Your own private photo library with AI that runs on your hardware.

Immich

Immich is like Google Photos, but you control it. Your photos stay on your server, AI features run locally, and you get automatic backups, single sign-on, and monitoring built in.

Immich is an open-source project. We make it easier to deploy and manage, but all credit goes to the original developers. Learn more about the Immich project →

Unified authentication with Auth0 (GitHub, email, passkeys)
Automatic encrypted backups that test themselves weekly
Works with or without GPUs—CPU mode is fine for most people

What you get

All the features you'd expect, plus the control and privacy of self-hosting.

AI that runs on your hardware

Face recognition, search by what's in the photo, and object detection all happen on your server. No sending photos to Google or Apple—everything stays private.

Share with family and teams

Create shared albums, set expiration dates on links, and control who sees what. Perfect for families or small teams who want to collaborate without accidentally sharing too much.

Control how long photos are kept

Set automatic deletion rules, create archives, or keep everything forever. You decide what happens to your photos.

What's included

Built-in safety features

Login integration, automatic backups, and monitoring come included. Everything runs on your server; we handle the setup and give you clear guides for when things go wrong.

Unified authentication

Single sign-on across all apps using Auth0. Sign in with GitHub, email, or passkeys. One account, all your apps—no need to create separate passwords.

Automatic backups that verify themselves

Your photos get backed up nightly, and every week we test that the backups actually work. You'll get an email if something's wrong.

See what's happening

Built-in dashboards show you GPU usage, how fast photos are processing, and if anything needs attention. Get alerts before problems become disasters.

Resource requirements

Plan your deployment with these hardware requirements. All tiers include overhead for Docker and supporting services.

Heavy Resource Usage

Minimum Configuration

Good for testing and small-scale use

CPU
2 cores
RAM
4GB
Storage
20GB + photo library
Capacity
1-5 users

Important notes

  • Photo library grows 10-20% from thumbnails and transcoded videos
  • Postgres database requires 1-3GB, must use local SSD (never network shares)
  • Machine learning features work on CPU but GPU accelerates processing
  • When using Docker resource limits, allocate minimum 2GB RAM to Postgres

How it works

Here's how everything fits together. All the setup files are in the docs if you want to customize things.

How photos get in

Automatic SSL certificates, secure connections, and everything runs through your private tunnel. No port forwarding needed.

Where photos are stored

Stores photos on your server with options to offload older stuff to cheaper storage. Built-in checksums make sure nothing gets corrupted.

Photo processing

AI features run in containers. If you have a GPU, it'll use it. Otherwise CPU works fine—just slower. Everything scales automatically.

Get started in three steps

Use the portal to deploy your app, set it up, and start using it—all through your web browser.

01

Enroll your device

Download the agent installer from the portal dashboard. The installer handles Docker setup and connects your device to your UnboundBytes account automatically.

02

Deploy Immich

Select Immich from the application catalog, choose your deployment target, and click deploy. The portal configures everything—containers, tunnels, and SSL certificates.

03

Access your instance

Your Immich instance is available at {yourname}.unboundbytes.com/immich. Share the URL with your team—no port forwarding or firewall configuration needed.

Common questions

Still have questions? Join our community chat or check out the support page for more help.

Can we run without GPUs?

Yes. CPU inference is supported with smaller models. You can add GPUs later and Immich will auto-detect them.

Where are backups stored?

Wherever you choose—S3-compatible object storage, on-premises cold storage, or Cloudflare R2. You hold the keys; we provide automation.

How do updates work?

We ship curated release channels. Approve updates in the portal and the agent performs rolling restarts with automatic rollback if health checks fail.

Learn more

Check out the docs, upstream projects, and support channels.