Pick the wrong server software and you'll spend the next six months fighting performance issues, plugin incompatibilities, or both. Pick the right one and your server "just works" while you focus on actually building a community.
In 2026, four options matter: Paper, Spigot, Purpur, and Folia. Vanilla and CraftBukkit exist but neither has been a serious choice for community servers in years. Here's the honest comparison.
The short version
| Software | Best for | Skip if |
|---|---|---|
| Paper | 99% of community servers up to ~50 concurrent | You need 100+ concurrent in one world |
| Spigot | Maximum plugin compatibility, very legacy setups | Any server built in the last 2 years |
| Purpur | Heavy customization, fork-of-Paper features | You don't actually need the extra config knobs |
| Folia | 100+ concurrent in a single world (multi-threaded) | You depend on plugins that aren't Folia-compatible |
Spigot — the historical default, now mostly outdated
Spigot was the standard from 2013 until about 2020. It's still maintained, still works, and still has the broadest plugin compatibility because every plugin author has always tested against it. But:
- Performance is mediocre. Spigot does minimal optimization beyond what CraftBukkit already had. Things Paper optimizes — chunk loading, entity ticking, item merging — Spigot doesn't touch.
- Anti-xray, anti-grief, and async chunk improvements aren't built in. You'll bolt them on with plugins, eating into your performance gains.
Use Spigot only if you're maintaining a 5+ year-old plugin ecosystem and migration cost outweighs the performance gain. For everyone else, Paper.
Paper — the modern default
Paper is a fork of Spigot maintained by the PaperMC team. It includes hundreds of optimizations, configurable mob behavior, async chunk loading, built-in anti-xray, and much better config defaults. It's also the base most other modern forks are built on (including Purpur and Folia).
What Paper gives you over Spigot:
- Async chunk loading (massive TPS improvement at moderate scale)
- Built-in anti-xray (no need for the Anti-Xray plugin)
- Detailed per-world tick performance reports (
/timings pasteor its successor) - Sensible defaults: mob limits, despawn ranges, view distances all configurable
- Active development with releases roughly every Minecraft update
What you give up: very little. ~99% of Spigot plugins work on Paper unchanged. The plugins that don't are usually doing something weird with NMS internals and probably need updating anyway.
This is the default recommendation for almost every server in 2026. Start here, switch only if you have a specific reason.
Purpur — Paper with extra knobs
Purpur is a fork of Paper that adds dozens of configurable game-mechanics options Paper doesn't expose. Things like:
- Configurable mob spawning rates per type
- Per-player view distance
- Phantom flame trails, ridable mobs, configurable explosion rules
- Skeleton aim accuracy, zombie speed, end crystal damage caps
Use Purpur if: you're building a heavily-modded survival, a custom kit PvP arena, or any server where the vanilla mechanics need tweaking beyond what Paper allows. The config file is huge — read it before you ship.
Skip Purpur if: you're not actually going to use the extra knobs. Running Purpur with default config is just Paper with slightly slower releases (Purpur lags Paper by a few days on major Minecraft updates).
Folia — Paper, but multi-threaded
This is the big 2024-2026 story. Folia is a Paper fork from the PaperMC team that actually parallelizes the main game loop. Vanilla Minecraft (and every fork before Folia) runs the entire world on a single CPU thread — which is why even a $200/mo server struggles past ~100 concurrent in one world.
Folia splits the world into "regions" (groups of 16x16 chunks) and ticks each region on its own thread. Your 8-core server can actually use all 8 cores.
What this means in practice:
- A single-world survival can comfortably handle 200-500 concurrent players where Paper would crawl at 80
- Anti-cheat, mob behavior, and chunk loading all parallelize cleanly
- The TPS metric stops being meaningful — Folia tracks "MSPT per region" instead
The catch: plugin compatibility. Folia's threading model breaks the standard Bukkit plugin assumption that "I'm running on the main thread." Many older plugins simply don't work. The plugin author has to specifically add Folia support.
As of 2026, the situation is much better than at Folia's launch:
- Folia-compatible: EssentialsX, LuckPerms, CoreProtect, WorldEdit, ViaVersion, NuVotifier, GriefDefender
- Not Folia-compatible (as of this writing): many older economy plugins, some legacy region/claim plugins, anything that wraps the BukkitScheduler heavily
Use Folia if: you have or expect 100+ concurrent in a single world AND your plugin stack is mostly modern. Survival servers, anarchy servers, and large factions servers are the natural fit.
Skip Folia if: you'd be happier on Paper with proxy-split backends (each <50 concurrent), or your plugin list includes anything that flatly doesn't work.
What about Pufferfish, Airplane, Leaf?
Smaller Paper forks with experimental optimizations. They sometimes deliver real performance gains but the maintenance situation is variable — some are abandoned, some are healthy. If you're not actively benchmarking and willing to migrate when a fork goes stale, stick with the big four.
Real-world picking advice
You're starting a new community SMP for ~20-50 players:
Paper. Don't overthink it.
You're running a 100+ concurrent survival/anarchy:
Folia, unless one of your critical plugins isn't compatible — then Paper with a proxy and split backends.
You're running a faction/PvP server with custom mechanics:
Purpur — the extra config will save you from writing custom plugins.
You're running a minigame network (Hypixel-style):
Paper backends behind Velocity. Each minigame backend is small (<50 concurrent), so Paper's single-thread is fine. Folia adds complexity here without solving a problem you have.
You're modernizing a 5+ year-old Spigot server:
Migrate to Paper in place (same plugins, same config files mostly). Run for a month. Then consider Purpur or Folia only if you hit a real limitation.
The migration itself
Paper from Spigot is usually a drop-in replacement:
- Stop server
- Back up the world and config
- Replace the server JAR with Paper
- Start server
- Check for plugin warnings in console; update the 0-3 plugins that complain
Paper to Folia is more disruptive — plugins have to be explicitly Folia-compatible, and the config file layout changes. Allocate a weekend to test in a staging environment before flipping production.
What we run / what most listed servers run
Of the servers actively listed on our toplist with successful Votifier integrations, the rough breakdown in mid-2026 is: 65% Paper, 15% Purpur, 12% Spigot (mostly older listings), 5% Folia, 3% other. Paper's dominance is real.
Once you've picked your software, the next step is making sure votes work: see our complete NuVotifier setup guide.