How to compare Cobblemon server hosts

Which Cobblemon build are you actually hosting?

"Cobblemon" on a plan page can mean the mod alone or a pack with well over a hundred parts, so pin the build before comparing anything. Modrinth showed Cobblemon 1.8.4 for Fabric on Minecraft 1.21.1, published 27 October 2026. Cobbleverse, built on Cobblemon 1.7.3 according to Shockbyte's pack page, had a newest Modrinth release of 1.7.42 from 21 July 2026, declaring 186 dependencies. Sizing advice written for one does not transfer to the other.

Modrinth marks both projects as required on client and server, so every player has to match the server. Ask whether a host can hold your version until the group is ready to update.

Licence terms can also decide this. Shockbyte reproduces Cobbleverse's terms: its custom structures, trainers, questlines, datapacks and configurations cannot be used on a public server that runs a commercial shop, and a server may not be named a "COBBLEVERSE Server". Settle that before you plan to sell ranks.

Does Cobblemon publish a RAM figure?

No hardware figure was found on 8 November 2026. The official site shows only its title without JavaScript. A wiki search for "server RAM" returned one changelog page, and the Config page lists settings without hardware guidance. The whole wiki was not read, so a figure could exist elsewhere.

Why do spawns and scattered players decide it?

Cobblemon spawns Pokémon in zones around players, so a group that scatters is the case to plan for. These defaults are the ones a Cobblemon owner can act on. None of the sources puts a server cost on them.

Defaults move between versions. The wiki page, last edited 1 February 2026, gives a default box count of 30 where the current source gives 40, so read your own server's main.json.

Other load has no published cost either. Modrinth says battles use the Pokémon Showdown engine, and Cobbleverse adds 1 to 7 star raids per Shockbyte. Space-Node, which sells hosting, says riding means owners should "expect faster exploration and more chunk loading" (5 May 2026). See MSPT and TPS explained for the 50 ms budget.

Whose Cobblemon RAM figures conflict?

Recommendations for the same pack differ by 2 GB, and figures spread further across server sizes.

Pack size is contested too. Pocketcraft's calculator groups Cobbleverse with 50 to 120 mod packs, while the Modrinth release declares 186 dependencies, which may include non-mod content. For how plan RAM becomes heap, see Java flags for modded servers and the cross-pack framework.

Which hosts sell a Cobblemon server, and on what terms?

Two hosts were confirmed to offer a Cobblemon pack on pages read 8 November 2026, listed in no order of merit. Neither has been tested, and other hosts may offer Cobblemon too.

On published structure only: Shockbyte's pack page gives the smaller figure (4 GB minimum, 6 GB recommended) and Sparked Host's directory gives 8 GB for the same pack, so the plan size you buy differs by the publisher you believe. Shockbyte's 72-hour window is longer than Sparked Host's 24 hours, and Sparked Host's is discretionary. Both rows are claims.

What should you ask a Cobblemon host before paying?

Ask these six questions, in writing where you can, because no plan page read for this page answers all of them.

1. Can the host install your exact pack version and hold it there until you choose to update?
2. How much of the plan's RAM becomes Java heap, and can you set it?
3. Can you edit main.json and the spawn settings above?
4. What CPU will your server actually run on?
5. Is world pre-generation limited? Shockbyte's terms (28 August 2026) say any pre-generation "must not exceed a radius of 10K" and world files above 5GB "must not include pre-generated chunks".
6. How long is the refund window? Shockbyte states 72 hours from ordering, once per client. Sparked Host's terms (29 October 2026) make Minecraft refunds discretionary, for unresolved technical issues, requested within 24 hours.

How can you judge a Cobblemon plan on your own server?

Use your real group during a normal session, after reading your host's terms, and compare the same server in four states. Silence in terms is not permission, and bots or load tools need the host's written agreement first; see independence and host terms.

1. Everyone standing at spawn.
2. Everyone travelling together.
3. Everyone walking in different directions into new terrain.
4. A raid or a busy base, if your pack has them.

The spark tick-loop guide describes the 20-tick, 50 millisecond target. If MSPT climbs only in state 3, suspect world generation and spawning. If it climbs everywhere while memory is comfortable, more RAM is unlikely to help. How to benchmark your own server covers the readings.

What would be measured, and why?

The planned protocol would run one pack on three hosts with players moving through new terrain and triggering spawns and raids. The load tool has not been chosen, and load profiles notes that bot movement may not trigger realistic spawning.

Test-server budget is not approved, no server exists, and each host's terms must be checked first. Next: the benchmarks hub or the method overview.