node install
5. Clear proxy env vars on Windows (blocks pip/git silently)

MIT licensed. PRs welcome, especially for additional custom node verifications.

Detailed writeup: English (Medium) / 日本語 (Qiita)

[Rogala]

I spent three days on it, it's just incredible, I don't even know what to say, everything is done in 15 minutes.
Such cries as yellow press, 3I/ATLAS is flying towards us

[hiroki-abe-58]

Glad it worked! 15 minutes is exactly what I was aiming for.

If you run into any issues with specific custom nodes or
workflows, feel free to open an issue — I'll add them to
the verified list.

[Rogala]

That was sarcasm, move over here

[hiroki-abe-58]

Different tools for different use cases.

ComfyUI Master Manager requires git clone + venv and is
explicitly "not compatible with Portable/Embedded distributions."

My setup is a portable, one-click package specifically for
Blackwell sm_120 + cu130 — no git, no venv, no prerequisites.

Happy to discuss technical differences if you have specific
questions.

[Auswolf-debug]

Here is my custom build :)

Checkpoint files will always be loaded safely.
Found comfy_kitchen backend triton: {'available': True, 'disabled': True, 'unavailable_reason': None, 'capabilities': ['apply_rope', 'apply_rope1', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'quantize_mxfp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8']}
Found comfy_kitchen backend eager: {'available': True, 'disabled': False, 'unavailable_reason': None, 'capabilities': ['apply_rope', 'apply_rope1', 'dequantize_mxfp8', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'quantize_mxfp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8', 'scaled_mm_mxfp8', 'scaled_mm_nvfp4']}
Found comfy_kitchen backend cuda: {'available': True, 'disabled': False, 'unavailable_reason': None, 'capabilities': ['apply_rope', 'apply_rope1', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'quantize_mxfp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8', 'scaled_mm_nvfp4']}
Total VRAM 16311 MB, total RAM 32532 MB
pytorch version: 2.10.0+cu130
xformers version: 0.0.34
Enabled fp16 accumulation.
Set vram state to: NORMAL_VRAM
Device: cuda:0 NVIDIA GeForce RTX 5060 Ti : cudaMallocAsync
Using async weight offloading with 2 streams
Enabled pinned memory 14639.0
working around nvidia conv3d memory bug.
Using sage attention
Python version: 3.14.3 (tags/v3.14.3:323c59a, Feb 3 2026, 16:04:56) [MSC v.1944 64 bit (AMD64)]
ComfyUI version: 0.12.3
ComfyUI frontend version: 1.38.13

python 14.3, 2.10.0+cu130, xformers, flash-attention2, sage-attention. Running latest comfyui kitchen with mxfp8 support.
Spent the day introducing mxfp8_e4m3 to the comfyui and it seems to be functional, loaders, loras and saving nodes.

nvfp4 vs mxfp8
Wan2.2 5B ti2v 4 steps @ 5 sec 1280x704
nvfp4 16s inference time
mxfp8 18s inference time
fp8 22s inference time

So it looks mxfp8 should be as fast as nvfp4 it just needs scaled_mm_mxfp8 for cuda. At current state it uses eager.

Not the greatest, but for a video made in 20 secs. Only have to enable the VAE to polish it up.