ComfyUI is a modular, node-based AI creation engine designed for visual professionals to generate images, video, audio, 3D models, and text. Utilizing a visual graph interface, the system supports complex, reusable workflows through subgraphs and templates, while offering an API for production pipeline integration and an "App Mode" for simplified UI exposure. Technically, the engine optimizes local execution via asynchronous queueing, partial graph re-execution, and intelligent VRAM/RAM management, including model offloading and quantized model support. It features extensive native support for state-of-the-art models, including Stable Diffusion (1.5, SDXL, SD3.5), Flux (1 and 2), Wan, HunyuanVideo, and various multimodal text models like Gemma 3 and 4. The platform is cross-platform, running on Windows, Linux, and macOS, with hardware acceleration for NVIDIA (CUDA), AMD (ROCm), Intel (XPU), Apple Silicon, Ascend, and Cambricon. Advanced capabilities include high bit-depth support (16-bit PNG, 32-bit EXR, 10-bit AVIF) and HDR media processing. The architecture is divided into three interconnected repositories: ComfyUI Core, Comfy Desktop, and ComfyUI Frontend, with a release cycle targeting major stable versions approximately every two weeks.