Jellyfin vs Plex in 2026: Which Media Server Fits Your Home?

Illustration of a home media server streaming movies to a living room TV

You have a drive full of movies, home videos and old TV box sets. You want them on the living room TV, the kids' tablets and your phone. In the Jellyfin vs Plex choice, the short answer is this. Pick Jellyfin if you want every feature free, no account and no company in the middle. Pick Plex if you want the most polished apps and the easiest setup, and you accept that watching away from home now needs a paid plan.

Who this is for: you have (or plan to have) a home server, a NAS or a spare computer, and you want your own Netflix-style library. Who it is not for: anyone who only streams from paid services and owns no media files. A media server only serves the files you already have.

What a media server does

A media server is a program that runs on one computer in your home. It scans your video and music folders, adds posters and descriptions, and streams the files to apps on your TVs, phones and browsers. Each person then watches in an app, called a client.

One word comes up a lot: transcoding. That means the server converts a video on the fly, because the device cannot play the original file or the connection is too slow. Transcoding is hard work for a processor. A graphics chip can do it much more easily, which is called hardware acceleration. It is one of the real differences between the two.

If you are new to running services at home, start with the guide to self-hosting in 2026, which explains what it really takes. Then come back here. A media server is also one of the most popular home lab projects, so it makes a good first service.

Jellyfin and Plex in one paragraph each

Jellyfin

Jellyfin is free and open-source software. Open source means anyone can read, change and share the code. As of October 2026, the Jellyfin website says the server and official apps are "free to download, now and always", with no hidden costs. It also says Jellyfin has no tracking and no central servers that collect your data. A volunteer community builds it. The project started in December 2018, when Emby (an older media server) moved its next release to closed source.

Plex

Plex is made by a company. The server is free to download, and so is watching your own videos at home. Plex also runs a free, ad-supported service with on-demand movies and live TV channels. The paid part is called Plex Pass. It unlocks premium features, such as hardware-accelerated streaming, downloads and skip-intro buttons. Since 29 April 2025, Plex also sells a cheaper Remote Watch Pass that only unlocks streaming away from home.

Jellyfin vs Plex: side by side

This table reflects the vendors' own pages as of October 2026. Plans change, so check them before you decide.

FactorJellyfinPlex
Cost modelFree. No paid tier.Free tier, plus paid Plex Pass (monthly, yearly or lifetime) and Remote Watch Pass
CodeOpen source (GNU GPL)Closed source
Account neededNo. Users live on your own server.Yes, in normal setup. You sign the server in to a Plex account.
Watching at homeFreeFree
Watching away from home (video)Free, but you set up the connection yourselfNeeds Plex Pass on the server owner's account, or a Plex Pass or Remote Watch Pass on the viewer's account
Hardware transcodingFreePlex Pass feature
AppsOfficial apps for Android, Android TV and Fire TV, iOS, Roku, LG, Samsung, Xbox and desktopApps for most phones, TVs, streaming sticks and consoles
Extra contentYour own media onlyYour media plus free ad-supported movies and live TV
Setup effortModerateEasy

The big 2025 and 2026 change: Plex remote streaming

This is the change that sent many people searching for a Plex alternative. Plex's support page on requirements for remote playback of personal media sets the rule. Since 29 April 2025, to stream your own video from a Plex server when you are not on the same home network, one of these must be true:

  • The server owner's account has an active Plex Pass. Then everyone who streams from that server is covered.
  • Your own account has an active Plex Pass.
  • Your own account has an active Remote Watch Pass.

The rule covers video only. Plex says it does not apply to music in its Plexamp app or to photos in its Plex Photos app. Plex rolled the rule out app by app. In an update to its blog post dated 29 April 2026, Plex said the rule now also applies on Amazon Fire TV, supported smart TVs such as Samsung, LG and Vizio, and PlayStation and Xbox consoles. It says all apps will follow in time.

What this means for you: if Grandma watches your movies from her house, someone now pays. If everyone watches at home, the free tier still works.

Plex has raised prices several times. Plex Pass went up on 29 April 2025, the plans page set a further lifetime price rise from 1 July 2026, and the Remote Watch Pass introductory price ended on 1 June 2026. I do not quote prices here because they keep moving. Check the current figures on the Plex plans page.

Apps and TVs

As of October 2026, the Jellyfin clients page lists official apps for Android phones, Android TV and Fire TV, iPhone and iPad, Roku, LG webOS TVs, Samsung Tizen TVs, Xbox, a desktop player and a Kodi add-on. For Apple TV, the Jellyfin project builds Swiftfin, a native app for iOS and tvOS. Many people also use third-party players with Jellyfin.

Plex has apps for a very wide range of devices, and in my view they are the more polished of the two. That is the honest trade. A paid team builds Plex. Volunteers build Jellyfin, and some Jellyfin TV apps feel plainer.

Before you choose, list every screen in your home. Install both apps on the one that matters most (usually the living room TV) and try them. Both servers are free to test at home.

Remote access: watching away from home

Plex handles the connection for you. You turn on Remote Access in the server settings, and Plex helps your apps find the server through your Plex account. Then the paid rules above apply for video.

Jellyfin has no company in the middle, so you build the path yourself. The Jellyfin networking docs say that opening a port straight to the internet is insecure and not recommended. Port forwarding means telling your router to send outside traffic to a device inside your home. The docs point you to a reverse proxy (a front door program that adds HTTPS encryption) or a VPN instead.

For most families, a private VPN is the simplest safe choice. Tailscale does this with no open ports. The guide Tailscale explained walks through it in plain steps.

Hardware transcoding

  • Jellyfin: hardware acceleration is free. As of October 2026, the Jellyfin docs list Intel Quick Sync Video, NVIDIA NVENC, AMD AMF, VA-API on Linux, Apple VideoToolbox on macOS and Rockchip on Linux.
  • Plex: Plex's support page says hardware-accelerated streaming is a premium feature that needs an active Plex Pass. It recommends Intel Quick Sync Video, and also supports NVIDIA cards on Windows and Linux. Plex says AMD support is not fully tested.

Intel Quick Sync is the video engine built into many Intel processors, including the small chips in many mini PCs. It is a common reason people pick an Intel machine for a media server.

Privacy and accounts

With Jellyfin, user accounts live on your server. Nothing leaves your home unless you set up remote access. With Plex, you sign the server in to a Plex account, and sign-ins go through Plex. That is what makes Plex easy, and it is also a dependency.

Two facts are worth knowing. In March 2025, Plex updated its privacy policy and terms with more detail on the data it collects and how it may use or share it. In September 2025, Plex emailed users to reset their passwords after a security incident. Plex said the exposed data included emails, usernames and hashed (scrambled) passwords. It also said it stores no payment card data. If you use Plex, turn on two-factor sign-in. Read the current Plex privacy policy yourself before you decide.

What you need

  • Hardware: an always-on computer with enough storage for your media. A mini PC with an Intel chip, an old desktop or a NAS that supports the server software all work. The home server setup guide helps you pick a hardware class.
  • Skill: Plex needs very little. Jellyfin needs a bit more, mostly for remote access. Many people run Jellyfin in Docker. The guide to containers vs virtual machines explains where that fits.
  • Time: plan an evening to install and scan a library. Messy file names take longer, because both servers use names to find the right posters.
  • Cost: Jellyfin has no license cost. For Plex, the cost depends on whether anyone watches away from home and whether you want premium features. See the plans page.

Choose Jellyfin if...

  • You want every feature, including hardware transcoding and remote streaming, with no subscription.
  • You do not want an outside account between your family and your files.
  • You are happy to set up a VPN or reverse proxy for remote viewing.
  • Your main screens have official Jellyfin apps, and you tested them.

Choose Plex if...

  • You want the easiest setup and the most polished apps, and that is worth paying for.
  • Everyone watches at home, so the free tier covers you.
  • You like the free ad-supported movies and live TV next to your own library.
  • Family members on other networks already have Plex accounts, and someone will pay for remote viewing.

What about Emby?

Emby is the third option. It has a free server and a paid subscription called Emby Premiere. As of October 2026, Emby's page says Premiere unlocks hardware-accelerated transcoding, DVR recording, offline downloads and the full mobile apps. According to the Jellyfin project, Emby moved to closed source with its 4.x release. Emby suits people who want a Plex-style paid product with less of a central account. Check the Premiere page for current terms.

The risky bit

  • Opening a port to the internet. This is the most common mistake. Use a VPN or a proper reverse proxy with HTTPS.
  • Weak passwords on a shared server. Give each person their own user, with a strong password. Turn off accounts you do not use.
  • One drive holds everything. If that drive dies, the library goes with it. A media server is not a backup.

The recovery step: back up the library and the settings

Two things need protection. The first is your media files. Keep a second copy on another drive, and ideally a third copy in another place. Home videos matter most, because you cannot download them again. The second is the server's config folder. It holds watch history, users and settings. Back it up too, so a rebuild takes minutes, not a weekend of rescanning.

Then test it. Restore the config to a fresh install once, and check that your users and watch history come back.

FAQ

Is Jellyfin really free?

Yes. As of October 2026, the Jellyfin project says the server and official apps are free, with no paid tier and no hidden costs. Your costs are the hardware and electricity.

Is Plex still free in 2026?

Partly. The server is free, and so is streaming your own video at home. Streaming your own video away from home now needs a Plex Pass or a Remote Watch Pass, as of October 2026. Check the Plex plans page for the current terms.

Can I watch Jellyfin away from home?

Yes, at no cost. You must set up the connection yourself. The safest simple way for most people is a private VPN, rather than opening a port on your router.

Does Jellyfin work on Roku and smart TVs?

As of October 2026, Jellyfin lists official apps for Roku, LG webOS, Samsung Tizen, Android TV and Fire TV. Test the app on your own TV before you move the whole family over.

Can I run Jellyfin and Plex on the same server?

Yes. Both can read the same media folders. This is a good way to compare them with your real library and your real TVs.

Is Emby better than Jellyfin?

It depends on your criteria. Emby is a paid product for its premium features, while Jellyfin is free and open source. Jellyfin started as a fork of Emby in 2018, so they share a common past.

Related guides

Snapshot vs Backup: How to Prove Your Home Server Backup Actually Works

Illustration comparing a snapshot on the same server with backups stored elsewhere and a tested restore

You took a snapshot of your home server before an update, so you feel safe. Then the disk dies, and the snapshot dies with it. That is the core of snapshot vs backup. A snapshot is a quick undo point that lives on the same storage as the original. A backup is a separate, complete copy you can restore somewhere else. You need both, and you only know a backup works after you restore it once.

Who this is for: you run a home server, a NAS or a Proxmox box with things you care about, such as family photos, a password vault or a smart home setup. Who it is not for: you only run throwaway test machines that you can rebuild in ten minutes. You can still read on, but you need less of it.

What a snapshot is

A snapshot freezes the state of a virtual machine (VM) or a disk at one moment. After that, the system writes new changes to a separate layer. If an update goes wrong, you roll back, and the VM returns to the frozen state, usually quickly. A rollback also throws away everything written since the snapshot, such as new photos or passwords. Save or back up recent changes before you roll back.

Proxmox can snapshot both VMs and containers, if the storage supports it. The guide to containers vs virtual machines explains the two. Snapshots are great for one job: a short-term undo button before a risky change. Take one before an upgrade. Test the upgrade. Then delete the snapshot when you are happy.

What a backup is

A backup is a full, independent copy of your data, stored somewhere else. It does not need the original disk to exist. If the server burns, you can restore the backup onto new hardware.

In Proxmox VE, the docs say plainly that its backups "are always full backups", with the VM or container configuration and all data. See the Proxmox VE backup and restore docs. Tools such as Proxmox Backup Server then store those backups in a smart way, so later runs only send what changed.

Snapshot vs backup: four reasons a snapshot is not a backup

  1. It lives on the same storage. A snapshot sits on the same disk or storage pool as the VM. If that disk fails, you lose the VM and every snapshot at once.
  2. It depends on the original. Most snapshots store only the changes since the freeze. They are not a complete copy, so they mean nothing without the original disk.
  3. It shares the same mistakes. Delete the VM by accident, and its snapshots usually go too. Ransomware (malware that locks your files and asks for money) that reaches the host can reach the snapshots too.
  4. It is not off-site and not long-term. Snapshots stay in the house, on the box. Old snapshots also grow as data changes, and they can slow the VM and fill the disk.

One more trap. An older Proxmox wiki page on live snapshots notes that "snapshots are not included in the backup", and current backup logs still report snapshots as not included. So a backup does not save your snapshot history either. Treat the two as separate tools.

A word on naming: Proxmox has a backup mode called "snapshot mode". It is still a real backup. The name only means the backup runs while the VM keeps working, so you have little downtime.

QuestionSnapshotBackup
Where does it live?Same storage as the VMSeparate storage, ideally another box and another place
Is it a complete copy?Usually no, it records changesYes
Survives a dead disk?NoYes, if stored elsewhere
Best useUndo before an updateRecover from loss, theft, fire or ransomware
How long to keepHours or daysWeeks, months or longer, by a plan

The 3-2-1 backup rule, in plain words

The US Cybersecurity and Infrastructure Security Agency (CISA) gives a simple rule. On its guide to backing up business data, it says to keep:

  • 3 copies of important files,
  • on 2 different types of storage (like a hard drive and the cloud),
  • with 1 copy stored off-site, away from your home or business.

For a home server, that can look like this. The live data sits on the server. A backup sits on a second box or a USB drive at home. A third copy sits off-site, such as a drive at a relative's house or an encrypted cloud copy. CISA also tells you to test your backups, make sure you can restore data fully and partly, and be able to roll back at least seven days. That test is the part most people skip.

Some app makers say the same. The Immich photo project's own README warns you to always follow a 3-2-1 backup plan for your photos and videos. If you followed the guide to self hosting in 2026, this is the step that makes it safe.

What you need for a first restore drill

  • Hardware: your server, plus a separate place for backups. A second small box, a NAS or an external drive all work. Do not count a second folder on the same disk.
  • Software: any backup tool that can restore. This guide uses Proxmox VE with Proxmox Backup Server as the worked example. On Synology, Unraid or TrueNAS, use the same drill: restore to a new location, never over the original, then check your marker.
  • Skill: you can click through the Proxmox web page and start a VM.
  • Time: set aside an evening. A small test VM restores faster than a big one.
  • Cost: Proxmox Backup Server is free and open source (AGPLv3), per its official introduction. Optional paid subscriptions exist for support and the enterprise update channel. Your real cost is the backup disk or second box. Check current terms on the Proxmox site, as of October 2026.

Proxmox Backup Server stores incremental, deduplicated backups (it skips data it already has), can encrypt them on your side before they leave the host, and checks them with checksums. It can also run "verify jobs" on a schedule to catch damaged backup data. It is new to many home users, so read its docs before you trust it with irreplaceable files. For the bigger picture of where this box fits, see the home server setup guide.

How to test a backup restore: a step-by-step drill

The goal is simple. Prove that a real backup turns back into a working machine, without touching the live one.

  1. Pick a test subject. Choose one VM or container that matters but is small. A password vault or a photo app is a good choice.
  2. Plant a marker. Create a small text file inside it, or note one recent photo or record. You will look for it after the restore.
  3. Run a backup now. In Proxmox VE, open the VM, go to Backup, and choose "Backup now" to your Proxmox Backup Server storage. Wait for the task to finish with no errors.
  4. Verify the backup. On Proxmox Backup Server, run a verify on that backup, or check that your scheduled verify job passed.
  5. Restore to a new ID. In Proxmox VE, open your backup storage, select the backup and choose Restore. Give it a new, unused VM ID, so you never overwrite the live VM. Check "Unique" so the copy gets new network hardware (MAC) addresses. Leave "Start after restore" unchecked.
  6. Keep it off the network. Before you start the copy, open its Hardware tab and disconnect its network card. Two copies of the same server on one network can clash. In Options, also turn off "Start at boot", or the copy may start with the host after a reboot.
  7. Start it and check it. Boot the restored VM. Log in from the console. Find your marker file or record. Open the app if you can.
  8. Try a single-file restore. On the backup storage, use the "File Restore" button to browse inside the backup and download one file. This is the restore you will need most often.
  9. Write down the results. Record the date, what you restored, how long it took and any problem.
  10. Clean up. Delete the test VM when you are done, so it does not run by mistake later.

If you prefer the command line, Proxmox VE also restores with qmrestore for VMs and pct restore for containers. Both accept a new ID. Check the docs for the exact options on your version.

The risky bit: how restore drills go wrong

  • Restoring over the live VM. Always use a new VM ID for a drill. One wrong click on the original ID replaces your live machine with an older copy.
  • Two servers, one network. A restored copy can grab the same IP address or talk to the same devices. Keep it disconnected.
  • Backups on the same disk. A "backup" stored next to the original fails with it. Move it to another device.
  • Lost encryption key. If you encrypt backups and lose the key, nobody can restore them. Store the key in at least two safe places, away from the server.
  • Snapshots left running. A forgotten snapshot keeps growing and can fill the disk. Delete snapshots once an update proves fine.
  • No off-site copy. Fire, flood and theft take every device in the house. Proxmox Backup Server can sync to a second Proxmox Backup Server (a "remote") with a sync job, which is one way to get a copy off-site.

The recovery step: make the drill a habit

One drill proves today's backup. It does not prove next month's. Put a reminder in your calendar and repeat the drill on a schedule. Rotate which VM you test. After any big change (new disk, new backup target, major upgrade), run a drill again. The guide to what Proxmox VE is covers the platform itself if you are still choosing.

Printable restore drill checklist

  1. I have 3 copies of my important data, on 2 types of storage, with 1 off-site.
  2. No backup lives on the same disk as the data it protects.
  3. Backup jobs ran without errors this week.
  4. A verify job checked the backups recently.
  5. I restored one VM or container to a new ID and it booted.
  6. I found my marker file inside the restored copy.
  7. I restored one single file from a backup.
  8. My encryption key is stored in two safe places off the server.
  9. No old snapshots are left over from past updates.
  10. I wrote down the date, the time it took and any problems.
  11. The next drill is in my calendar.

FAQ

Is a snapshot a backup?

No. A snapshot is a short-term undo point on the same storage as the original. If that storage fails, the snapshot is gone too. A backup is a separate, complete copy you can restore elsewhere.

Should I use snapshots or backups?

Use both, for different jobs. Take a snapshot right before an update so you can roll back fast. Keep scheduled backups on another device, with one copy off-site, for real disasters.

What is the 3-2-1 backup rule?

Keep 3 copies of important files, on 2 different types of storage, with 1 copy off-site. CISA recommends it for small businesses, and it works well for a home server too.

How often should I test a backup restore?

Test on a regular schedule, and after any big change to your server or backup setup. Pick a rhythm you will keep, and put it in your calendar. A small, regular drill beats a big one you never do.

Can Proxmox Backup Server restore a single file?

Yes. In Proxmox VE, the File Restore button on the backup storage lets you browse a backup stored on Proxmox Backup Server and download single files. Check the docs for your version.

Does a Proxmox backup include my snapshots?

No. Proxmox backups do not include snapshots, and the backup log says so when it finds them. The backup holds the VM's current state and configuration, not its snapshot history.

Related guides

Tailscale Explained

Illustration of devices securely reaching a home server through a private tunnel without opening ports

You are at a friend's house and want to show them a photo stored on your home server. Or you are on vacation and want to check the ad blocker is still running. Tailscale solves this. It is a service that joins your phone, laptop and home server into one private network, built on the WireGuard VPN, so they can reach each other from anywhere. You do not need to open any ports on your router.

This guide explains how Tailscale works in plain words, what the free plan covers as of October 2026, how to set it up, and what can go wrong.

Who this is for (and who it is not for)

This is for you if you run a home server, a NAS or a mini PC, and you want to reach it safely from outside the house.

This is not for you if you want to publish a public website for strangers. Tailscale is for your own devices and people you invite. It is also not for you if you refuse to depend on any outside company. In that case, read the sections on Headscale and plain WireGuard below.

If you have not built the server yet, start with the home server setup guide. If you are still picking hardware, the guide to what to look for in a mini PC home server helps.

The problem Tailscale solves

Your home router acts like a front door with no mail slot. Devices inside can go out to the internet. Nothing outside can come in, unless you let it.

The old way to let it in is port forwarding. You tell the router: "send anything that arrives on this door number to my server." It works. But that door is now open to the whole internet. Automated scanners find open ports fast, then try passwords and known bugs. If the app behind the door has a flaw, an attacker can get in.

That is why you should never expose admin pages, such as the Proxmox web interface, to the internet.

Tailscale takes a different path. Each device makes an outgoing connection, which your router already allows. Then the devices find each other and connect. Tailscale's docs say it works across firewalls and NAT (the way your router shares one public address with every device at home) "without requiring port forwarding or complex firewall rules."

What is Tailscale, and how does it work?

Tailscale has two parts.

  • The app on each device. You install it on your server, phone and laptop. It creates encrypted WireGuard tunnels to your other devices. WireGuard describes itself as "an extremely simple yet fast and modern VPN that utilizes state-of-the-art cryptography."
  • The coordination server. Tailscale runs this. It keeps a list of your devices and hands out their keys and addresses, so they can find each other. It acts like a phone book, not a switchboard.

Your devices and the people you invite form a private network. Tailscale calls this a tailnet. According to the Tailscale docs, devices connect peer to peer (directly, device to device) with end-to-end encryption.

Sometimes a direct link is not possible, for example on a strict hotel or office network. Then traffic goes through a Tailscale relay, called a DERP server. As of October 2026, you can also let one of your own devices act as a peer relay. A relay only forwards data that is already encrypted. Tailscale's docs state that private keys never leave the device, so a relay cannot decrypt your traffic.

You do not create a Tailscale password. You sign in with an account you already have. As of October 2026, Tailscale's docs list Apple, Google, GitHub, Microsoft, Okta and OneLogin, plus passkeys and custom sign-in (OIDC) providers.

Tailscale vs WireGuard and other options

Tailscale is not the only way in. Here is how the common options compare for a home user.

OptionOpen ports needed?Setup effortWho runs the coordinationBest for
TailscaleNoLow: install and sign inTailscale (the company)Most DIYers and families
Headscale + Tailscale appsYes, for your Headscale serverMedium to highYouPeople who want no outside control server
Plain WireGuardYes, one UDP port on your routerMedium: you manage keys and configsYouPeople who want full control and fewer moving parts
Cloudflare TunnelNoMediumCloudflarePublishing a web app to the public
Port forwardingYes, one per serviceLow at firstYouNever for admin pages

Plain WireGuard

Tailscale uses WireGuard under the hood. You can also run WireGuard by itself. You set up a server at home, open one UDP port on your router and create a key pair for each device. WireGuard's site says a connection is made "by exchanging very simple public keys", much like SSH keys. It is fast and lean. The cost is your time: you add each device by hand and keep the configs in order.

Cloudflare Tunnel

A Cloudflare Tunnel also avoids open ports. A small program called cloudflared makes an outgoing connection from your server to Cloudflare's network. Visitors reach your app through Cloudflare. That suits a public web page, such as a family blog. You usually need a domain managed in Cloudflare. The trade-off is that your traffic passes through Cloudflare, and anything you publish is open to the internet unless you add a login in front of it. For private admin access, a VPN such as Tailscale is the simpler fit.

Headscale: the self-hosted option

Headscale is "an open source, self-hosted implementation of the Tailscale control server." You run the phone book yourself and point the Tailscale apps at it. Its README says it covers a single tailnet, suited to personal use or a small open-source group. It also states it is not associated with Tailscale Inc. The catch: your devices must reach your Headscale server from wherever they are. That usually means opening a port or renting a small cloud server. Start with Tailscale. Move to Headscale later if you want to.

What you need: hardware, skill, time and cost

  • Hardware. Nothing new. Tailscale has apps for Linux, Windows, macOS, iOS and Android. For other devices, check the install docs.
  • Skill. You can install an app and paste one command on a Linux server.
  • Time. An evening for your first devices and settings.
  • Cost. As of 9 October 2026, the Tailscale pricing page lists a free Personal plan with up to 6 users and unlimited user devices. Some features have caps on the free plan. Plans change, so check the page before you rely on it. Paid plans exist for teams and businesses.

One caveat. Most of the Tailscale client code is open source (BSD-3-Clause license), but the coordination server is not. You depend on the company for that part, unless you use Headscale.

How to set up Tailscale: the outline

These steps show the shape of the job. Follow the current Tailscale docs, because screens change.

  1. Create your tailnet. Go to the Tailscale site and sign in with your Google, Microsoft, Apple or GitHub account. Use the account you protect best.
  2. Install it on your home server. On a Linux server or VM, the docs give a one-line install: curl -fsSL https://tailscale.com/install.sh | sh. Then run sudo tailscale up. It prints a link. Open the link and sign in to add the server.
  3. Install it on your phone or laptop. Get the app from your app store. Sign in with the same account.
  4. Find your server by name. MagicDNS gives each device a name, so you type the server's name instead of a number address. Tailscale's docs say it is on by default for tailnets created on or after 20 October 2022.
  5. Test from outside. Turn off Wi-Fi on your phone and use cellular data. Open your photo app or dashboard by the server's name.
  6. Turn off key expiry for the server. By default, device keys expire after 180 days, and the device must sign in again. Tailscale's docs suggest turning this off for trusted servers that are hard to reach. Leave it on for phones and laptops.

If your server runs Proxmox, install Tailscale inside the VM or container that runs your apps, not on the host by habit. An unprivileged LXC container also needs access to the /dev/net/tun device. Tailscale's LXC guide shows how. The guide to what Proxmox VE is explains the host and guest layers.

Two useful extras

  • Exit node. An exit node is a device that carries all your internet traffic, not just traffic to your server. Set your home server as one, and on hotel Wi-Fi your phone browses as if it were at home. On Linux, turn on IP forwarding, then run sudo tailscale set --advertise-exit-node. An admin then approves it in the admin console.
  • Subnet router. Some devices cannot run Tailscale, such as a printer or a smart TV. A subnet router is a Tailscale device that lets your tailnet reach them on your home network.

Sharing with family

You have two choices.

  • Invite them as users. They join your tailnet. The free plan counts them toward the 6-user limit listed on the pricing page. By default, they can reach every device, so set access rules first (see below).
  • Share one device. You share only the media server, for example. Tailscale's docs say the recipient gets access to "only the shared machine" and nothing else. They accept it with their own Tailscale account. Sharing is available on all plans.

For grandparents who only want the photo app, sharing one device is the cleaner choice.

The risky bit: what can go wrong

  • A weak sign-in account. Your tailnet is only as safe as the Google or Microsoft account you sign in with. Turn on two-step login there. If someone takes that account, they can add their own device to your network.
  • Open access rules. Tailscale's docs say the default policy lets every device in the tailnet reach every other device. That is fine for one person. Once you add family, write access control rules (ACLs, the list of who can reach what) so a guest phone cannot reach your Proxmox admin page.
  • Treating the tailnet as a fortress. A device on your tailnet can still be infected. Keep strong passwords on your apps, and keep the server updated.
  • A forgotten old device. A lost laptop stays in the tailnet until you remove it. Check the device list now and then.
  • Opening ports anyway. Once Tailscale works, close any old port forwards on your router. Otherwise you keep both risks.

The recovery step: removal and rollback

  • Remove a lost device. Open the admin console, find the device and remove it. Its access ends.
  • Keep a way in at home. Tailscale is an extra path, not the only path. Make sure you can still reach your server on your home network without it, in case you lock yourself out with an access rule.
  • Undo an access rule. Keep a copy of your last working policy before you edit it. If a change blocks you, paste the old one back.
  • Remove Tailscale completely. On the server, sign out with sudo tailscale logout, then uninstall it as Tailscale's uninstall guide describes. Remove the device in the admin console too.

For more ideas on what to run once you can reach your server from anywhere, see these home lab projects.

FAQ

Is Tailscale free?

As of 9 October 2026, Tailscale lists a free Personal plan with up to 6 users and unlimited user devices. Some features have caps on that plan. Check the pricing page for current terms before you rely on them.

Is Tailscale a VPN?

Yes. It is a mesh VPN built on WireGuard. Instead of all traffic going through one central server, your devices connect to each other directly where they can.

Can Tailscale see my traffic?

Tailscale's docs say traffic is encrypted end to end, and private keys stay on your devices. Even its relay servers cannot decrypt traffic. The company does run the coordination server, which knows which devices are in your tailnet.

What is the difference between Tailscale and WireGuard?

WireGuard is the VPN protocol. Tailscale is a service built on it that handles keys, addresses and connections for you. With plain WireGuard, you do that work yourself and open a port on your router.

Do I need to open ports on my router for Tailscale?

No. Each device makes an outgoing connection, which home routers allow. If a direct link fails, traffic goes through an encrypted relay instead.

Can I self-host Tailscale?

Not the official coordination server. Headscale is an open source, self-hosted replacement for it, run by a separate project. It covers a single tailnet, and your devices must be able to reach it from outside the house.

Related guides

Mini PC Home Server: What to Look For in 2026 (and What to Skip)

Illustration of a mini PC being inspected with icons for memory, storage, networking and low power

You have seen the small boxes, about the size of a sandwich, that people run their whole home server on. A good mini PC home server needs five things: a CPU that supports virtualization, enough memory (ideally upgradable), room for at least one fast SSD, a wired network port, and low power draw at idle. If you want a media server, add a CPU with Intel Quick Sync for video. This guide shows you how to check each one before you pay.

This is a criteria guide, not a tested ranking, so I do not name a "best" model. You get the questions to ask of any listing.

Who this is for (and who it is not for)

This is for you if you want one small, quiet box that runs all day. Maybe it backs up the family photos, streams movies to the TV, blocks ads for every device in the house or runs a few test virtual machines.

This is not for you if your main job is bulk storage. A mini PC usually has room for one or two drives. For years of 4K video on big hard drives, a NAS (a storage box with several drive bays) or an old desktop tower fits better. The home server setup guide compares those hardware classes. This post goes deep on one.

If you are still asking why people run their own apps at home at all, start with why people are bringing their apps home in 2026.

What to look for in a mini PC home server

Work through these eight points. The first three matter most, because you cannot fix them later.

1. The CPU: two classes in plain words

The CPU (the main chip) sets what the box can do. Mini PCs for home servers fall into two rough classes.

  • Intel N-series. Small, low-power chips. As of October 2026, Intel's own N-series comparison chart lists the N95, N100, N150, N200 and N250 with 4 cores. It lists the Core i3-N300 and i3-N305, and the Core 3 N350 and N355, with 8 cores. The N100 and the newer N150 often appear in budget listings.
  • Intel Core and AMD Ryzen laptop-class chips. More cores and more speed. They suit you if you plan to run many virtual machines (VMs, full computers made of software) or heavier apps.

Here is a rule of thumb, not a tested result. A 4-core N-series box is a common pick for a few light services, such as an ad blocker, a media server and a password manager. Check each app's own needs. The photo app Immich, for example, recommends 4 CPU cores and 8 GB of RAM. For several VMs at once, or a local AI model, step up to a Core or Ryzen chip.

For an example, see Intel's N150 specification page. As of October 2026, it lists 4 cores and 4 threads, a launch in Q1 2025, a maximum of 16 GB of memory on one memory channel, and support for Quick Sync Video, VT-x and VT-d.

2. Memory: soldered or upgradable

Memory (RAM) is the working space for every app and VM. On a home server, it often runs out before the CPU does.

Check two things in the listing:

  • How much is fitted. Add up what your apps need. Proxmox VE asks for at least 2 GB for itself, plus memory for each guest. Immich alone recommends 8 GB. For a few apps, 16 GB gives you room to grow.
  • Is it soldered or in a slot? Soldered memory is fixed to the board. You can never add more. Memory in a slot (often called SO-DIMM) can be swapped for a bigger stick later.

Also check the chip's maximum on its spec page. If a listing claims more than Intel lists, treat that claim with care.

If you plan to try local AI, memory matters even more. The guide to running AI on your own computer explains why.

3. Storage: count the slots

Look for:

  • NVMe slots. NVMe is a fast type of SSD shaped like a stick of gum. One slot is the minimum. Two slots let you mirror the system drive or keep apps and data apart.
  • A 2.5-inch bay. Some boxes have room for one laptop-size SSD or hard drive.
  • USB ports for backup. A fast USB port lets you plug in an external drive for backups.

Avoid boxes where the only storage is eMMC (a slow flash chip soldered to the board). You cannot replace it when it wears out.

4. Network: one wired port, ideally 2.5GbE

A server should use a cable, not Wi-Fi. Wi-Fi drops and slows big file copies.

Many recent mini PCs have 2.5GbE ports. That means 2.5 gigabit Ethernet, two and a half times the speed of the common 1 gigabit port. It only helps if your router or switch also supports 2.5GbE. For most homes, one fast port is plenty.

Two ports help if you want the box to act as your firewall (a router OS such as OPNsense).

5. Power draw: why idle matters most

A home server runs all day, every day, and is idle most of that time. So idle power matters more than peak power.

Intel lists a "processor base power" of 6 W for the N150. That is the chip alone. Memory, SSD, fan and power brick all add to it, so it is not the figure on your bill.

The honest way to know is to measure. Run the box for a day on a plug-in power meter, then multiply by your electricity rate.

6. Noise and heat

If the box sits in your living room, noise matters. Fanless models are silent but get warm, so give them space. Others have a small fan that may whine under load. Read owner reviews for noise, not seller claims, and keep the box out of closed cabinets.

7. Intel Quick Sync for media

Transcoding means converting a video on the fly, so a phone or an old TV can play it. It is heavy work for a CPU alone.

Intel Quick Sync Video is a media engine built into the graphics part of many Intel chips. It does that video work in dedicated hardware, which leaves the CPU free for other jobs. If the media server runs in a VM or container, you must pass the graphics device through to it. The Jellyfin hardware acceleration docs list Intel Quick Sync as a supported method, next to NVIDIA, AMD and others. AMD Ryzen chips have their own media engine, so check the media server's docs for your exact chip.

If your family watches on a mix of phones, tablets and older TVs, Quick Sync is worth having.

8. Virtualization support

If you want to run Proxmox VE or any VMs, the CPU needs hardware virtualization. Intel calls it VT-x. AMD calls it AMD-V. The Proxmox VE requirements page recommends a CPU with the Intel VT or AMD-V flag, and VT-d or AMD-Vi for passing a device straight through to a VM.

Most current mini PC chips support this, but some boxes ship with it switched off. You turn it on in the BIOS (the setup screen before the OS loads). The guide on how to enable virtualization in the BIOS walks you through it. To see what you can then run, read what Proxmox VE is and why home labs run it.

While you are in the BIOS, look for a setting that powers the box back on after a power outage. The name varies by maker. Without it, your server stays off after an outage until you press the button.

The alternative: a used business mini PC

Companies replace their office PCs every few years. Many are small "tiny" or "micro" desktops that end up on the used market.

They were built to run all day in an office. Memory is usually in slots. Many have an NVMe slot and a 2.5-inch bay. Many use an older Intel Core chip with Quick Sync. Check the exact model on Intel's spec page.

They also have limits. They are older, so newer features may be missing. They usually have one 1 gigabit port. Some need a proprietary power brick, so make sure the seller includes it. And a used box has no warranty in most cases.

For a first home server, it is a sensible way to learn what you need before you buy new.

Mini PC home server checklist

Use this table when you read any listing. Open the CPU's official spec page for every "check" item.

ItemWhat to look forRed flag
CPU classN-series for light services. Core or Ryzen for many VMs.CPU model not named in the listing
VirtualizationVT-x or AMD-V listed on the official spec pageNo BIOS option to enable it
Memory16 GB, or room to reach itSoldered 8 GB if you want VMs
Memory maximumMatches the chip's official maximumSeller claims more than the spec page
StorageAt least one NVMe slot. Two is better.eMMC only
Extra bayA 2.5-inch bay or a fast USB portNo way to add a drive
NetworkOne wired port, 2.5GbE if your network supports itWi-Fi only
MediaIntel Quick Sync, or a GPU your media server supportsNo hardware video support, if you need to transcode
PowerLow idle draw, measured by youNo reviews with measured figures
NoiseFanless, or a quiet fan in owner reviewsReviews that mention fan whine
Power outageBIOS option to power on after an outageNo such option

What you need besides the box

  • Hardware. The mini PC, an SSD if none is fitted, a network cable and, ideally, a small UPS (a battery that keeps the box running through short power outages).
  • Skill. You need to install an OS from a USB stick and use a web interface. You do not need to code.
  • Time. An evening to set up, then an hour a month for updates.
  • Cost. I do not quote prices, because they change by the week. The CPU class, the memory and storage fitted, and new versus used drive the price.

The risky bit: what can go wrong

  • Buying a box you cannot upgrade. Soldered memory and eMMC storage lock you in.
  • Trusting the listing over the spec page. Listings round up and copy text. The maker's page is the source of truth.
  • Putting all your data on one SSD. One drive is one point of failure. A dead SSD takes the photos with it, unless you have a backup.
  • Asking a small chip to do a big job. An N100 box running ten VMs will feel slow. Match the box to the jobs.

The recovery step: plan for the box to fail

Every box fails one day. Plan for it from day one.

  1. Keep your data on a backup that is not inside the mini PC. An external drive plus one off-site copy is a good start.
  2. Back up your VMs and containers on a schedule. Proxmox VE has backup jobs built in.
  3. Write down the IP address, the services and where each one keeps its data.
  4. Test a restore. Bring one VM or one folder back from the backup, and check it opens.

With these steps, a dead mini PC is a bad afternoon, not a lost photo library.

FAQ

Is an N100 mini PC good enough for a home server?

For a few light services, often yes. It is a common pick for an ad blocker, a media server with Quick Sync and a password manager. Check each app's requirements first. For many VMs at once, choose a Core or Ryzen chip with more memory.

What is the difference between the N100 and the N150?

The N150 is the newer chip. Intel lists it with a Q1 2025 launch. Both are 4-core chips in the same family, per Intel's N-series chart. Compare the two spec pages on Intel's site for the exact differences.

Can a mini PC run Proxmox?

Yes, if the CPU supports VT-x or AMD-V and you turn it on in the BIOS. Proxmox recommends a CPU with that flag. Memory is usually the limit, so get as much as the box allows.

Do I need 2.5GbE on a home server?

Not always. It helps with big file copies, but only if your router or switch also supports 2.5GbE. For streaming and small apps, a 1 gigabit port is fine.

Is a used office PC better than a new mini PC?

It can be. Used business mini PCs often have slotted memory and room for drives. They are older and rarely have a warranty.

Related guides

How to Run AI on Your Own Computer: A Plain-English Guide to Local LLMs

Illustration of an AI model running privately and offline on a laptop

You use a chatbot every day, and one evening you wonder where your questions go. A local LLM is the answer to that worry. It is an AI chat model that runs on your own computer, so your words stay on your machine. Free tools such as Ollama and LM Studio make the setup short on a recent laptop. The trade-off is simple: local models are smaller than the big cloud ones, so they know less and make more mistakes.

Who this is for: you are curious about AI, you have a fairly recent laptop, desktop or mini PC, and you want private, offline AI without a subscription. Who it is not for: anyone who needs the best possible answers for work today, or whose computer has little memory (the memory section below explains how to check). A cloud service will serve you better.

What a local LLM is, in plain words

LLM stands for large language model. It is the type of AI behind chat tools: a program that has read a huge amount of text and predicts good next words. The "model" is one big file, often several gigabytes. A "runner" is the app that loads that file and lets you chat with it.

When you use a cloud chatbot, the model runs in a data center. Your question travels over the internet, and the answer comes back. With a local LLM, the model file sits on your own disk. Your computer does the work. Once you download the model, you can unplug the internet and keep chatting.

You may also see the words "open model" or "open weights". That means the company that made the model lets you download it. Each model has its own license (the rules for use), so check the license on the model's page before you use it for a business.

Why run a local LLM?

  • Privacy. Your notes, letters and family questions stay on your computer. The Ollama FAQ says, as of October 2026, that Ollama does not see your prompts or data when you run models locally.
  • It works offline. On a plane, in a cabin, or during an internet outage, it still answers.
  • No monthly fee for local use. Ollama and LM Studio are free to run with local models. Your cost is the hardware you already own and the power it uses.

What a local LLM cannot do (yet)

Be honest with yourself here. A model that fits on a laptop is much smaller than the models behind the big cloud services. In practice that means:

  • More mistakes. Small models get facts wrong more often. The Home Assistant docs put it plainly: smaller models are more likely to make mistakes than larger ones. Check anything that matters.
  • Old knowledge. A model only knows what it learned up to its training date. Most local setups do not search the web for you.
  • Short memory by default. The "context window" is how much text the model can keep in mind at once. As of October 2026, Ollama's docs say it uses a 4k context (about a few pages of text) on machines with under 24 GiB of graphics memory. You can raise it, but that uses more memory.
  • Speed depends on your machine. A model that answers quickly on one computer can crawl on another. Only a test on your own hardware tells you.

For drafting, summaries of your own notes and brainstorming, a local model is often good enough. For legal, medical or money decisions, no model is a source of truth.

Ollama vs LM Studio: which tool to start with

Both tools download models and run them on your machine. They feel different. These facts are as of October 2026. Check the vendor pages before you rely on them.

FactorOllamaLM Studio
What it feels likeA small background service you drive from a terminal (a text window for commands), plus a desktop app on Mac and WindowsA desktop app with a chat window, a model browser and settings
LicenseOpen source, MIT licenseFree, but closed source; terms allow personal and internal business use
PlatformsmacOS 14 or newer (Intel Macs run on the processor only), Windows 10 22H2 or newer, LinuxmacOS 14 or newer on Apple Silicon, Windows (x64 and Arm), Linux (Ubuntu 20.04 or newer)
Where models come fromThe Ollama model libraryHugging Face, a large public model hub, searched inside the app
Chat with your documentsDrag a text file or PDF into the Mac or Windows appBuilt in, offline
Lets other apps use the modelYes, a local API (a door other programs use) on port 11434Yes, a local server that copies the OpenAI API style
Optional cloud modelsYes, after you sign inYes, on paid plans
Best forHome servers, scripts, and feeding other self-hosted appsTrying models on a laptop with no terminal at all

Sources: the Ollama GitHub repo and docs, and LM Studio's system requirements and app terms.

My suggestion: if you want to click, not type, start with LM Studio. If you plan to run AI on a home server and connect other apps to it later, start with Ollama. This is a judgment on ease of use, not a tested ranking.

What you need: memory first

The single biggest limit is memory. The whole model must fit in memory while it runs, plus room for the conversation. If it does not fit, it runs very slowly or not at all.

There are two kinds of memory to know:

  • RAM is your computer's main working memory.
  • VRAM is the separate memory on a graphics card (GPU). Models run much faster when they fit in VRAM. Apple Silicon Macs share one pool of memory between the processor and graphics, which helps.

Here is how to size it without guesswork. Check the official guidance for the model before you download it. As of October 2026, the Ollama quickstart says the gemma4:e2b download is about 7.2 GB, and it recommends 8 GB of available graphics memory, or unified memory on a Mac. With less, Ollama can use system RAM, but answers come more slowly. Bigger models need more. Each model's page on the Ollama library lists the size of every version.

LM Studio's own system requirements recommend 16 GB of RAM or more. On Windows and Linux x64, the processor must support AVX2, a set of math instructions. Check your processor's spec page if you are not sure. A graphics card is optional. Ollama supports NVIDIA, AMD and Apple graphics, and runs on the processor alone if there is no supported card. Expect slower answers that way.

You also need disk space. Ollama's docs warn that models can take tens to hundreds of gigabytes once you collect a few. Start with one small model.

Your first local LLM with Ollama, step by step

These steps follow the Ollama quickstart as of October 2026. Commands change, so check the docs if one fails.

  1. Install Ollama. On Mac or Windows, download the app from ollama.com and run it. No administrator rights are needed on Windows. On Linux, open a terminal and run the official install script. (It is good practice to open the script URL in a browser and read it first.)
curl -fsSL https://ollama.com/install.sh | sh
  1. Download a small model. This is the model the Ollama quickstart uses. The download is about 7.2 GB, so it takes a while on a slow connection.
ollama pull gemma4:e2b
  1. Start a chat. Type a question at the >>> prompt and press Enter.
ollama run gemma4:e2b
  1. Leave the chat. Type /bye or press Ctrl+D.
  2. See what is loaded. This shows running models and whether they use the processor or the graphics card.
ollama ps
ollama ls

Want a chat window instead of a terminal? As of October 2026, Ollama's Mac and Windows apps include one.

The same thing in LM Studio

Install the app for your system. Open the model search, pick a small model, and download it. Compare its file size with your free memory, as above. Load it and start chatting. That is the whole job, and you never open a terminal.

The risky bit: what to keep away from it

A local LLM is private, but it is not automatically safe. Watch for these.

  • Do not open it to the internet. As of October 2026, Ollama listens only on your own computer (address 127.0.0.1, port 11434) by default. Its API has no password, so anyone who can reach that port can use it. The OLLAMA_HOST setting can change that. Never forward that port on your router. If you want to reach it from your phone, use a private VPN instead.
  • Know when you are using the cloud. Both tools now offer optional cloud models. In Ollama, model names that contain "cloud" run on Ollama's servers, not yours. If privacy is your reason, pick a local model.
  • Do not hand it the keys. A chat model that only talks is low risk. An AI agent that can run commands, read files or send email is a different thing. Keep agents in a virtual machine, a computer-in-a-computer you can reset. The Hermes vs OpenClaw comparison explains how.
  • Do not paste secrets. Passwords, bank details and other people's private data do not belong in any chat, local or not. Chat history can be saved on disk.
  • Get models from known places. Use the official Ollama library or well-known publishers on Hugging Face.
  • Check its answers. It will sound confident when it is wrong.

Connect it to AI agents (when you are ready)

A local model can also be the "brain" for a self-hosted agent. Both Hermes Agent and OpenClaw can use Ollama. Hermes needs a much bigger context than Ollama's small default, so you must raise that setting. Follow the step-by-step guides to install Hermes Agent in an Ubuntu VM or run OpenClaw with Docker Compose in a VM. Expect an agent on a local model to need more memory than chat alone.

The recovery step: clean up or remove it

Local AI is easy to undo. Models are just files.

  • Free memory now: ollama stop gemma4:e2b. Ollama also unloads an idle model after five minutes by default.
  • Free disk space: ollama rm gemma4:e2b deletes that model.
  • Find the files: models live in ~/.ollama/models on a Mac, /usr/share/ollama/.ollama/models on Linux, and C:\Users\%username%\.ollama\models on Windows.
  • Remove Ollama on Linux: the Linux docs list the full steps. They start with these two commands.
sudo systemctl stop ollama
sudo systemctl disable ollama

On Mac and Windows, uninstall the app the normal way, then delete the models folder. In LM Studio, delete models from its model list. If you test on a spare machine or a virtual machine, a snapshot taken before you start is the simplest undo of all. The guide to starting a home lab shows cheap ways to get one.

Local AI is one part of a bigger move to run apps at home. The overview of self-hosting in 2026 shows where it fits.

FAQ

Is running an LLM locally free?

The software can be. As of October 2026, Ollama is open source and LM Studio is free for personal and internal business use with local models. You pay for hardware and power. Both also offer optional cloud models, which you do not need.

Can I run a local LLM without a graphics card?

Yes. Ollama runs on the processor alone when there is no supported graphics card. Answers come more slowly. Start with the smallest version of a model and see if the speed suits you.

How much RAM do I need for a local LLM?

It depends on the model. Check its official page before you download. As of October 2026, the Ollama quickstart recommends 8 GB of available graphics memory, or unified memory on a Mac, for its starter model. LM Studio recommends 16 GB of RAM or more.

Is a local LLM as good as ChatGPT?

Usually not. Models small enough for home hardware make more mistakes and know less than the large cloud models. They are good enough for many everyday writing and summary tasks, and they keep your data at home.

Is Ollama or LM Studio better?

Neither wins for everyone. LM Studio suits people who want a point-and-click app. Ollama suits people who want a background service that other apps and home servers can use. You can install both.

Is a local LLM private?

When you run a local model, your prompts stay on your computer. Privacy breaks if you choose a cloud model, expose the port to the internet, or connect it to tools that send data out. Check each of those.

Related guides

Best NAS for Home in 2026: How to Choose One (and Never Lose Your Data)

Illustration of a home NAS with a separate backup drive and an off-site cloud copy protecting its data

My own 6 TB NAS failed quietly one day, and it took everything with it. It was the only copy. So here is the short answer first. The best NAS for home fits your main job (photos, files, media or apps). It uses NAS-rated drives, and it sits inside a real backup plan. A NAS is a great home for your data. It is not a safe on its own.

Who this is for: you have family photos on old phones and documents on one laptop. Or your media library has outgrown a USB drive. You want one place for all of it, at home. Who it is not for: you have a small amount of data and no wish to maintain hardware. A cloud service is simpler for you, and that is a fine choice.

What is a NAS, in plain words?

NAS stands for network attached storage. It is a small computer with several drive bays that sits on your home network. Every device in the house can save files to it, like a shared drive that never leaves home. Most modern NAS boxes also run apps, such as a photo library, a media server or a backup target for your laptops.

Here is how a NAS compares with the two other common options.

OptionBest forEffortMain risk
Cloud storage serviceSmall to medium data, no hardwareVery lowMonthly cost grows with data, and you trust a third party
NAS (appliance)Lots of files and photos, shared by a householdLow to mediumOne box holds everything unless you back it up
Mini PC home serverApps, containers and virtual machinesMedium to highUsually only one or two drives, so less room for storage

Many people use two: a NAS for storage plus a cloud backup of the important folders. The home server setup guide compares hardware classes and operating systems in more depth, so this post stays on the NAS.

The best NAS for home in 2026: brands and who they suit

This is not a tested ranking, because I have not benchmarked these boxes. Here is what each brand is known for, as of October 2026. Specs below are from each vendor's product page. Check every detail there before you buy.

The established brands

  • Synology. Known for its polished DSM software and mature photo, backup and file sync apps. Its Plus series is the usual home choice. For example, the DS925+ is a four-bay model with an AMD Ryzen V1500B processor. It has two 2.5GbE ports (2.5 times faster than standard gigabit wired networking). Its processor has no built-in video engine, so it is a weak pick for media transcoding.
  • QNAP. Known for lots of hardware options and features, such as faster networking and expansion cards. QTS has many settings, which some beginners find harder.
  • Asustor. Known for good value hardware and its ADM software. A solid middle ground for people who want an appliance.
  • TerraMaster. Known for budget-friendly hardware. TOS is a simpler system, so check it has the apps you need.

The challengers

  • UGREEN NASync. A newer NAS line from a well-known accessory brand. The DXP4800 Plus has four bays, an Intel Pentium Gold 8505 processor, and one 10GbE and one 2.5GbE port. Its UGOS Pro software includes an AI photo album that sorts by faces, places and things.
  • Minisforum. Known for mini PCs, it now makes NAS boxes too. The N5 Pro has five drive bays and an AMD Ryzen AI 9 HX PRO 370. It has 10GbE plus 5GbE networking and supports ECC memory (memory that fixes small errors). It ships with MinisCloud OS and also lists Windows and Linux support.
  • ZimaCube. From IceWhale, it runs ZimaOS, an app-first system. The ZimaCube 2 has six SATA bays plus a separate bay for NVMe drives.

DIY and open options

  • TrueNAS Community Edition. Free, open source and built on ZFS. As of October 2026, the TrueNAS software status page recommends 25.10 "Goldeye" for general users.
  • Unraid. A paid license with tiers. Its main strength is mixing drives of different sizes. Some tiers include one year of updates, then an optional renewal.

The self-hosting in 2026 guide covers these platforms as part of the bigger trend of bringing apps home.

Trends worth knowing

  • AI photo search. Several brands now tag faces and objects on the box, without a cloud.
  • Docker apps. Most brands now run containers (small, packaged apps). The guide to containers vs virtual machines explains the idea.

Who it suits: a shortlist by use case

Criteria: ease of setup, app ecosystem, hardware headroom and freedom to choose drives.

Your main jobLook atWhy
Family photos and files, minimal fussSynology Plus, UGREEN NASyncMature or simple apps, guided setup
Media server with video transcodingModels with Intel Quick Sync or a strong built-in GPUHardware video conversion keeps playback smooth
Apps, containers and some AIMinisforum N5 Pro, ZimaCube 2, higher-end QNAP modelsMore CPU and memory headroom (check the exact model)
Mixed old drives, tinkeringUnraid on your own hardwareMix drive sizes and brands, large app catalog
Data safety first, comfortable learningTrueNAS Community EditionZFS checksums and snapshots, free (Synology's Btrfs also checksums)

What to look for in a home NAS

Bays

Buy more bays than you need today. Two bays give you a mirror (two copies of each file on two drives). Four bays give you room to grow and better protection options. Adding drives later is easier than moving to a new box.

Drives: NAS-rated, CMR, not the cheapest

Use drives made for NAS use. They are built to run all day and to work in a group. Check the spec sheet for CMR (conventional magnetic recording). Avoid SMR (shingled magnetic recording) drives in a RAID group. SMR drives write in overlapping bands, and they can be very slow during a rebuild. A slow rebuild leaves your data exposed for longer. Major drive makers now list the recording type. WD's Red Plus and Red Pro lines are CMR. Check before you buy.

RAID, SHR and ZFS in plain words

RAID combines several drives so the box survives a failed drive. RAID 1 mirrors two drives. RAID 5 survives one failed drive. RAID 6 survives two. SHR is Synology's flexible version that handles mixed drive sizes. ZFS (used by TrueNAS) and Btrfs (used by Synology and others) add checksums. A checksum is a fingerprint for each block of data. The system can spot silent corruption and repair it from a good copy, if the pool has one.

None of these is a backup. They keep the box running when one drive dies. They do not save you from a deleted folder, ransomware, a fire or a failed enclosure.

RAM and CPU

File sharing alone needs little power. Photo AI, Docker apps and virtual machines need more memory and a stronger CPU. Look for upgradable RAM slots. For a media server, an Intel chip with Quick Sync (Intel's built-in video engine) helps with transcoding. Transcoding means converting video on the fly for a phone or TV. Not every NAS chip has a video engine, so check the spec sheet.

Networking

Gigabit is enough for photos and documents. 2.5GbE helps with big video files. 10GbE helps only if your whole network supports it.

Power and noise

A NAS runs all day, so power use adds up. Hard drives also click and hum. Put the box in a ventilated spot away from the bedroom. Most vendors publish power figures on the spec sheet.

Apps and update support

The software matters as much as the hardware. Check the photo, backup and sync apps you will use every day. Also check how long the vendor supports a model with security updates. A NAS that stops getting updates becomes a risk on your network.

Drive compatibility lock-in

Some vendors limit which drives you can use. Synology changed its policy twice in 2025. As of October 2026, its drive compatibility FAQ for 2025 and later models sets the rules. DS Plus models on DSM 7.3 accept unlisted third-party hard drives and 2.5-inch SATA SSDs for new storage pools. Drives on its incompatibility list are refused. M.2 NVMe drives still need to be on the compatibility list to create a pool or cache. The rules apply only to models launched from 2025. Read the current page before you buy, because policies can change again.

What you need: hardware, skill, time and cost

  • Hardware: the NAS, NAS-rated drives, a wired network connection and, ideally, a small UPS (a battery that keeps the box on through a power cut).
  • Skill: an appliance needs about the skill of setting up a new router. DIY options need more comfort with menus and terms.
  • Time: an afternoon for setup, then a few minutes a month for updates and checks.
  • Cost: drives often cost more than the box. Cost rises with bay count, CPU class, 10GbE and RAM. Do not forget a second place for backups.

RAID is not a backup: the safeguards that matter

This is the part I wish I had read first. A NAS with RAID is still one box in one room. Here is how to protect what is on it.

  1. Follow 3-2-1. The US Cybersecurity and Infrastructure Security Agency (CISA) advises keeping 3 copies of important files, on 2 different types of storage, with 1 copy off-site. See its guide to backing up data. For a home, that can be the NAS, a USB drive you rotate, and a cloud backup.
  2. Turn on snapshots. A snapshot is a quick undo point for a folder. It helps when you delete a file or when ransomware encrypts a share. Snapshots live on the same drives, so they are not a backup.
  3. Back up off-site. Most NAS brands have a built-in app that copies data to a cloud storage service or to a second NAS at a relative's house.
  4. Watch drive health. Turn on SMART monitoring (the drive's own health report) and email alerts. Schedule regular scrubs, which read every block and check it for errors.
  5. Use a UPS. A sudden power cut during a write can damage data. Connect the UPS by USB so the NAS shuts down cleanly.
  6. Replace drives early. When a drive shows errors, replace it. Keep a spare on the shelf if you can.
  7. Test a restore. Pick a folder, restore it from the backup to another place, and open the files. A backup you have never restored is a hope, not a plan.

NAS security: keep it off the open internet

NAS boxes are a favorite target because they hold everything. The DeadBolt ransomware began attacking certain QNAP NAS devices on 25 January 2022. It encrypted files and replaced the login page with a ransom note, per QNAP's security advisory QSA-22-02. QNAP's main advice was to update to the fixed version immediately. From 21 February 2022, DeadBolt also hit Asustor devices. Its advisory AS-2022-002 advises a strong password, changing the default web ports, disabling unused services, regular backups and prompt updates. These are examples, not outliers. Synology has also warned of brute-force attacks on its NAS boxes, so every brand needs these steps.

  • Never expose the admin page to the internet. Do not forward its ports on your router.
  • Turn on automatic security updates for the system and its apps.
  • Disable the default admin account and create your own admin with a long, unique password.
  • Turn on two-factor sign-in (a code from your phone as well as the password).
  • Disable UPnP (a feature that lets devices open router ports by themselves) on the NAS and the router. Also turn off vendor relay services you do not use.
  • Reach the NAS from outside through a VPN. A private mesh network such as Tailscale lets your phone reach the NAS without opening any ports. The home server setup guide covers safe remote access.

Common mistakes

  • One copy only. This is what got me. If the NAS is the only copy, one failure ends everything.
  • The cheapest drives. Desktop drives are not built for all-day group work.
  • SMR drives in RAID. Slow rebuilds increase the chance of a second failure.
  • All drives from one batch. Some builders spread purchases so drives do not share one batch.
  • Ignoring warnings. A "degraded" message means act today, not next month.
  • No UPS. A power cut during a write can damage data.
  • Vendor lock-in. Check drive and app rules before you commit years of data.
  • End-of-support models. An old NAS with no updates should not touch the internet at all.

If your NAS already died: recover data safely

First, breathe. My loss was total because that NAS held my only copy. Your drives may still be fine. Sometimes the fault is the power supply or the board, not the drives. What you do in the next hour matters most.

  1. Stop. Do not click "initialize", "repair", "rebuild" or "create new volume". Any of these can write over your data. If you connect the drives to a Windows PC and it offers to initialize or format a disk, click Cancel.
  2. Power it down and listen. If a drive clicks, grinds or does not spin up, stop here and contact a data recovery service. Each extra power-on can make a failing drive worse.
  3. Label the drives. Before you pull any drive, mark its bay number on a sticker. Order can matter for reassembly.
  4. Clone first if it is your only copy. For irreplaceable data, a professional service is safest. At the least, have each drive cloned (copied sector by sector) before anyone reads it.
  5. Read, never write. Most home NAS boxes use standard Linux storage, so another Linux machine can often read the drives. Synology publishes a guide on how to use a PC to recover data when a Synology NAS malfunctions. You run Ubuntu 18.04 on a PC with all the drives connected, force the RAID to assemble, then mount the volume read-only. Only the mount is read-only. The forced assembly can write RAID data to the drives. As of October 2026, the guide covers DSM 6.2 and later with Btrfs or ext4 volumes, not volumes with a read-write SSD cache. Encrypted volumes also need their recovery key. Synology warns that the drives may not work in a new NAS afterward.
  6. Copy off, then stop again. Copy your files to a different drive. Do not try to fix the original volume.
  7. Know when to call a pro. If the steps fail, stop. Synology's guide says the same: more repair attempts may cause more damage, and the last option is a data recovery company. Recovery is never guaranteed.

Not comfortable with a Linux terminal? Skip step 5 and go straight to a professional.

FAQ

What is the best NAS for home use?

There is no single best one. Pick by your main job. An easy appliance such as Synology or UGREEN suits photos and files. A stronger box suits apps and media, and TrueNAS or Unraid suits tinkerers. Then add a backup.

Is RAID a backup?

No. RAID keeps the NAS running when one drive fails. It does not protect you from deletion, ransomware, theft, fire or a failed enclosure. You still need copies on other storage, with one off-site.

Can I use any hard drive in a Synology NAS?

As of October 2026, Synology's FAQ covers models launched from 2025. On DSM 7.3, DS Plus models accept unlisted hard drives and 2.5-inch SATA SSDs for new pools. M.2 NVMe pools and caches still need listed drives. Models launched before 2025 are not covered by these rules. Other 2025 series are stricter. Check the current FAQ for your model.

How do I recover data from a NAS that will not turn on?

Do not initialize, format or rebuild anything. If a drive clicks or the data is irreplaceable, go to a professional recovery service. Otherwise, label the drives and follow the vendor's PC recovery guide, if one exists.

Is it safe to access my NAS from outside the house?

Yes, through a VPN or a private mesh network such as Tailscale. Never forward the admin page port on your router.

NAS or mini PC: which should I buy?

Buy a NAS if storage is the main job and you want several drives. Buy a mini PC if apps and virtual machines matter more. Many home labs run both, with the NAS as storage and the mini PC as the app server.

Related guides

Self-Hosting in 2026: Why People Are Bringing Their Apps Home (and What It Really Takes)

Illustration of apps moving from the cloud back to a small home server on a shelf

Your phone says your photo storage is full again. The streaming bill went up. A new AI tool wants a monthly fee and all your documents. Self hosting is the other option: you run the app on a computer you own, in your home, and your data stays there. In 2026 it is easier than it has ever been, but it is still a small job you take on, not a product you buy. This guide shows why people do it, which tools to know, and what it really takes.

Who this is for: you are curious, you have an old laptop, a mini PC or a NAS (a network storage box), and you want to run one or two useful apps at home safely. Who it is not for: you want zero upkeep, or nobody in your house will ever check a backup. In that case, a good paid cloud service is the better choice.

What self hosting means, in plain words

Most apps you use run on a company's computers. You rent access, and you follow their rules on price, storage and privacy.

To self host is to run your own copy of an app on your own hardware. A "server" here is just a computer that stays on and serves apps to your other devices. It can be a ten-year-old desktop under the stairs. Your phone, TV and laptop then talk to that box.

Why self hosting is growing now

No single reason is new. Together, they explain why normal people, not only IT staff, now run a box at home.

  • Subscription fatigue. Many households now pay for photo storage, music, video, cloud drives and AI tools, all on monthly plans. Self hosting replaces some of them with a one-time hardware cost and your time.
  • Privacy and control. Family photos and documents are personal. When they sit on your box, no company can scan them, change the terms or close your account.
  • Local AI. You can now run an AI chat model on your own computer. Your questions and files do not leave the house.
  • Small, quiet hardware. Modern mini PCs are small enough to sit on a shelf and quiet enough to forget about. Used office PCs are another common start.
  • A push toward open source. After Broadcom bought VMware, it changed how VMware products are licensed. Some small users and home labbers moved to open source platforms such as Proxmox as a result.

The self hosting map: who does what

Sort the projects by the job they do. This is a map of well-known names, not a tested ranking. Check each project's own site before you commit.

Platforms: the base layer

The platform is the software that runs on the box itself. Your apps sit on top of it.

PlatformWhat it isCost model (as of October 2026)Good fit if
Proxmox VEA hypervisor (software that runs many virtual computers on one machine) with a web page to manage itOpen source (AGPLv3), free to use. Optional paid support subscription.You want to learn, and run many separate apps safely apart
UnraidA storage and apps system that mixes drives of different sizesPaid license, bought once, with a set period of updates or lifetime updates by tier. Free trial.You have a pile of mixed hard drives and want simple app installs
TrueNAS Community EditionA storage system built on ZFS (a file system that checks your data for damage)Free, open sourceSafe storage comes first, apps second
Synology DSMThe operating system that comes on every Synology NASComes with the Synology hardwareYou want a ready-made box and a polished interface
umbrelOS, ZimaOSAppliance-style home server systems with a one-click app storeumbrelOS: free for personal use under a noncommercial license. ZimaOS: a free edition with limits on disks and users, plus a paid Plus license.You want the easiest start and accept fewer controls

Check the current terms on each vendor page. For example, Unraid lists its tiers and trial on the Unraid license page. CasaOS, from the maker of ZimaOS, is still open source, but its team said in January 2026 that it is in a maintenance phase, with new work going to ZimaOS. For more on Proxmox itself, read what Proxmox VE is and why home labs run it.

What it means for you: there is no single right platform. Pick the one that matches your first goal. Storage first points to TrueNAS, Unraid or Synology. Learning and many apps points to Proxmox. Fastest start points to an appliance system.

Apps: the reason you do this

All app details below are as of October 2026.

  • Immich replaces a cloud photo service. It backs up phones and shows a timeline, faces and maps. Immich announced its first stable release, v2.0.0, on 1 October 2025 (see the Immich v2.0.0 release post).
  • Jellyfin is a media server for your own movies, shows and music. The project says the server and official apps are free, with no tracking.
  • Home Assistant runs your smart home locally. Its own tagline puts "local control and privacy first."
  • Nextcloud gives you files, calendar and contacts in one place, like a private cloud drive.
  • Vaultwarden is an unofficial server that works with Bitwarden password apps. It is not made by Bitwarden, Inc., and its own docs tell you to back up its data.

What it means for you: start with one app that solves a real annoyance. Family photos are a common first pick, because the value is clear. They are also the data you can least afford to lose, so backups come first.

Remote access: reaching home from outside

Your apps are most useful when you can reach them from your phone on the train. The old way was "port forwarding", which opens a door on your router to the whole internet. That invites break-ins.

A safer pattern is a private network between your own devices. Tailscale is the well-known option here. It is built on WireGuard (a modern, fast VPN protocol) and works across home routers without port forwarding. As of October 2026, the Tailscale pricing page lists a free Personal plan with up to 6 users and unlimited devices. Verify that before you plan around it. Plain WireGuard is the do-it-yourself version, with more setup.

What it means for you: do not open ports on your router for your first apps. Use a private network instead.

Local AI: the new arrival

Ollama is a free tool (MIT license) that downloads and runs AI language models on macOS, Windows or Linux, per its Ollama GitHub page. Open WebUI adds a chat page in your browser, much like the big AI chat sites. Note that Open WebUI changed its license in 2025 to add a branding clause. It stays free to use, but read its license page if you plan to rebrand it.

The limit is memory. Bigger models need more RAM or a graphics card with more video memory. Check the model page for its size before you download it. Some people go further and run AI agents that can use tools. If that interests you, run them in a separate virtual machine. The guide to Hermes vs OpenClaw on virtual machines shows why.

What it means for you: local AI is real and private, but a small box runs small models. Expect it to be slower and less capable than the big cloud services.

Trends to watch, and what they mean for you

  • Free tiers with limits. Appliance systems now mix free and paid editions. ZimaOS caps disks and users on its free edition, and umbrelOS is free only for noncommercial use. What it means for you: read the free tier's limits before you build on it.
  • Licenses that change. Unraid moved new buyers to tiers with a set period of updates in 2024. Open WebUI added a branding clause in 2025. What it means for you: check the license again before a big upgrade.
  • Apps reaching "stable". Immich's stable release is one example. What it means for you: fewer breaking updates, but you still read release notes before you update.
  • AI at home. What it means for you: you can keep sensitive questions private, but you take on the job of keeping the AI box updated and fenced off.
  • Containers everywhere. Many self-hosted apps now ship as containers (a packaged app that shares the host's operating system core). What it means for you: installs are faster. The guide on containers vs virtual machines explains when to use each.

What self hosting really takes

Hardware

Most people start with what they own: an old laptop, an old desktop or a NAS. A small, low-power mini PC is the common next step. The box runs all day, so power use matters more than raw speed. Check the CPU supports hardware virtualization if you plan to run virtual machines, and enable virtualization in the BIOS first. The home server setup guide walks through hardware classes and a first OS.

Skill

You do not need a degree. You need to be calm with a web dashboard, a few typed commands and an install guide read line by line.

Time

Plan a weekend for the first app, including backups. After that, budget a little time each month for updates and a quick check that backups still run. If nobody will do that, the box will slowly become a risk.

Cost

Prices change, so I do not quote them. What drives cost is simple: the hardware, the electricity to run it all day, extra drives for backups, and any paid license (Unraid, for example). Some apps also have optional paid tiers. Check each vendor page as of the day you buy.

The risky bit: what can go wrong

  • One disk, one copy. If your only copy of the family photos is on one drive in the box, one dead drive loses them. Immich's own README warns you to follow a 3-2-1 backup plan (3 copies, on 2 types of storage, 1 off-site) for this reason.
  • Exposed apps. An app reachable from the open internet gets attacked. Keep apps private and reach them through a private network.
  • Weak account security. Your remote access account is now the key to your house. Turn on two-factor login.
  • Silent failures. Backup jobs can fail for months without anyone noticing. Check them.
  • Too much at once. Ten apps in a weekend is ten things to update. Add one at a time.

The recovery step: before you trust it

Before you move anything important onto the box, set up a backup and prove you can restore it. A copy you have never restored is a hope, not a backup. Keep at least one copy off the box, and one copy away from the house.

Also know how to undo it. If self hosting is not for you, export your data in a standard format (photos as files, passwords as an export) and move back to a cloud service. Keep the old cloud account until the new setup has run well for a few months.

Who should not self host

  • You want to host email. It is hard to keep mail from landing in spam, and most people should not try.
  • You run a small business with customer data and legal duties. Get professional help, or start with the guide to virtualization for small business.

FAQ

What does self hosting mean?

Self hosting means you run an app on hardware you control, usually at home, instead of using a company's cloud service. You get control and privacy. You also take on updates, security and backups.

Is self hosting cheaper than subscriptions?

Sometimes, over a few years, but not always. You pay for hardware, electricity, backup drives and your time. Add up your current subscriptions, then compare with those costs before you decide.

Do I need a powerful computer to self host?

No. Many first apps run well on an old laptop, an old desktop or a small mini PC. Local AI is the exception, because larger models need much more memory.

Is self hosting safe?

It can be, if you keep apps off the open internet, use a private network, turn on two-factor login and keep tested backups. Most problems come from exposed apps and missing backups.

What should I self host first?

Pick one app that fixes a real annoyance, such as a photo library or a network ad blocker. Set up its backup on day one. Add the next app only when the first runs without fuss.

Can I self host AI?

Yes. Tools such as Ollama run language models on your own computer, and Open WebUI gives them a chat page. Expect smaller and slower models than the big cloud services on modest hardware.

Related guides