Start here
Test install
- 2–4 cores
- 4 GB RAM
- 20 GB free space
- SSD recommended
Good for trying Arciin and testing uploads.
Your server, your control.
Install
No installer to babysit, no cloud account. Paste one line on your Linux server — Arciin prepares PostgreSQL, Redis, storage, API, web, and worker for you.
One-line install$ curl -fsSL https://get.arciin.com/install.sh | bashDebian · Ubuntu · WSL · Docker · then open http://localhost
install.sh
$ curl -fsSL https://get.arciin.com/install.sh | bash
What the script does
- Checks your OS and installs system dependencies
- Creates the storage layout under /srv/arciin-storage/arciin
- Runs database migrations and starts all services
- Prints the local URL and setup token for first claim
Claim your Arciin instance.
Setup
After the script finishes, open the printed URL on your LAN. Enter your setup token, name the instance, create the owner account, and choose where files live. Setup locks forever after the first successful claim — no public signup, no cloud account.
First user becomes OWNERarciin.local/setup
Default libraries appear immediately — Videos, Images, Music, Documents, and Inbox. Drag files anywhere in the dashboard and Arciin classifies them for you.
Install flow
From zero to your private command center.
Four steps — most people are past step two in under five minutes on a fresh Linux box.
Step 01
Run the install script
One curl command on Debian, Ubuntu, WSL, or Docker. Dependencies, storage paths, and services are configured automatically.
api · web · worker · postgres · redis
01
Run the install script
One curl command on Debian, Ubuntu, WSL, or Docker. Dependencies, storage paths, and services are configured automatically.
02
Open your instance
Visit the printed URL on your LAN — typically http://localhost:3000 or your server IP. No external signup.
03
Claim with your setup token
Enter ARCIIN_SETUP_TOKEN, name your instance, create the owner account, and choose your storage root.
04
Start using Arciin
Default libraries appear, uploads work globally, mobile PWA can pair, and you can connect AI models on the Models page.
Install methods
Native or Docker — your call.
Fastest on a fresh Debian or Ubuntu server (or WSL). Installs dependencies and starts all services for you with one command.
$ curl -fsSL https://get.arciin.com/install.sh | bashSystem requirements
Start small. Scale with your library.
Arciin runs on your hardware. Cloud AI works on normal servers. Local AI, large libraries, and 4K transcoding need real RAM, disk, and often a GPU.
Supported OS
Install native or with Docker Compose
Native installDocker Compose
Debian 12+Ubuntu 22.04+Ubuntu 24.04+WSL2Any Linux + Docker
Recommended
Best for most peopleRecommended server
- 4 cores
- 8 GB RAM
- 100 GB+ storage
- SSD/NVMe for app + database
- HDD, external drive, or NAS for files
Best for daily self-hosted use — libraries, thumbnails, metadata jobs, mobile uploads, vault, APIs, and Plex/Jellyfin folder routing.
Local AI
Local AI server
- 6–8 cores
- 16 GB RAM minimum
- 8 GB+ VRAM recommended
- SSD/NVMe for models and cache
Best for small local models (e.g. Ollama) and personal AI workflows. Cloud AI still works on lighter hardware.
Media + AI
Media + local AI
- 8+ cores
- 32 GB RAM recommended
- 12–16 GB+ VRAM recommended
- Wired Ethernet recommended
Best for larger libraries, local AI, vision, and media workflows where model + media jobs share one machine.
Production
Power user / team
- 8–12+ cores
- 32–64 GB RAM
- GPU/iGPU recommended for transcoding
- SSD/NVMe + large HDD/NAS
Best for teams, production use, heavier 4K media workflows, and multiple concurrent jobs.
Storage & network
Where files live and how you reach them.
Specs above are about compute. These details cover disks and connectivity — separate from which hardware tier you pick.
- Storage
- Any mounted disk, external drive, NAS mount, or server storage path
- Docker: bind-mount any location into the container
- Default native path:
/srv/arciin-storage/arciin
- Network
- LAN first — browse and upload with no cloud account
- Optional remote access: domain, reverse proxy, or Cloudflare Tunnel
- Wired Ethernet recommended for media-heavy and team servers
Before you size hardware
Cloud AI, local AI, and 4K media.
These notes sit outside the size cards so you can match the tier above to how you actually run AI and media.
Cloud AI vs local AI
Cloud AI works on normal Arciin hardware because the model runs through your connected provider. Local AI runs on your server, so RAM, VRAM, model size, quantization, and context length matter. For local AI, 16 GB RAM is the starting point; 32 GB+ is recommended for a smoother experience with larger models, vision, or media workflows.
4K media note
Arciin organizes your media folders. Plex or Jellyfin handles playback. Direct play is usually light, but 4K transcoding can require a strong CPU or hardware acceleration. For serious media use, use wired Ethernet and a GPU or iGPU that supports transcoding.
After install
What to do first.
Pair mobile PWA
Install the companion app on your phone and pair with a six-digit code on your network.
Connect AI models
Add OpenAI, Anthropic, Gemini, DeepSeek, or local Ollama keys on the Models page.
Drag-and-drop uploads
Drop files anywhere in the dashboard — Arciin classifies and routes them to libraries.
Invite your team
Create local users with Owner, Admin, Member, or Viewer roles — no per-seat billing.