ComfyUI build based on Python 3.12, Torch 2.7 and CUDA 12.8 with support for Nvidia 5000 series.
Introducing an updated and scalable ComfyUI build based on Python 3.12, Torch 2.7 and CUDA 12.8, with full support for Nvidia 3000, 4000 and 5000 series graphics cards.
This build greatly simplifies ComfyUI installation and customization, saves time, and already includes all the necessary libraries and modules to work with neural networks, image generation, animations and 3D content.
Version with built-in Triton 3.3.0 - for modern graphics cards (5000/4000/3000/3000 series) with support for SageAttention, q8_kernels and LTX FP8 attentions.
Included libraries and modules:insightface - support for InstantID and Reactordlib, facexdlib - for PulID, Old Photo Restorationnunchaku - for generation with Flux SVDQ.
Libraries for HunYuan 3D, StableSR, etc.
Supported assemblies and models:LTX 13B, Wan 2.1, Reactor NSFW, Union 2, FramePack, Supir,Flux Tools, Flux Redax, Flux PulID, Flux SVDQ, Flux Fill Subject,Old Photo Restoration, HunYuan 3D, SDXL InstantID, SDXL IPAdapter, StableSR, Virtual Fitting Room and others.
Download ComfyUI ready build, run locally on your PC and start using the features of modern neural networks without long customization.
- download comfyui build
- comfyui for nvidia 5000
- torch 2.7 cuda 12.8 python 3.12
- ready comfyui build
- comfyui with triton 3.3.0
- comfyui for sdxl instantid and reactor
- install comfyui without errors
- image generation on local PC
- generative AI without the cloud
...
