Paper, Purpur or Fabric: picking server software in 2026
Plugins or mods, performance or fidelity, stability or features. A straight comparison of the Minecraft server platforms and who each one is actually for.
Every Minecraft server owner eventually asks the same question, usually while their tick rate is on fire: am I running the right server software? Here is a practical comparison of the realistic options, and an honest answer about when the choice matters.
The three families
Paper and its downstream forks
Paper is the default answer for most public servers. It is a fork of the vanilla server with extensive performance work and a plugin API inherited from Bukkit and Spigot. That plugin ecosystem is the real product: land claims, permissions, economy, anti-cheat and moderation tooling all exist and are maintained.
Purpur sits downstream of Paper and adds a long list of configuration toggles — gameplay tweaks, mob behaviour switches, ridable entities and a lot of small quality-of-life options. It is Paper plus knobs.
- Use Paper if: you run a public server, you need plugins, and you want the most-tested option.
- Use Purpur if: you want the same thing but with gameplay toggles you would otherwise write a plugin for.
- Trade-off: the plugin API cannot change the game the way a mod can. Adding a new block or a new dimension is not a plugin job.
Fabric
Fabric is a lightweight mod loader. Mods run on both the server and, usually, the client — which is the important part. Anything that adds genuinely new content needs the client to know about it.
- Use Fabric if: you are running a modpack, a small friends server with content mods, or anything where the world itself is different.
- Trade-off: every player installs the same mod set at the same versions. That is fine for twelve friends and painful for a public server.
NeoForge and the Forge lineage
The heavier modding platform, and where most large content modpacks live. Same client-install requirement, larger memory footprint, and the deepest content available.
- Use it if: the modpack you want to run is built for it. That is genuinely the whole decision.
The comparison that actually matters
| Need | Answer |
|---|---|
| Public server with land claims and moderation | Paper |
| Same, plus gameplay toggles | Purpur |
| New blocks, dimensions, machines | Fabric or NeoForge |
| Bedrock crossplay | Paper family with a bridge |
| Thousands of players | A proxy in front of several Paper backends |
Where performance actually comes from
This is the part most owners get wrong. Switching from Paper to Purpur will not save a server that is dying, because server software is rarely the bottleneck. In rough order of real-world impact:
- Entity count. Mob farms, item frames, dropped items. This is almost always the answer.
- View distance and simulation distance. Dropping simulation distance from 10 to 6 is often worth more than every config tweak combined.
- Single-core CPU speed. Minecraft's main thread does not parallelise. A high-clock consumer CPU beats a many-core server chip.
- Disk speed for chunk I/O. An SSD is not optional on a busy world.
- The plugin you installed and forgot about. Profile before you blame the platform.
If your server is struggling, spend an hour with a timings or spark profile before you spend a weekend migrating platforms. The profile will usually name the culprit outright.
Migrating without losing a world
- Back up first, and verify the backup restores. Not "copy the folder somewhere" — actually restore it once.
- Paper to Purpur is close to drop-in. Purpur back to Paper is fine too; the extra config is ignored.
- Paper to Fabric is not a migration, it is a rebuild. Plugins do not transfer and the world may carry data those plugins wrote.
- Never migrate and update the game version in the same step. One variable at a time.
- Announce it. A surprise three-hour outage costs more players than any performance gain wins back.
The short version
Public server, plugins, players who install nothing: Paper, possibly Purpur. Small server where the world itself should be different: Fabric. A specific big modpack: whatever that pack targets. And if the real problem is performance, the answer is nearly always fewer entities rather than different software.
Server owners comparing setups can find plenty of live examples on the Minecraft server list — most listings state their platform and version, which is a decent way to see what comparable communities settled on.