news

Revolutionize Video Creation: Wan2.1 Now Supported in ComfyUI

Unlock Efficiency with ComfyUI Frontend 1.10: New Selection Toolbox & Workflow Templates

Unlock Next-Level Video Generation with ComfyUI's LTX-Video 0.9.5 Integration

"Revolutionizing Video Generation: Latest Updates on WAN 2.1 and Hunyuan Image to Video Models"

Revolutionizing AI-Generated Content: NVIDIA's Cosmos Integrates with ComfyUI

Revolutionize Video Creation: HunyuanVideo-I2V, the Ultimate Image-to-Video Solution

Revolutionizing Video Generation: 3 Groundbreaking AI Models You Need to Know

Discover ComfyUI-RMBG: The Powerful Background Removal Tool for ComfyUI Users

ACE+ Unified FFT Model: The Ultimate Solution for Image Generation and Editing Needs

Rewriting Industry Standards: The FLUX AI Outfit-Changing System Explained

Introducing ComfyUI 1.10: Smarter Selections and Faster Workflows Ahead

ComfyUI Desktop Unleashed: Discover the Top Benefits of the Latest Release

Revolutionize Your Workflow: 5 Reasons to Choose ComfyUI for AI Image Creation

Unlock ComfyUI on M3 Chip MacBook Pro: A Step-by-Step Installation Guide

From Zero to Hero: Mastering ComfyUI Plugin Installation in Minutes

Edit Images with Ease: Gemini 2.0 Flash Experimental Plugin Now Available for ComfyUI

Create Music in Minutes: DiffRhythm's Breakthrough Technology Now on the Cloud

🎵 Generate Music in Minutes: Introducing ComfyUI_DiffRhythm v1.0.0!

ComfyUI Copilot: The Ultimate Game-Changer for Efficient UI Workflow

Unlock the Power of 3D Generation: Hunyuan Model Update in ComfyUI

"Revolutionizing 3D Generation: ComfyUI Now Supports Hunyuan3D 2.0!"

Unlocking the Power of Multimodal AI: Jieyue Xingchen's Step-Video-TI2V Model Revolutionizes Image-to-Video Generation

ComfyUI-Manager: A New Chapter in Custom Node Evolution

Say Goodbye to Installation Nightmares: ComfyUI's Game-Changing Update is Here!

ComfyUI Updates: Native Wan2.1 FLF2V Support for Stunning 720p Video Creation

HiDream-I1 Goes Native: Explore the Latest Capabilities and Workflows

ComfyUI Integrates OpenAI's Latest Image Generation Model: GPT-Image-1

Comfy Just Got a Major Boost: 62 New API Nodes and Counting!

Unlock Endless Audio Possibilities: ComfyUI Now Supports ACE-Step!

ComfyUI API Keys Now Available: Streamline Your API Node Experience

Wan2.1-VACE Native Support and Ace-Step Workflow Refined

ComfyUI API Nodes Wave 2 is LIVE!

Introducing the Comfy Bounty Program

GPT Killer: FLUX.1 Kontext API Node Day-0 Support

Node Confusion Solved: ComfyUI's New Built-In Documentation

Introducing Subgraphs: A Powerful New Feature Coming to ComfyUI

ComfyUI custom node V3 version: dependency resolution and standard specification details

Unlock the Power of Image-to-Video: HunyuanVideo I2V Now Available in ComfyUI

ComfyUI Now Offers Native Support for NVIDIA’s Powerful Cosmos-Predict2 Model Suite

ComfyUI Subgraph I/O System: Beta Enhancements Released

ComfyUI Now Offers Native Support for NVIDIA’s Powerful Cosmos-Predict2 Model Suite

CN

ComfyUI.org

2025-06-20 16:24:34

Model Overview

Cosmos-Predict2 is a family of high-performance pre-trained world foundation models developed by NVIDIA, designed specifically for generating physics-aware images, videos, and world states for Physical AI development.

These models are available for commercial use under the NVIDIA Open Model License.

Supported models can be found at: Comfy-Org/Cosmos_Predict2_repackaged

Getting Started

Update ComfyUI to the latest version
Navigate to Workflow → Template to locate Cosmos templates, or download the example workflows below
Download the models as instructed and run

Cosmos-Predict2 Text-to-Image Workflow

mc4jujbrqv7uu3tj6us5059791b-2e27-48a2-9205-7bffe4f59b76_2048x770.webp

Example Outputs

mc4jxfr506rop21qf8d8c46513d-2363-4ec3-8b72-115807041b65_1024x1024.webp

mc4jxyeui8bx0ge4awi13e44b42-fb28-4609-aed0-897063589088_1024x1024.webp

mc4jya0skw8z6gl2tr2fe65fb4-fe6b-4c99-aa50-e77bbaa7adfc_1280x720.webp

mc4jymfgfh0c882tqy2d18c8ef-aa1f-4b5a-a9b9-c4e96493862b_1360x768.webp

mc4jyme0gftdfsdlue0b50571a-78b2-4835-9b9f-94bee92a906c_1360x768.webp

For details, refer to the documentation.

Cosmos-Predict2 Video2World (Image-to-Video) Workflow

mc4k1bf88jgmrv5i45904636782-3c81-4207-926b-22371dcb4a96_2048x1040.webp

Example Output

For additional details, see the documentation. Start creating today!

Unlock the Power of Image-to-Video: HunyuanVideo I2V Now Available in ComfyUI

ComfyUI Subgraph I/O System: Beta Enhancements Released

Recommend

Discover the Ultimate Eastern Art Creation Workflow with AI

Unlock Eastern Pixar-style art creation with this workflow! Generate high-quality images with Flux.1 and Lora models. Download now and enhance your digital illustrations!

Boost Your Image Generation Game with Stable Diffusion, JOY Caption Two, and LORA

Unlock AI-powered image generation with Stable Diffusion, JOY Caption Two, and FLUX. Discover how to reverse-engineer prompts from reference images and create stunning new visuals. Learn more and start creating now!

"Revolutionizing 3D Generation: ComfyUI Now Supports Hunyuan3D 2.0!"

Unlock 3D Generation with Hunyuan3D 2.0! Discover how ComfyUI's native support for Tencent's open-source model empowers high-fidelity 3D creation - try it now!

Transforming Line Art into 3D-Style Renders: A Deep Dive into ControlNet and Dual CLIP Encoding

Unlock Stunning Art: Transform line art into vibrant illustrations & 3D-style renders with ControlNet-guided generation & super-resolution. Learn how to use this AI workflow for breathtaking results.

Unveiling the Past: Transforming Ancient Paintings into Hyper-Realistic Photos

Transform Ancient Portraits into Hyper-Realistic Photos with AI. Discover how to use SDXL models & multi-ControlNet guidance to bring historical figures to life.

Summary

Unlock the power of physics-aware image and video generation with Cosmos-Predict2, NVIDIA's high-performance pre-trained models. Get started today and discover the possibilities!

Chapter