TrueTick vs Exaroton: two pay-as-you-go hosts, compared

If you've decided that paying a flat monthly fee for a server nobody is on is a bad deal, you've probably already found Exaroton. It popularized hourly, pay-as-you-go Minecraft hosting, it has a loyal following, and it earned that following. This page is written by the paid competitor, so every claim we make about them below links to their own documentation — check it before you take our word for anything.

Here's the shape of this comparison up front, because most "vs" pages bury it: Exaroton meters by the gigabyte-hour, stops your server when it empties, starts it again when someone joins, parks them in a lobby while it boots, and shows live TPS. We do all of that too, and at July 2026 exchange rates their rate is lower than ours. Two things actually differ, and one of them is why this company exists.

Who each one is actually for

Exaroton is for players who want the cheapest metered hosting from an established name, on a CPU core shared with other servers — which, to be fair to them, is fine a lot of the time.

TrueTick is for players who want that same model on a core the platform refuses to share, and who want to see what's eating the tick, not just that the tick is slow.

The comparison

Both of us reject "pay full price whether you play or not". Below is where that stops being the same product. Checked July 2026 — every Exaroton row links to their own docs:

	Exaroton	TrueTick
Metered rate	1 credit per GB of RAM per hour; 100 credits = €1, so €0.01 per GB-hour. They don't bill whole hours either — they "calculate the exact price and round it up to the next hundredth of a credit"	$0.012 per active GB-hour — at July 2026 exchange rates, their rate is about 5% below ours
Always-on, per GB	€7.20 per month at 24/7 (30 × 24 h × €0.01) — "servers on exaroton don't have a fixed monthly price", metered is the only mode	$3/GB-month on the flat plan — at July 2026 exchange rates, under half their metered cost
Idle cost	€0 — AutoStop stops an empty server for you, free	$0 — sleeps on idle
Wake on join	AutoStart boots the server when someone connects, free	wake-on-join, same idea
While it boots	Offline waiting lobby — players wait in a holding world and are transferred when it's up	wake-limbo — same, players wait in-game and get transferred
CPU	0.5 shared cores per GB — "a shared CPU core doesn't mean exclusive usage rights for a specific core but general usage equal to a single core if possible"	A dedicated share per active server; admission control refuses to overcommit a node
Live metrics	player count, RAM, CPU and TPS, in their panel	the same, plus MSPT on Paper, Purpur, Forge and NeoForge — nothing to install. Where the core prints percentiles (Paper, Purpur, Vanilla and Fabric from Minecraft 1.20.3) we also show the p95 tail; Forge, NeoForge and older Paper/Purpur print none, and the panel says so instead of drawing a flat line. On those same 1.20.3+ cores the rate is counted off the world's own tick counter rather than inferred from tick time; below 1.20.3 Vanilla and Fabric have no tick command at all
RAM	2 to 16 GB self-serve; more if you "contact our support if you have a valid use case"	up to 16 GB self-serve; 24 or 32 GB on request

Where Exaroton is the right call

• You want the lowest metered rate. Theirs is lower. €0.01 per GB-hour against our $0.012 — at July 2026 exchange rates that's roughly 5% in their favour, every hour. If you're counting cents on a small server, that's a real difference and it isn't in our favour.
• Their argument for shared cores is a real argument, not a dodge. Their docs say assigning cores exclusively "would be much more expensive and is usually unnecessary since the full CPU power is only required during short load peaks." For a lot of servers that is simply true, and they price accordingly.
• You'd rather have a mature product with years of track record than a newer one.

If your server is small, quiet and occasional, this is a close call and their price wins it.

Where it hits a ceiling

Their case for sharing cores rests on one assumption: that load peaks are short and don't line up. For most workloads that's a good assumption. Minecraft isn't most workloads — prime time is the same Friday evening for everyone on the node. The peaks correlate. A shared core is at its worst exactly when your server is at its busiest, which is exactly when you're watching.

That's the assumption we don't make:

• A core we don't share. Every active server gets a dedicated share, and admission control blocks creating or starting a server the node can't actually carry. When we're full, the panel says we're full instead of selling you a seat anyway.
• MSPT and its tail, not just TPS. Exaroton shows TPS, so you'll see the drop. MSPT is the tick budget — it tells you what is eating the tick, which is the part you can act on. On Paper, Purpur, Forge and NeoForge it's in the panel with nothing to install; on Vanilla and Fabric it arrives from Minecraft 1.20.3. The p95 tail is where lag actually lives — a world can average a comfortable 48 ms and still stutter — but only cores answering the vanilla tick command print percentiles, so the tail appears on Paper, Purpur, Vanilla and Fabric from 1.20.3 and nowhere else. Where a core prints none, the panel says so rather than guess.
• A flat plan for when metered stops making sense. If the server really is up 24/7, metering is the wrong tool at any rate — theirs works out to €7.20/GB-month against our $3/GB-month flat, which at July 2026 exchange rates is under half.

Verdict

If you want the cheapest metered hosting from an established name, and a shared core is fine for your server, take Exaroton. They're good at this, at July 2026 exchange rates their rate beats ours, and we'd rather say so than pretend otherwise.

If you want the same model on a core the platform refuses to oversell, MSPT for when something does go wrong, or a flat plan that costs less than metered on an always-on server — check the maths for your server or create one and watch the tick line yourself.