• Config depth — some things you might want to tweak are locked
• Optimization ceiling — you can squeeze more TPS with forks that focus on perf

Pick Paper if: you run a server for 5 to 30 players, care about plugin stability, and don't want to think about this decision again for 12 months.

Purpur — Paper plus hundreds more knobs

Purpur is a Paper fork that adds a huge number of config options. Every setting Paper has, Purpur has. Plus hundreds more — things like "can endermen pick up blocks," rideable mob stat curves, custom AI tweaks, and dozens of gameplay-tunable details.

Why Purpur is great:

• 100% Paper plugin compatibility (it is Paper under the hood)
• Wider config surface for gameplay tuning
• Negligible performance difference from Paper
• Active development, ships quickly after Paper releases

Where Purpur falls short:

• The extra options can feel overwhelming to new admins
• If you're not going to actually use them, you gained nothing

Pick Purpur if: you run a community server where players request gameplay tweaks ("can we make phantoms rarer", "can we speed up crop growth"), or you're running a specific game mode that benefits from fine-tuning.

Pufferfish — the performance-focused Paper fork

Pufferfish was built specifically for high-load servers. Think 40+ players, large redstone contraptions, heavy mob farms, modded Paper hybrids.

Why Pufferfish is great:

• Measurable TPS gains under heavy load (internal testing shows 8 to 15% on 40+ player scenarios with heavy entity counts)
• Async entity handling — mobs don't block the main thread the way they do on vanilla Paper
• Smaller but very active optimization community
• Near-complete Paper plugin compatibility

Where Pufferfish falls short:

• Overkill for small servers — on 5 to 10 player servers you won't notice any difference
• A few edge-case plugins that touch entity internals can behave oddly
• Smaller community means fewer guides and Stack Overflow answers

Pick Pufferfish if: you run a mid-to-large server where TPS drops hurt, or Paper-compatible modded setups where entity handling is your bottleneck.

Leaves — the newer experimental fork

Leaves is newer than the other three. It's a Paper-based fork that takes inspiration from Folia (Paper's experimental multithreaded server) and applies some of those ideas more aggressively to standard workloads.

Why Leaves is interesting:

• Active development with frequent patches
• Picks up new Folia-style optimizations before they land in Paper proper
• Config philosophy similar to Purpur — wider tuning surface

Why Leaves is risky:

• Less proven, fewer real-world deployments, thinner battle-testing
• Documentation is still catching up
• Plugin compatibility has occasional edge cases

Pick Leaves if: you like trying new things, have a backup plan if something breaks, and specifically want to run workloads where async scheduling matters.

What the benchmarks actually say

We run all four forks in production across different workload profiles. The honest short version:

• Under 10 players, vanilla-ish gameplay: all four behave identically. You will not notice a difference.
• 10 to 30 players with 15 to 25 plugins: Paper and Purpur feel the same. Pufferfish shows a marginal TPS improvement under spikes but nothing you feel day to day.
• 40+ players, heavy farms, large worlds: Pufferfish separates from the pack. TPS under load is meaningfully better. Paper still works but you'll see more hitches on mob-dense ticks.
• Modded Paper hybrid (Mohist or Arclight): stick with Paper. Modded compatibility layers target Paper specifically and struggle with fork-specific changes.

Your mileage will vary based on plugin mix, hardware, and playstyle. These are patterns, not guarantees.

How to switch forks (it's a 3-minute job)

Switching between Paper-based forks is safer and simpler than most people expect:

1. Stop the server and back up worlds folder + plugin configs
2. Replace the server jar — download the new fork's jar, drop it in place, update the start script if the filename changed
3. Start the server, watch the console for plugin warnings on first boot. Most plugins will just work.

On BruceNode you can do this from the panel's version picker — no SFTP required. One-click reinstall covers Paper, Purpur, Spigot, Fabric, Forge, and NeoForge.

The decision tree

• 2 to 10 players, no heavy plugins: Paper. Move on.
• 10 to 30 players, community server with gameplay tweaks: Purpur.
• 30+ players, TPS is your bottleneck: Pufferfish.
• Modded Paper hybrid: Paper. Don't fork further.
• Bleeding-edge curious: Leaves.

Final take

Most servers will never need anything other than Paper. The "which fork is best" conversation online can feel bigger than the actual performance gap on real servers. Paper is the right default. Purpur is the right upgrade when you want more knobs. Pufferfish is the right upgrade when you actually have a TPS problem. Leaves is the right bet if you enjoy experimentation.

We built BruceNode to make this specific decision painless — one-click installs for every major Minecraft fork, so you can try Paper today, swap to Purpur tomorrow, and benchmark Pufferfish next week without ever touching SFTP.

New customers: 20% off your first invoice with code BRUCE20. Try it at brucenode.com/minecraft-hosting.

Related reading:

• How to install Fabric on a Minecraft server (2026)
• Minecraft server RAM calculator
• Minecraft hosting plans