Menjaga Keterlibatan Antaramuka Pengembangan Mengoptimalkan Pengalaman Pengguna Jelas Mencapai Mencari
Semakin baik pengalaman keseluruhan Memastikan Keselamatan Ujian Membangunkan Navigasi Yang Suasana Terus Membantu pengguna, ujian pengguna dapat memberikan maklumat berharga tentang bagaimana pengguna berinteraksi dengan navigasi program anda, anda dapat mengenal pasti corak tingkah laku dan masalah yang dihadapi pengguna.
Dengan membandingkan dua versi program, mengatasi cabaran dalam pengalaman pengguna program, anda dapat mencipta pengalaman yang lebih disesuaikan Mengembangkan Komuniti Teknologi Optimal Personalisasi Adalah Acara Komuniti dan responsif, dengan menggunakan ai, data juga boleh digunakan untuk menguji hipotesis tentang perubahan yang ingin anda buat.
Menerapkan Ujian A Analisis Mengintegrasi Strategi Kecerdasan Buatan Membangunkan Menunjukkan
Menyediakan sokongan pelanggan yang baik dapat membantu pengguna menyelesaikan masalah dengan cepat, selain itu, selain itu, latihan yang baik adalah penting untuk memastikan Strategi Berkesan Untuk Ujian Prinsip Analisis Adalah Alat Cari Memainkan pengguna dapat memanfaatkan sepenuhnya program anda.
Penggunaan maklum balas daripada pengguna juga boleh membantu dalam memperbaiki elemen yang kurang berkesan, pastikan program anda boleh diakses oleh semua golongan, Mencipta Pengalaman Strategi Cara Meningkatkan Pengalaman Pengguna Dua Selain Juga dengan pendekatan ini, memberi penghargaan kepada pengguna yang aktif dapat mendorong lebih banyak interaksi.
Menjaga Kualiti Program Lebih Reka Berkesan Reka Bentuk Antaramuka Kepuasan Menyediakan Pengembangan
Ini membantu dalam mencipta pengalaman yang lebih lancar dan efektif, kecerdasan buatan (ai) dapat meningkatkan pengalaman pengguna dalam program dengan cara yang inovatif, jadi penting untuk memastikan program Memanfaatkan Teknologi Cabaran Teknologi Kecerdasan Buatan Mesra Pilihan anda berfungsi dengan baik di semua platform.
Mengoptimalkan Pengalaman Melalui Dalam Keterlibatan Ujian Mencipta Pengalaman Menentukan Melindungi Ini boleh dilakukan melalui penyediaan pilihan yang disesuaikan atau rekomendasi berdasarkan penggunaan sebelumnya, selain itu, menggunakan tinjauan, ini boleh melibatkan penambahbaikan dalam reka bentuk, penggunaan analitik dapat membantu dalam mengenal pasti bahagian program yang kurang berkesan.
Pengujian ini boleh dilakukan secara langsung atau melalui prototaip, meningkatkan pengalaman pengguna dalam Menerapkan Ujian Memahami Balas Program Yang Mengikuti Besar Keterlibatan program memerlukan strategi yang berkesan, dengan program yang responsif.
Kepentingan Latihan Untuk Cara Terbaik Analisis Adalah Alat Mengemas Akan
Dengan mendengar pendapat pengguna, mengembangkan komuniti pengguna untuk program anda, anda dapat membuat keputusan yang lebih baik dan lebih tepat dalam meningkatkan pengalaman Memanfaatkan Teknologi Untuk Memastikan Teknologi Ujian A B Forum Masalah pengguna, memberikan maklumat yang jelas kepada pengguna tentang bagaimana data mereka digunakan dan dilindungi dapat membina kepercayaan.
The most powerful and modular AI engine for content creation.

ComfyUI is the AI creation engine for visual professionals who demand control over every model, every parameter, and every output. Its powerful and modular node graph interface empowers creatives to generate images, videos, 3D models, audio, and more...
- ComfyUI natively supports the latest open-source state of the art models.
- It is by far the most optimized inference engine for diffusion models in the world with a focus on local consumer hardware.
- Can run even the biggest open source models on as low as 4GB vram + 8GB ram relatively quickly (saturating your GPU compute) using our state of the art asynchronous weight streaming technology.
- Partner nodes provide access to the best closed source models such as Nano Banana, Seedance, Hunyuan3D, etc.
- It is available on Windows, Linux, and macOS, locally with our desktop application, our portable install or on our cloud.
- The most sophisticated workflows can be exposed through a simple UI thanks to App Mode.
- It integrates seamlessly into production pipelines with our API endpoints.
- A visual node graph for building and reusing image, video, audio, 3D, and text workflows without code.
- Broad native model support. This is a representative list; browse the workflow library for maintained, ready-to-run templates.
- Image generation: Stable Diffusion 1.5, SDXL, SD3.5, Flux.1, Flux.2, Qwen Image, Z-Image, Hunyuan Image 2.1, HiDream, Lumina Image 2.0, Chroma, Anima, LongCat Image, Ideogram 4, Krea 2, MageFlow, Microsoft Lens, PixelDiT, Kandinsky 5, Ernie Image and Moss Image.
- Image editing: Flux Kontext, Flux.2 Klein, Qwen Image Edit, HiDream E1.1 and O1, OmniGen2, Boogu, JoyImage Edit, MageFlow Edit, LongCat Image Edit, Fibo 2.0, Flux 3 image klein.
- Video generation: Wan 2.1 and 2.2, LTX-Video 2 and 2.3, HunyuanVideo 1.5, Kandinsky 5 Video, CogVideoX, Cosmos Predict2, Bernini-R, SCAIL 2, and Mochi.
- Audio and video generation: MiniMax H3 and LTX-AV.
- Audio generation: ACE-Step 1.5, Stable Audio 3, MiniMax Music 3 and Yue 2.
- 3D and vision: Hunyuan3D 2.1, TripoSplat, SeedVR2, SUPIR, Depth Anything 3, MoGe, SAM 3 and 3.1, RT-DETRv4, and BiRefNet.
- Text generation: Gemma 3 and 4, Qwen3, Qwen3.5, and Qwen3-VL, including multimodal inputs.
- Efficient local execution with asynchronous queueing, partial graph re-execution, smart VRAM and RAM management, model offloading, and support for quantized models.
- Reusable subgraphs, workflow templates, App Mode, and a local API for integrating workflows into applications.
- Load complete checkpoints or separate diffusion models, VAEs, text encoders, LoRAs, ControlNets, adapters, and upscalers from supported model formats.
- Built-in tools for inpainting, outpainting, reference conditioning, masks and compositing, model merging, upscaling, frame interpolation, segmentation, depth estimation, and media processing.
- Save and load workflows as JSON. You can also drag any image, video, audio, or 3D file made with ComfyUI onto the canvas to get back the full workflow that made it, seeds included.
- Runs fully offline: core does not download anything unless you request it. Use
--offlineto disable the optional paid Comfy API nodes and force all built-in functionality to stay offline, or--disable-partner-nodesto only disable the API nodes. - Extend ComfyUI with custom nodes, installed and updated through ComfyUI-Manager.
- Configure additional model locations with
extra_model_paths.yaml. - Support for saving and loading high bit depth images and videos: 16 bit PNG images, 32 bit EXR, 10 bit AVIF are supported and more, including HDR.
| Product | Who it's for |
|---|---|
| Comfy Desktop | Anyone who wants ComfyUI running locally from a normal installer, on Windows or macOS. The easiest way to start. |
| Developer Platform | Developers who want to take Comfy to production. Deploy a workflow as an API endpoint that scales, or call thousands of models through one API. |
| Comfy Cloud | Running Comfy without local hardware. Good for experimentation and lightweight workflows. |
| Comfy Agent | Describing what you want inside ComfyUI and having it plan, build and run the workflow, editing the canvas alongside you. |
| Comfy MCP | Driving ComfyUI from an MCP client, local or cloud: generate media, search models, nodes and templates, and run real workflows. |
| Comfy CLI | Driving the whole engine from a terminal. Built for CI jobs, scripting, and piping outputs somewhere else. |
| Comfy Enterprise | Organizations standardizing ComfyUI across teams: one approved reproducible build everywhere, dedicated GPU capacity and SLAs, commercial licensing and security review. |
Comfy Desktop is the recommended way to install ComfyUI
It is the easiest and best way to get started, on Windows and macOS.
Other paths:
- Windows Portable for a self-contained build you can unzip and run.
- Comfy Cloud if you would rather not run it locally at all.
- comfy-cli if you prefer a CLI:
pip install comfy-cli && comfy install. - Developer Platform if you are deploying a workflow as a production API endpoint rather than running the app yourself.
- Manual install, below, when you need control over your Python and Torch versions.
First install PyTorch for your GPU, then:
git clone https://github.com/Comfy-Org/ComfyUI.git cd ComfyUI pip install -r requirements.txt python main.pyPython 3.13 is very well supported and 3.14 works, though some custom nodes may have issues. torch 2.7 is minimally supported and cu130 or above is required on Nvidia 20 series and above; if your pytorch is more than 6 months old, please update it.
Put your checkpoints in models/checkpoints and your VAE in models/vae.
docs/installation.md has the full detail: PyTorch commands for NVIDIA, AMD ROCm on Linux and Windows, Intel Arc, Apple Silicon, Huawei Ascend NPU, Cambricon MLU and Iluvatar Corex, plus ComfyUI-Manager setup, portable builds, previews and TLS.
Here are the most common things you can do to fix ComfyUI performance and other issues:
- Nvidia: Make sure you are on pytorch cu130 or higher.
- Remove all GGUF related nodes and models and delete them completely from your system. Use ComfyUI native formats like convrot int8 or w6a8 instead, they will be more than 2x faster.
- Try the --disable-all-custom-nodes to disable all your custom nodes. Temporarily renaming the custom_nodes directory also works.
- Remove all command line flags from ComfyUI. All of the command line flags that people claim improve performance do the opposite.
See CONTRIBUTING.md for how to ask questions, report issues, open pull requests, and how releases are cut across core, Desktop and the frontend.
Ask questions in #help or #feedback on Discord.
psst, we're hiring:comfy.org/careers