<h1 id="comfyui-auto-installer-v5-(python-rewrite)">🚀 ComfyUI Auto-Installer — v5 (Python Rewrite)</h1>Version 5 is a full rewrite from the ground up in Python, replacing all the PowerShell scripts from previous versions. It's cross-platform, faster, smarter, and now ships with a TUI manager, Docker images, and GPU-optimized inference out of the box.If you are upgrading from the PowerShell version (v4.x), a one-command migration preserves all your models, outputs, and custom nodes: <code>irm https://get.umeai.art/migrate.ps1 | iex</code><h2 id="quick-start-(one-liner)">⚡ Quick Start (One-Liner)</h2><ul><li>Windows (PowerShell): <code>irm https://get.umeai.art/comfyui.ps1 | iex</code></li><li>Linux / macOS: <code>curl -fsSL https://get.umeai.art/comfyui.sh | sh</code></li></ul>Only requires Git — everything else (Python, uv, dependencies) is handled automatically.<h2 id="what's-new-in-v5">✨ What's New in v5</h2><ul><li>Full Python rewrite — no more PowerShell dependency</li><li>Cross-platform — Windows, Linux, macOS, and Docker</li><li>TUI Manager — interactive terminal UI to launch, update, download models, and configure settings</li><li>VRAM-aware model catalog — 7 model families with quantization recommendations based on your GPU</li><li>GPU auto-detection — NVIDIA (CUDA 13.0/12.8), AMD (ROCm/DirectML), Apple Silicon (MPS)</li><li>SageAttention 2 + 3 — pre-compiled wheels including RTX 50XX Blackwell support</li><li>One-click update — update ComfyUI, all nodes, and dependencies with a single command</li><li>Model security scanner — detects malicious pickle code in .ckpt/.pt files</li><li>Junction architecture — models and outputs persist independently from ComfyUI updates</li><li>Docker ready — 4 image variants including a cloud version with JupyterLab for RunPod</li></ul><h2 id="prerequisites">📋 Prerequisites</h2><ul><li>Git</li><li>GPU: NVIDIA (CUDA 12.x+), AMD (Radeon RX 6000+), or Apple Silicon (M1+)</li><li>Internet connection</li><li>Note: Python is automatically installed via uv if not present. No manual Python setup required.</li></ul><h3 id="model-catalog-(7-families)">🎨 Model Catalog (7 Families)</h3>Interactive model downloader with VRAM-based recommendations (★ markers) and SHA-256 integrity checks. Each bundle offers multiple quantization variants (fp16, fp8, GGUF Q3→Q8). Downloads are accelerated via aria2c with HuggingFace + ModelScope fallback:<ul><li>FLUX (Image): Dev, Fill</li><li>Z-IMAGE (Image): Turbo</li><li>WAN 2.1 (Video): T2V, I2V 480p</li><li>WAN 2.2 (Video): I2V, Fun Inpaint, Fun Camera</li><li>HiDream (Image): Dev</li><li>QWEN (Image Edit): Image Edit</li><li>LTX-2 (Video + Audio): Dev</li></ul><h3 id="34-custom-nodes-included">🧩 34 Custom Nodes Included</h3>Additive manifest — never removes user-installed nodes.<ul><li>Core (always installed): ComfyUI-Manager</li><li>UmeAiRT Tier: ComfyUI-UmeAiRT-Sync, ComfyUI-UmeAiRT-Toolkit, ComfyUI-Crystools, ComfyUI-nunchaku</li><li>Full Tier (all of the above +): ComfyUI-Impact-Pack, ComfyUI-Impact-Subpack, ComfyUI-GGUF, ComfyUI-mxToolkit, ComfyUI-Custom-Scripts, ComfyUI-KJNodes, ComfyUI-WanVideoWrapper, ComfyUI-VideoHelperSuite, ComfyUI-Frame-Interpolation, rgthree-comfy, ComfyUI-Easy-Use, ComfyUI-HunyuanVideoMultiLora, ComfyUI-Florence2, ComfyUI-MultiGPU, ComfyUI-WanStartEndFramesNative, ComfyUI-Image-Saver, ComfyUI_UltimateSDUpscale, comfyui_controlnet_aux, x-flux-comfyui, ComfyUI-Detail-Daemon, wlsh_nodes, ComfyUI_essentials, ComfyUI-wanBlockswap, Derfuu_ComfyUI_ModdedNodes, ComfyUI_LayerStyle, ComfyUI-Upscaler-Tensorrt, comfyui-vrgamedevgirl, comfyui-int-and-float, was-node-suite-comfyui</li></ul><h3 id="gpu-optimizations-(auto-installed)">⚙️ GPU Optimizations (Auto-Installed)</h3><ul><li>PyTorch 2.10: CUDA 13.0/12.8, ROCm 7.1, DirectML, MPS</li><li>xformers: Memory-efficient attention</li><li>Triton: triton-windows / triton (Linux)</li><li>SageAttention 2: Unified ABI3 wheels (Windows), per-arch SM80–SM100 (Linux)</li><li>SageAttention 3: RTX 50XX Blackwell native (Windows + Linux)</li><li>FlashAttention: Linux + NVIDIA only</li><li>Nunchaku &amp; InsightFace: Pre-compiled wheels</li><li>Additional Python packages auto-installed: facexlib, onnxruntime-gpu, nvidia-ml-py, cupy-cuda13x, imageio-ffmpeg, hf_xet, cython, rotary_embedding_torch, blend_modes, segment_anything, gguf, and more.</li></ul><h3 id="docker-support">🐳 Docker Support</h3>Requires Docker and an NVIDIA GPU: <code>docker run --gpus all -p 8188:8188 -v comfyui-data:/data registry.gitlab.com/umeairt-studio/comfyui-auto_installer-python:latest</code><ul><li>latest: ~4 GB — Ready to go with pre-installed PyTorch</li><li>latest-cloud: ~4.5 GB — + JupyterLab for RunPod / cloud</li><li>latest-lite: ~2 GB — Minimal (installs PyTorch on first run)</li><li>latest-lite-cloud: ~2 GB — Lite + JupyterLab</li></ul><h3 id="security">🔒 Security</h3><ul><li>No external script execution — all logic is internalized</li><li>Secure subprocess calls — no shell=True</li><li>HTTPS only — all URLs validated</li><li>SHA-256 integrity checks on all model downloads</li><li>Pickle model scanner — detects malicious code in .ckpt/.pt files</li><li>Zip-slip prevention on archive extraction</li><li>CI runs Bandit + pip-audit on every push</li></ul><h3 id="post-installation">📂 Post-Installation</h3>Three launcher scripts are generated:<ul><li>UmeAiRT-Start-ComfyUI: Launch (Performance mode + SageAttention)</li><li>UmeAiRT-Start-ComfyUI_LowVRAM: Launch with --lowvram --fp8 for ≤8 GB VRAM</li><li>UmeAiRT-Manager: TUI manager (update, download, reinstall, settings)</li></ul><h3 id="links">🔗 Links</h3><ul><li>Source code: GitLab (<a target="_blank" rel="ugc" href="https://gitlab.com/UmeAiRT-Studio/ComfyUI-Auto_installer-Python">https://gitlab.com/UmeAiRT-Studio/ComfyUI-Auto_installer-Python</a>)</li><li>Mirror: Codeberg (<a target="_blank" rel="ugc" href="https://codeberg.org/UmeAiRT">https://codeberg.org/UmeAiRT</a>)</li><li>Ecosystem: UmeAiRT Studio (<a target="_blank" rel="ugc" href="https://umeai.art">https://umeai.art</a>)</li></ul>

【TOOL】ComfyUI Installer

This massive update finalizes the transition of the PowerShell installer into a highly resilient, modern, and secure maintenance tool for your ComfyUI ecosystem. While the Python version is actively being developed as our new ultimate solution, this PowerShell legacy version has received premium features to keep the current environments rock-solid!

Files with SHA256: ee3ffc94d6712911a01afa1eff91233d7db26768b17c7a176bd8320f18ac6977