[{"data":1,"prerenderedAt":242},["ShallowReactive",2],{"seo-verification":3,"blog-how-to-host-comfyui-on-a-vps-ma-en":6,"promo-bar-likely":241},{"google":4,"bing":5},"EycwPY2XMyTkVzas3n1ygeNJFGAH513qrMjfDljzsMQ","",{"key":7,"data":8},"blog-how-to-host-comfyui-on-a-vps-ma-en",{"id":9,"slug":10,"slugs":11,"title":15,"excerpt":16,"readTime":17,"views":18,"isPinned":19,"noindex":19,"publishedAt":20,"updatedAt":21,"category":22,"categories":28,"featuredImage":30,"bgImage":31,"posterImage":32,"relatedSolution":30,"intro":33,"sections":34,"ctaTitle":186,"ctaBody":187,"ctaButton":188,"ctaUrl":189,"relatedPosts":190},19,"how-to-host-comfyui-on-a-vps",{"fr":12,"en":10,"ar":13,"es":14},"heberger-comfyui-vps","كيفية-استضافة-comfyui-على-خادم-vps","como-alojar-comfyui-en-un-vps","How to host ComfyUI on a VPS: complete guide 2026","Deploy ComfyUI on a VPS in under an hour: hardware requirements, Docker\u002Fvenv install, SDXL\u002FFlux downloads, reverse proxy and troubleshooting.",9,3,false,"2026-06-01T00:00:00+00:00","2026-11-08T01:13:13+00:00",{"id":23,"name":24,"slug":25,"color":26,"icon":27},1,"Artificial Intelligence","intelligence-artificielle","bg-purple-500\u002F10 text-purple-400","ia",[29],{"id":23,"name":24,"slug":25,"color":26,"icon":27},null,"\u002Fblog\u002Fcovers\u002Fbg.svg","\u002Fblog\u002Fcovers\u002Fheberger-comfyui-vps-poster.svg","ComfyUI is the reference node-based interface for driving Stable Diffusion and diffusion models through reproducible visual workflows. Hosting it on a VPS, ideally with a GPU, turns it into an image-generation studio accessible remotely and automatable via API. This guide covers the full installation: hardware sizing, Docker or Python venv setup, model downloads from HuggingFace, reverse proxy configuration and common error fixes.",[35,39,49,52,85,88,104,116,119,131,134,173,176,183],{"type":36,"title":37,"body":38},"h2","Why self-host ComfyUI on a VPS","ComfyUI organizes image generation into node graphs: each step (model loading, prompt encoding, sampling, VAE) is a linkable, reusable block, which makes workflows reproducible and shareable in JSON format. Unlike an online generation service, self-hosting gives you control over the checkpoint models, the LoRAs, the ControlNets and the extensions, without censorship or quota. On a GPU VPS, you get a studio available 24\u002F7 that an entire creative team can use remotely, and whose API lets you industrialize generation from your own scripts or pipelines.",{"type":40,"title":41,"items":42},"ul","Concrete benefits of self-hosting",[43,44,45,46,47,48],"Reproducible node-based workflows, exportable as JSON and shareable within the team","Free library of checkpoints, LoRAs and ControlNets with no quota or censorship","HTTP API to automate generation from your scripts and pipelines","Remote GPU accessible 24\u002F7 without tying up a local workstation","Installation of custom nodes (community extensions) without restriction","Cost control: a GPU VPS by the hour or the month rather than paying per image",{"type":36,"title":50,"body":51},"Hardware requirements by use case","Sizing depends directly on the execution mode and target models. In CPU mode (slow, for testing and prototyping), a 4 vCPU VPS with 8 GB RAM is enough to load an SDXL checkpoint, but expect several minutes per image. In GPU mode, the bottleneck is VRAM: 8 GB of NVIDIA VRAM can run SDXL in fp16 thanks to dynamic VRAM management (DynamicVRAM), enabled by default on NVIDIA GPUs with PyTorch 2.8 or later; 12 to 16 GB of VRAM are recommended for Flux.1 at full precision. For storage: an SDXL checkpoint weighs around 6 to 7 GB, Flux.1 schnell (full precision) is 23.8 GB, its fp8 version 17.2 GB. Plan for at least 50 GB of SSD, ideally 100 GB if you intend to store multiple models and their LoRAs. System RAM must be at least 16 GB when GPU-to-RAM offloading is active.",{"type":53,"title":54,"headers":55,"rows":61},"comparison","Minimum configuration by scenario",[56,57,58,59,60],"Scenario","vCPU","RAM","VRAM","Disk",[62,68,73,79],[63,64,65,66,67],"CPU test (SDXL, slow)","4","8 GB","— (no GPU)","50 GB",[69,64,70,71,72],"GPU SDXL comfortable","16 GB","8 GB NVIDIA","80 GB",[74,75,76,77,78],"GPU Flux.1 (recommended)","8","32 GB","16 GB NVIDIA","100 GB",[80,81,82,83,84],"Production multi-user","8+","32 GB+","24 GB NVIDIA","200 GB+",{"type":36,"title":86,"body":87},"Two installation methods: Docker vs Python venv","ComfyUI can be deployed two ways: via Docker (isolation, reproducibility, simplified GPU dependency management) or via a Python virtual environment (closer to bare-metal, more flexible for experimental custom nodes). On a production VPS, Docker is recommended for ease of maintenance and version isolation.",{"type":89,"title":90,"steps":91},"steps","Method A — Python venv installation (direct hardware access)",[92,95,98,101],{"title":93,"body":94},"Install system dependencies","On Ubuntu 24.04 (Python 3.12 is in its official repositories, not in those of 22.04): `apt update && apt install -y git python3.12 python3.12-venv python3-pip`. ComfyUI supports Python 3.12 and 3.13; version 3.13 is very well supported, 3.14 may cause compatibility issues with some custom nodes.",{"title":96,"body":97},"Clone the repository and create the venv","`git clone https:\u002F\u002Fgithub.com\u002FComfy-Org\u002FComfyUI.git \u002Fopt\u002Fcomfyui && cd \u002Fopt\u002Fcomfyui && python3.12 -m venv venv && source venv\u002Fbin\u002Factivate && pip install -r requirements.txt`",{"title":99,"body":100},"Install PyTorch with CUDA or CPU support","For NVIDIA GPU (CUDA): `pip install torch torchvision --index-url https:\u002F\u002Fdownload.pytorch.org\u002Fwhl\u002Fcu130` (ComfyUI requires a CUDA 13.0 or later PyTorch build on NVIDIA 20 series GPUs and newer). For CPU-only mode: `pip install torch torchvision`. PyTorch 2.7 is the minimum supported version; a newer version is strongly recommended.",{"title":102,"body":103},"Launch ComfyUI","GPU mode: `python main.py --listen 127.0.0.1`. CPU mode: `python main.py --cpu --listen 127.0.0.1`. With `--listen 127.0.0.1` (the default), ComfyUI only listens on the VPS loopback interface: that is enough for the SSH tunnel and the reverse proxy, and nothing is exposed without authentication. The interface is available on port 8188.",{"type":89,"title":105,"steps":106},"Method B — Docker deployment with GPU",[107,110,113],{"title":108,"body":109},"Prepare the GPU VPS","On a VPS with an NVIDIA GPU, install the drivers then the NVIDIA Container Toolkit so that Docker can access the GPU. Validate with `docker run --rm --gpus all nvidia\u002Fcuda:12.4.0-base-ubuntu22.04 nvidia-smi`.",{"title":111,"body":112},"Run ComfyUI in a container","Start a ComfyUI image with GPU access and persistent volumes: `docker run -d --gpus all -p 127.0.0.1:8188:8188 -v \u002Fopt\u002Fcomfyui\u002Fmodels:\u002Fopt\u002FComfyUI\u002Fmodels -v \u002Fopt\u002Fcomfyui\u002Foutput:\u002Fopt\u002FComfyUI\u002Foutput --name comfyui ghcr.io\u002Fai-dock\u002Fcomfyui:latest-cuda`. Restricting to 127.0.0.1 avoids direct exposure. Adjust the image tag to match your VPS CUDA version.",{"title":114,"body":115},"Verify the GPU is detected","After startup: `docker logs comfyui | grep -i 'cuda\\|gpu\\|device'`. ComfyUI displays the selected device at startup. If you see `Using CPU`, your GPU is not accessible from the container — check the NVIDIA Container Toolkit.",{"type":36,"title":117,"body":118},"Downloading models from HuggingFace","Models are downloaded from HuggingFace using `wget` or the HuggingFace CLI (`pip install huggingface_hub`). Each file type has its dedicated folder in the ComfyUI tree. For SDXL: place the checkpoint `.safetensors` file in `models\u002Fcheckpoints\u002F`. For Flux.1: the architecture differs — the diffusion model goes in `models\u002Fdiffusion_models\u002F` (or `models\u002Funet\u002F` depending on the version), and Flux requires two text encoders in `models\u002Ftext_encoders\u002F`: `clip_l.safetensors` and `t5xxl_fp16.safetensors` (or `t5xxl_fp8_e4m3fn_scaled.safetensors` to save VRAM). The VAE (`ae.safetensors`) goes in `models\u002Fvae\u002F`. Flux.1 schnell, under the Apache 2.0 license, is published in `black-forest-labs\u002FFLUX.1-schnell` on HuggingFace (23.8 GB full precision); Comfy-Org publishes a 17.2 GB fp8 version in `Comfy-Org\u002FFlux-1-Schnell`. Black Forest Labs repositories are gated: you must accept the conditions on HuggingFace before downloading, and Flux.1 dev is also under a non-commercial license.",{"type":89,"title":120,"steps":121},"Nginx reverse proxy with authentication",[122,125,128],{"title":123,"body":124},"Create the basic auth file","`apt install -y apache2-utils && htpasswd -c \u002Fetc\u002Fnginx\u002F.htpasswd your_user`. ComfyUI has no native authentication: without this step, your instance is open to everyone.",{"title":126,"body":127},"Configure the Nginx virtual host","Create `\u002Fetc\u002Fnginx\u002Fsites-available\u002Fcomfyui` with: `server { listen 443 ssl; server_name comfy.yourdomain.com; ssl_certificate \u002Fetc\u002Fletsencrypt\u002Flive\u002Fcomfy.yourdomain.com\u002Ffullchain.pem; ssl_certificate_key \u002Fetc\u002Fletsencrypt\u002Flive\u002Fcomfy.yourdomain.com\u002Fprivkey.pem; auth_basic \"ComfyUI\"; auth_basic_user_file \u002Fetc\u002Fnginx\u002F.htpasswd; location \u002F { proxy_pass http:\u002F\u002F127.0.0.1:8188; proxy_read_timeout 300s; proxy_send_timeout 300s; proxy_http_version 1.1; proxy_set_header Upgrade $http_upgrade; proxy_set_header Connection \"upgrade\"; proxy_set_header Host $host; } }`. The WebSocket upgrade is mandatory for ComfyUI's real-time API.",{"title":129,"body":130},"Obtain the Let's Encrypt certificate and enable","`certbot --nginx -d comfy.yourdomain.com && ln -s \u002Fetc\u002Fnginx\u002Fsites-available\u002Fcomfyui \u002Fetc\u002Fnginx\u002Fsites-enabled\u002F && nginx -t && systemctl reload nginx`. Then close port 8188 at the firewall: `ufw deny 8188`.",{"type":36,"title":132,"body":133},"Installing ComfyUI Manager and custom nodes","ComfyUI Manager is the essential extension for managing custom nodes from the graphical interface. In Python venv: `cd \u002Fopt\u002Fcomfyui && source venv\u002Fbin\u002Factivate && pip install -r manager_requirements.txt`. Then relaunch ComfyUI with `python main.py --enable-manager --listen 127.0.0.1`. A \"Manager\" icon appears in the interface: you can install, update and disable popular custom nodes (WAS Node Suite, ControlNet Preprocessors, IP-Adapter, etc.) without any command line. In Docker, mount a volume on `custom_nodes\u002F` so that installations survive container restarts.",{"type":53,"title":135,"headers":136,"rows":140},"ComfyUI versus AUTOMATIC1111 (Stable Diffusion WebUI)",[137,138,139],"Criterion","ComfyUI","AUTOMATIC1111",[141,145,149,153,157,161,165,169],[142,143,144],"Approach","Visual node-based workflows","Classic tabbed interface",[146,147,148],"Reproducibility","Excellent (workflow exported as JSON)","Limited to the entered parameters",[150,151,152],"VRAM consumption","Optimized, handles small GPUs better","More demanding at equal configuration",[154,155,156],"Learning curve","Steeper (graph logic)","More approachable for beginners",[158,159,160],"API automation","Native and granular","API present but less flexible",[162,163,164],"Recent models (Flux, SD3)","Fast, reference-grade support","SD3 since 1.10.0, no native Flux support",[166,167,168],"Custom nodes \u002F extensions","Very rich node ecosystem","Large extension catalog",[170,171,172],"Ideal use case","Advanced pipelines and automation","Fast interactive generation",{"type":36,"title":174,"body":175},"Troubleshooting: 4 common errors","On a freshly configured VPS, several errors come up consistently. Here are the causes and fixes.",{"type":40,"title":177,"items":178},"Frequent errors and solutions",[179,180,181,182],"**CUDA not available \u002F Using CPU**: ComfyUI did not detect a GPU. Causes: PyTorch installed without CUDA support (`pip install torch` without the CUDA index), or missing NVIDIA drivers. Check with `python -c \"import torch; print(torch.cuda.is_available())\"`. If `False`, reinstall PyTorch with `--index-url https:\u002F\u002Fdownload.pytorch.org\u002Fwhl\u002Fcu130`. In Docker, verify that the NVIDIA Container Toolkit is installed and that you launch with `--gpus all`.","**CUDA out of memory (OOM)**: the model does not fit in VRAM. On NVIDIA GPUs with PyTorch 2.8 or later, dynamic VRAM management is on by default and `--lowvram` has no effect; otherwise (or with `--disable-dynamic-vram`), add `--lowvram`, which runs the text encoders on the CPU. For Flux on 8 GB VRAM, also use the fp8 variant of the model. As a last resort, `--novram` offloads everything to RAM (very slow). Reducing the generation resolution (512×512 instead of 1024×1024) also helps immediately.","**ERROR: Could not find model \u002F model not found**: the file is not in the right place. ComfyUI looks for checkpoints in `models\u002Fcheckpoints\u002F`, Flux diffusion models in `models\u002Fdiffusion_models\u002F` (or `models\u002Funet\u002F`), text encoders in `models\u002Ftext_encoders\u002F`. A `.safetensors` file in the wrong subfolder will not appear in the interface. Refresh the list with the \"Refresh\" button in the model loading node.","**Port 8188 already in use**: a ComfyUI process or another application is already using the port. `lsof -i :8188` identifies the PID. Launch ComfyUI on another port with `--port 8189` and update your Nginx config accordingly. In Docker, the conflict may come from a stopped but not removed container: `docker rm comfyui` before restarting.",{"type":184,"body":185},"tip","To industrialize generation, leverage the API: submit your workflows via POST to `\u002Fprompt` and retrieve the results through the `\u002Fws` WebSocket that notifies the end of each task. On 8 GB of VRAM, SDXL runs thanks to dynamic VRAM management, enabled by default on NVIDIA GPUs with PyTorch 2.8 or later; `--lowvram` only matters when it is disabled. For remote access without a certificate (development), use an SSH tunnel: `ssh -L 8188:localhost:8188 user@your-vps` — ComfyUI stays accessible on `http:\u002F\u002Flocalhost:8188` from your workstation without any public exposure.","Set up your remote ComfyUI studio","Image generation demands a GPU and generous storage for the checkpoints. The ServOrbit Cloud VPS offers the resources and preconfigured Docker to deploy ComfyUI and expose it securely.","Deploy on a Cloud VPS","\u002Fvps-cloud",[191,207,225],{"id":23,"slug":192,"slugs":193,"title":197,"excerpt":198,"readTime":17,"views":23,"isPinned":19,"noindex":19,"publishedAt":20,"updatedAt":199,"category":200,"categories":201,"featuredImage":30,"bgImage":31,"posterImage":203,"relatedSolution":204},"how-to-host-open-webui-on-a-vps",{"fr":194,"en":192,"ar":195,"es":196},"heberger-open-webui","كيفية-استضافة-open-webui-على-خادم-vps","como-alojar-open-webui-en-un-vps","How to Host Open WebUI on a VPS","A complete guide to deploying Open WebUI on your own VPS. Keep control of your data, avoid cloud API costs, and access your LLMs from a clean web interface.","2026-11-09T20:55:18+00:00",{"id":23,"name":24,"slug":25,"color":26,"icon":27},[202],{"id":23,"name":24,"slug":25,"color":26,"icon":27},"\u002Fblog\u002Fcovers\u002Fheberger-open-webui-poster.svg",{"categorySlug":205,"appSlug":206},"artificial-intelligence","open-webui",{"id":208,"slug":209,"slugs":210,"title":214,"excerpt":215,"readTime":216,"views":23,"isPinned":19,"noindex":19,"publishedAt":217,"updatedAt":218,"category":219,"categories":220,"featuredImage":30,"bgImage":31,"posterImage":222,"relatedSolution":223},2,"deploying-dify-on-a-vps-ai-workflows-under-control",{"fr":211,"en":209,"ar":212,"es":213},"deployer-dify-vps","نشر-dify-على-vps-سير-عمل-الذكاء-الاصطناعي-تحت-السيطرة","desplegar-dify-en-un-vps","Deploy Dify on a VPS: Complete Guide from Installation to Plugins","Install Dify on a ServOrbit VPS: requirements, Docker Compose, offline plugins, known Docker limits, backup before upgrading, and troubleshooting.",15,"2026-02-11T00:00:00+00:00","2026-10-29T03:10:23+00:00",{"id":23,"name":24,"slug":25,"color":26,"icon":27},[221],{"id":23,"name":24,"slug":25,"color":26,"icon":27},"\u002Fblog\u002Fcovers\u002Fdeployer-dify-vps-poster.svg",{"categorySlug":25,"appSlug":224},"dify",{"id":226,"slug":227,"slugs":228,"title":232,"excerpt":233,"readTime":18,"views":234,"isPinned":19,"noindex":19,"publishedAt":235,"updatedAt":236,"category":237,"categories":238,"featuredImage":30,"bgImage":31,"posterImage":240,"relatedSolution":30},20,"how-to-host-stable-diffusion-webui-on-a-vps",{"fr":229,"en":227,"ar":230,"es":231},"heberger-stable-diffusion-webui-vps","كيفية-استضافة-stable-diffusion-webui-على-خادم-vps","como-alojar-stable-diffusion-webui-en-un-vps","How to host Stable Diffusion WebUI on a VPS","Deploy Stable Diffusion WebUI (AUTOMATIC1111) on your own VPS: private, unlimited AI image generation with no subscription.",4,"2026-05-31T00:00:00+00:00","2026-09-07T11:26:10+00:00",{"id":23,"name":24,"slug":25,"color":26,"icon":27},[239],{"id":23,"name":24,"slug":25,"color":26,"icon":27},"\u002Fblog\u002Fcovers\u002Fheberger-stable-diffusion-webui-vps-poster.svg",true,1794270422440]