Flux 2 Setup Guide: Get It Running on ComfyUI in 2025

Video Description

How to install Flux 2 in ComfyUI on Windows and RunPod in 2025 – step by step. In this video we fix the “Missing Models” error, download the correct Flux 2 dev, text encoder and VAE files, and wire a clean no-spaghetti workflow that works both locally and in the cloud. If Flux 2 won’t load in ComfyUI, or you keep getting model not found errors on RunPod, this tutorial walks through the exact setup, folder paths, and a working multi-reference graph. ----------------------------------- 📦 PATREON WORKFLOW & DISCORD ----------------------------------- My advanced Flux 2 ComfyUI workflow includes: • Up to 10 reference image slots (face, background, outfit, pose, logos, palettes, etc.) • rgthree bypass switches to toggle reference slots on/off Additional Downloads Notes, folder maps and ready-to-paste download commands, prompt pack, flux 2 guide, and much more! Get the workflow + join the Doss AI Discord here: 👉 https://www.patreon.com/JamesDossAI ----------------------------------- 👑 WHAT THIS VIDEO COVERS ----------------------------------- • Install Flux 2 for ComfyUI on RunPod and local Windows • Fix the “Missing Models” popup when loading the Flux 2 workflow • Download the 3 required Flux 2 model files: – Flux 2 dev diffusion checkpoint – Mistral 3 small Flux 2 text encoder – Flux 2 VAE • Put each file in the correct ComfyUI /models/ subfolder • Restart ComfyUI so the models show up in the dropdowns • Hide messy node links with Lite Graph → “Hidden” mode • Use my clean 10-reference Flux 2 workflow (no graph spaghetti) • Enable / disable reference slots with rgthree bypass switches • Run a real multi-reference example: – face (ref 1) – jungle background (ref 2) – outfit (ref 3) – pose sketch (ref 4) ⚠️ This video is focused on INSTALL + WIRING + FIRST IMAGES. The next video is a full **“How to Use Flux 2 in ComfyUI”** guide (prompts, character consistency, palettes, edits). Subscribe so you don’t miss it. ----------------------------------- 🧩 PREREQUISITES ----------------------------------- You should already have: • ComfyUI installed (local or RunPod) • A basic ComfyUI template running on RunPod If you’re brand new, start here: 🔹 ComfyUI base install / RunPod template tutorial: 👉 https://youtu.be/j37l9q3u1nk RunPod template used in this video: 👉 https://console.runpod.io/deploy?template=lyoe0y42qv&ref=313ph7qf ***MAKE SURE TO UPDATE ALL*** So you are on the most updated ComfyUI Version. ----------------------------------- 📁 FLUX 2 MODEL FILES & FOLDERS ----------------------------------- Drop these into your ComfyUI /models/ folders: ComfyUI/models/text_encoders └─ mistral_3_small_flux2_bf16.safetensors https://huggingface.co/Comfy-Org/flux2-dev/resolve/main/split_files/text_encoders/mistral_3_small_flux2_bf16.safetensors ComfyUI/models/diffusion_models └─ flux2_dev_fp8mixed.safetensors https://huggingface.co/Comfy-Org/flux2-dev/resolve/main/split_files/diffusion_models/flux2_dev_fp8mixed.safetensors ComfyUI/models/vae └─ flux2-vae.safetensors https://huggingface.co/Comfy-Org/flux2-dev/resolve/main/split_files/vae/flux2-vae.safetensors After copying the files, **restart ComfyUI and refresh your browser** so the models appear in the Load Diffusion Model, Load CLIP and Load VAE dropdowns. ----------------------------------- 💻 OPTIONAL – WGET COMMANDS (RUNPOD SHELL) ----------------------------------- From inside your ComfyUI/models directory on RunPod you can also use: Text encoder: wget "https://huggingface.co/Comfy-Org/flux2-dev/resolve/main/split_files/text_encoders/mistral_3_small_flux2_bf16.safetensors" -O "text_encoders/mistral_3_small_flux2_bf16.safetensors" Diffusion model: wget "https://huggingface.co/Comfy-Org/flux2-dev/resolve/main/split_files/diffusion_models/flux2_dev_fp8mixed.safetensors" -O "diffusion_models/flux2_dev_fp8mixed.safetensors" VAE: wget "https://huggingface.co/Comfy-Org/flux2-dev/resolve/main/split_files/vae/flux2-vae.safetensors" -O "vae/flux2-vae.safetensors" ----------------------------------- ⏱️ CHAPTERS ----------------------------------- 00:00 – Flux 2 Overview & Required Files (Quick) 01:10 – Wire the Models & Clean ComfyUI Workflow 04:47 – Flux 2 In Action (Multi-Reference Demo) 05:44 – Next Steps: Mastering Flux 2 in ComfyUI ----------------------------------- 📜 CREDITS & LICENSE ----------------------------------- Flux 2 / FLUX.2 models are created and released by Black Forest Labs. Official model page and license: https://github.com/black-forest-labs/flux2 https://huggingface.co/black-forest-labs