Folia vs Paper: Which Should You Pick in 2026?

If you run a Minecraft server with more than a few dozen concurrent players, you've probably seen Folia come up - usually framed as "Paper, but actually multithreaded." That framing is half-right and half-misleading, and the gap between those halves is exactly where the painful surprises live.

This post is the honest version. What Folia actually does, what it gives up, who it's for in 2026, and the specific question that decides everything: does your plugin stack support it?

The 30-second version

If you're already running Paper happily, Folia is almost certainly not the right move. If you're routinely hitting TPS floors with 500+ players spread across a large overworld, it's worth a serious look.

What Folia actually does differently

Vanilla and Paper both run the world simulation on a single thread - the "main thread." Every entity tick, every block update, every chunk load happens sequentially. Modern CPUs are multi-core, but Paper can only use one core for the actual game loop. (It uses other cores for things like async chunk I/O, but the game logic itself is single-threaded.)

Folia changes this by splitting the world into regions - groups of nearby chunks - and ticking each region on its own thread. Regions that aren't interacting can run truly in parallel. Players standing 2,000 blocks apart don't share a thread, so a redstone contraption in one region can't lag the other.

The practical implication: on a server with a multi-core CPU and players spread across the world, you can roughly multiply your tick-rate headroom by the number of CPU cores. A server that maxed out at 200 concurrent players on Paper might handle 1,000+ on Folia if everything else cooperates.

The "if everything else cooperates" problem

Here's the part people gloss over.

The Bukkit/Paper API was designed assuming a single main thread. Plugins call Bukkit.getOnlinePlayers(), walk the list, schedule tasks - all of it written assuming there's one thread, one world state, one ordering. Folia breaks that assumption fundamentally.

In Folia, you can't safely:

• Iterate Bukkit.getOnlinePlayers() from anywhere (players might be in different regions, each on its own thread)
• Run a BukkitRunnable with runTaskLater - the scheduler API was replaced with region-aware variants
• Access an entity from outside its owning region's thread without explicit thread-jumping
• Hold a static cache of world state and assume it's consistent

This means most existing plugins simply do not work on Folia. They don't crash on startup - they crash on first interaction, or worse, corrupt state silently.

The 2026 plugin compatibility landscape

The Folia ecosystem has grown since launch in 2023, but it's still niche compared to Paper. Rough state of things as of mid-2026:

Generally Folia-compatible:

• Modern PaperMC core plugins (LuckPerms, Geyser-Spigot for newer builds, ViaVersion, Vault forks)
• A handful of optimization plugins (Chunky, ServerOptimizer with Folia mode)
• Newer custom plugins written by devs who target Folia from day one

Generally NOT compatible:

• Most economy plugins (EssentialsX is not Folia-ready, despite long-standing requests)
• Most claim/land plugins (GriefDefender, GriefPrevention - partial or no support)
• Most quest/skill systems (mcMMO, Heroes, etc.)
• Older PlaceholderAPI expansions
• Anything that hasn't seen an update in 12+ months

If your server runs anything in the "NOT compatible" list and you can't replace it, Folia is off the table.

When Folia is the right call

There are real use cases where Folia wins decisively:

1. Massive survival or skyblock servers

500+ concurrent players spread across a multi-thousand-block overworld, where players rarely cluster. Each region ticks independently, you get near-linear scaling with CPU cores.

2. Minigame networks where players are scattered

If your network has 200 players each on their own isolated island/arena, those plot up beautifully into Folia's region model. The threading wins are dramatic.

3. New servers being built from scratch

If you haven't accumulated technical debt yet, you can pick Folia-native plugins from day one. The cost is "available plugin list is smaller"; the benefit is "you'll never hit Paper's single-thread ceiling."

When Folia is the wrong call

1. Existing servers with a settled plugin stack

Migration cost is enormous. You'll spend weeks finding Folia-compatible replacements for every plugin, and some won't have viable alternatives. Players will notice missing features.

2. Servers under ~200 concurrent players

Paper's single thread handles this load fine on modern CPUs. The complexity tax of Folia isn't worth it.

3. Servers where players cluster

PvP arenas, spawn cities, event servers - if everyone's in the same chunk, they're all in the same region. Folia degrades back to single-thread performance and you've gained nothing.

4. Anything depending on cross-world atomic operations

If you have an economy plugin that needs to debit player A in world 1 and credit player B in world 2 in a single atomic step, that's hard in Folia. The model assumes regions are mostly independent.

Benchmarks (with caveats)

Folia's repo and various community tests show 3-8x TPS improvements over Paper under ideal conditions (players spread out, lightweight plugins, multi-core CPU). I've seen real-world deployments hit 4-6x on properly suited workloads.

The catch is that real servers rarely hit ideal conditions. Cluster spawn → single region pegged. Boss fight in arena → single region pegged. Lag machine triggered by one player → single region pegged. Folia improves the average case dramatically but the worst case still bottlenecks on one thread.

The migration path (if you decide to go)

If you've checked your plugin list and Folia is viable, here's the sane migration order:

1. Spin up a parallel test server. Same world, same plugin list, but Folia jar. Don't touch production.
2. Verify every plugin loads cleanly and produces no thread-safety warnings in the console.
3. Stress test with synthetic players (bot fleet) for 24h+ to surface threading bugs.
4. Run a private beta with 20-50 real players for a week. You will find issues that bots miss.
5. Cut over with a maintenance window. Have rollback plan ready - Paper world saves are forward-compatible from Folia world saves usually, but not always.

Budget at least 2-3 weeks for a careful migration of a medium-sized server. More if you have custom plugins that need rewriting.

The "should I switch" decision tree

• Do you currently have player counts that exceed Paper's single-thread capacity? → If no, stay on Paper.
• Are all your critical plugins explicitly Folia-compatible? → If no, stay on Paper.
• Are your players spread out across a large world or many independent areas? → If no (everyone clusters), stay on Paper - Folia won't help.
• Do you have time and budget to manage a careful migration? → If no, stay on Paper.

If you answered yes to all four, Folia is worth the effort. If you answered no to any one, Paper is the safer call - probably permanently.

What we ship for

For context: at the time of writing we build our plugins targeting Paper 1.20.4+ as the primary platform. Folia compatibility is something we add on request for clients who need it - the testing matrix is larger and the writing style is more constrained, so it costs more. If you need a Folia-native plugin built from scratch, we take custom orders with that platform pre-scoped.

Bottom line

Folia is a remarkable piece of engineering and the right answer for a specific kind of large server. For 95% of servers, Paper is still the right default in 2026 - more stable, more compatible, easier to staff, and fast enough.

If you're hitting a real performance ceiling with Paper on a multi-core server with players spread out, evaluate Folia seriously. Otherwise, stick with Paper and spend your engineering budget on optimization instead - which we covered in Optimizing Paper Server TPS.