Using ComfyUI Cloud

Create Amazing AI Apps

Develop ComfyUI workflows online and publish AI Apps to earn revenue

Cutting-edge AI Workflows Community

Share/Study/Edit/Run Online

The Best ComfyUI IDE and Runtime

Offer the latest, most complete nodes and high-performance GPUs

StarryNight

“The Workspace offers amazing flexibility and value. The free development environment and workflow editing make it so easy to get started, and the fast cloud GPUs ensure quick execution. I love that I only pay for the runtime I actually use—it's fair, simple, and cost-effective!”

Astros

“RunningHub's Workspace is incredibly fast and reliable! The pre-installed nodes cover everything I need, and every workflow runs smoothly without errors or red flags. It's the perfect platform for quick, hassle-free AI creation”

SarahC

“RunningHub's Workspace is a powerhouse! With all the nodes pre-installed and ready to go, I can start working on my AI projects in no time. The speed is amazing, and the workflows run without a hitch. It's a must-use platform for anyone looking to create AI apps fast and efficiently!”

Easily Empower Your System with APIs

Easily integrate AI workflows into your system

Invoke RunningHub Workflows on Your Platform

Check Multi-task Running Status of Workflows via API

Design Business Logic to Control Workflow Invocation Strategy

A Big Collection of Models, Always Fresh

Stable Diffusion, Flux, Cogvideo and other popular models

Track the Latest ComfyUI Nodes

Massive pre-installed nodes, daily update

Wan2.1

Wan2.1 is a comprehensive open-source suite of video foundation models. With SOTA performance, it supports consumer-grade GPUs (T2V-1.3B model requires only 8.19GB VRAM). Capable of multi-tasks like Text-to-Video and Visual Text Generation, it also features a powerful Wan-VAE, driving new breakthroughs in video fields.

VACE

VACE is an all-in-one model for video creation and editing, covering tasks like Reference-to-Video (R2V), Video-to-Video Editing (V2V), and Masked V2V, with free task composition. Equipped with capabilities like "Move-Anything," it streamlines workflows and unlocks diverse possibilities for video creation.

FramePack

FramePack is a neural structure for progressive video generation, compressing frame contexts to fixed lengths. This keeps workload independent of video length, letting 13B models process many frames on laptop GPUs with image diffusion-like batch sizes, blending video/image diffusion experiences.

ComfyUI_InfiniteYou

InfiniteYou (InfU) is an early framework using FLUX-like Diffusion Transformers for high-fidelity identity-preserved image generation. It employs InfuseNet and multi-stage training to enhance identity similarity and image quality, outperforming baselines with plug-and-play compatibility.

ComfyUI-LBMWrapper

LBM (Latent Bridge Matching) by gojasper removes distractions (e.g., passersby) & adjusts lighting via latent space matching. Bypassing pixels, it supports tasks like normal estimation. Open-source for fast, precise edits in photography/professional use.

Step1X-Edit

Step1X-Edit is a cutting-edge image editor rivaling closed-source models. It processes instructions via multimodal LLMs, integrates diffusion decoding, trains on high-quality data, outperforms open baselines on GEdit-Bench, and nears proprietary models.

Spark-TTS

Spark-TTS is an advanced text-to-speech system leveraging large language models (LLMs) for highly accurate and natural-sounding voice synthesis. It balances efficiency, flexibility, and power for both research and production applications.

ICEdit

In-Context Edit revolutionizes image editing with only 0.1% training data, outperforming GPT-4o in ID persistence. Official ComfyUI workflow requires just 4GB VRAM, enabling efficient and precise edits for low-cost, high-quality image creation.

Discover What's New in ComfyUI

Explore the latest innovations and follow industry trends

Flux.1 Tools

Flux.1 Tools introduces cutting-edge features for creators, including Fill (Inpainting & Outpainting) for seamless image editing, Depth and Canny tools for advanced visual control similar to ControlNet, and Redux for effortless style transfer. RunningHub has officially integrated the full suite of Flux.1 Tools now!

CogVideo

CogVideo is an advanced text-to-video generation model using hierarchical training and pretrained image models like CogView2 for smooth, coherent videos. It excels in text-video alignment and multimodal understanding, with open-source tools for easy integration and innovation

Tencent's Hunyuan Video Model

Tencent's Hunyuan Video Model is an advanced text-to-video generation system that combines a Multimodal Large Language Model (MLLM) and 3D Variational Autoencoder (3D VAE) to produce high-quality videos with low computational cost. It features a prompt rewrite mechanism for better semantic alignment and excels in motion quality and text-to-video fidelity, making it a cutting-edge tool for diverse creative applications