Best VPS for a Monero Node in 2026

VPS for a Monero node in 2026: pruned monerod on Prime $9.60 annual-eff, archive on Summit. XMR billing, 18080 open, Core from $4.80, Forge $39.20.
VPS for a Monero node in 2026: pruned monerod on Prime $9.60 annual-eff, archive on Summit. XMR billing, 18080 open, Core from $4.80, Forge $39.20.
12 locations · 19 coins · SLA 99.95% · no KYC · public canary
VPS from $4.80* — Core on the annual term (list $6/mo), one invoice for 12 months, no auto-charges. Dedicated with IPMI from $39.20/mo. Crypto invoice, no KYC.
Launch nowWhat the ranking demands of a node host
monerod publishes its requirements and never renegotiates them: NVMe with steady random-read behavior, memory with headroom for the transaction pool and the LMDB map, and an unobstructed P2P path on TCP 18080. The axes for this table are those three plus the house rules — settlement in XMR without identification, and an acceptable-use policy that receives nodes as tenants with rights rather than suspects on probation.
Metering is the quiet disqualifier. A public node converses with peers around the clock, and a per-gigabyte bill turns that conversation into a liability; unmetered ports are table stakes here, not perks. Burst-allocated memory earns the same verdict — chain databases fail in ways that mimic disk corruption — so only KVM-class platforms with documented allocation qualified for a row at all.
Sizing constants come from the project’s own documentation: roughly 70 GB for a pruned store, past 200 GB and climbing for a full copy as of 2026. Rival claims were re-checked on 2026-09-13 against public pages, and where an operator’s published specifications decided a placement, the placement says so in plain language.
Declared interest: the first row is ours, graded on the identical axes with cons printed like everyone else’s, and nothing on the page pays a referral. The ranking optimizes for the machine’s behavior under your workload — not for a vendor’s margin — which is why the location section later argues openly with folklore.
Speed of provisioning earned points too: a node that appears in days is a node you quietly stop patching. Credit-to-SSH in about a minute keeps the chain in the category of things you operate rather than things you wait for, and that difference compounds across every upgrade you will ever run.
Resource math for monerod
| Mode | Storage | Memory | Fit at VPSbit |
|---|---|---|---|
| Pruned (~2/3 kept) | ~70 GB | 4–8 GB | Prime $12 (160 GB NVMe) |
| Full node | ~200 GB, still climbing | 8–16 GB | Titan $24 (320 GB NVMe) |
| Node plus wallet | RPC as above, plus wallet data | 16 GB comfortable | Titan |
Begin from the chain’s dimensions. A pruned store — about two-thirds of history retained — occupies near 70 GB and enforces every consensus rule; a full copy passed 200 GB during 2026 and inherits growth with each new output the network creates. Memory: 4–8 GB carries pruned duty, and 8–16 GB keeps wallets, RPC callers and a sociable peer set from queueing behind each other.
Ports define the node’s role in the network. TCP 18080 carries P2P conversation and must face the internet; 18089 serves restricted RPC when you deliberately share the node with other wallets, best wrapped in TLS. From zero, NVMe completes the first sync in hours — spinning disks still measure it in days — and pruning freezes the footprint permanently afterward.
Wallet artifacts grow alongside: cache, address indices, view-only wallets. Reserve headroom beyond the raw chain figure — on Prime’s 160 GB a pruned store never revisits the question; Titan’s 320 GB gives a full node years of margin; Summit’s 2×2 TB of bare-metal NVMe ends the trimming conversation entirely, for the archives that must never forget.
One trap catches almost everyone exactly once: LMDB maps aggressively, so free disk outranks free RAM as the resource to guard. Mount the database on the local NVMe — never a network allowance — and leave a fifth of the volume unallocated so the map can stretch the way its documentation promises it will.
Four hosts, one table
| Host | Strengths | KYC |
|---|---|---|
| VPSbit | Billing in XMR natively, NVMe disks, unmetered traffic, node-tolerant AUP | None |
| Privex | Culture built around crypto users | None |
| FlokiNET | Privacy-first stance, present in 4 locations | None |
| 1984 Hosting | Icelandic jurisdiction, modest hardware | None |
Row one wins where nodes can feel it: Monero settles the invoice natively, NVMe backs every tier, ports run unmetered, the AUP names nodes as welcome, and 18080 answers at all twelve locations. Strong points: allocation documented on named silicon, credit-to-SSH in about a minute, a metal path for archives. Weak points: coins-only checkout, ticket support without a time commitment at entry, backups on the customer, and published premiums on four of the flags.
Privex’s row exists on the strength of its front door: built by crypto users, reachable through Tor or I2P, XMR native, identification absent. The ceiling is breadth — small plans, a short catalog, a Sweden-weighted map. A private remote node fits inside those walls with room to spare; a public node with strangers depending on it meets the RAM limit before it meets the community.
FlokiNET contributes the conviction column — privacy-first posture, four locations, cash-by-mail behind native XMR — discounted by hand provisioning, Iceland setup fees, and hardware that stays modest. 1984 Hosting supplies the jurisdiction column: Icelandic law, Monero accepted, one country, 1 GB-class entries, bandwidth that never sprints. Both are honest; neither was built to carry a fleet.
The exclusions were deliberate: no burst-memory reseller, no metered bargain, no operator whose abuse desk mistakes popularity for an offense. Those three filters remove more Monero-node candidates than any price comparison ever will — which is why this table is short, and why it can afford to be.
Disk planning and peer traffic
Plan around storage first. Pruned duty sits comfortably on Prime at $12; a public node that wallets depend on wants the whole chain — Titan’s 320 GB today, Summit’s 2×2 TB when nothing may ever be trimmed. Reversibility belongs to you on KVM: re-sync or resize at will, while a volume that fills mid-sync answers to nobody but the re-download queue and your calendar.
Mount the database on local NVMe, keep LMDB’s headroom, and let the unmetered port carry the permanent murmur of 18080 without arithmetic anxiety. At 2.5 Gbps on Prime or 10 Gbps on Titan, the initial download becomes an evening’s background task — the port stops being a variable you think about, which is the entire point of removing the quota.
Upgrades get snapshots, without exception: releases occasionally alter database behavior, and a rehearsed restore beats a re-sync every time it has actually been tested. Walk disk growth monthly, check new binaries against the project’s signed hashes, and refuse builds from anywhere else — validation is only as honest as the binary performing it, and that binary is your responsibility.
Peer health is the last habit: watch the connection count and the height against a reference you distrust for everything else. A node that loses peers after an upgrade is telling you something before the logs finish loading — listen early, and the fix stays small instead of becoming a re-sync with an audience.
When a second daemon or a demanding wallet service arrives, climb a tier rather than rationing: Prime for the personal role, Titan for public duty, metal for everything at once. Account, invoice and terms ride along unchanged through every step of that ladder, which is what makes the climb boring in the good sense.
Location choices without mythology
monerod runs identically at all twelve sites — same image, same ports, same unmetered uplinks — so choose by milliseconds to the wallets that will consult the node. Reykjavik attaches geothermal power and Icelandic practice with no mandatory data retention; Zurich adds Tier IV construction, a 100 Gbps roof, and Swiss FADP process beyond EU reach. Published premiums apply at both: +20% and +35% on list, respectively. Latency to the peers of a node matters more than the flag on the rack.
Latency arithmetic for Europe points at Frankfurt or Amsterdam and their peering depth, at base list; wallets concentrated in Asia find Tokyo, Singapore, Hong Kong or Taipei answering in the low hundreds of milliseconds. The invoice is issued from Saint Kitts and Nevis regardless — geography selects the law over the rack, never the questions asked at checkout.
The mythology worth retiring: the belief that one specific country legitimizes a node. Legitimacy comes from running near your users, under a statute you have actually read, on hardware whose specification you can verify. Everything else — retention practice, facility tier, peering depth — is published per location and equivalent in spirit across the entire map.
A pattern that survives contact with production: primary in Zurich for the wallets you own, a pruned standby in Reykjavik, failover arranged at the wallet or DNS layer. Two entry-tier machines undercut one oversized machine and shrug off a regional incident — arithmetic this site repeats often because it keeps turning out to be true.
Sockets, binaries and upgrade habits
Source discipline comes first: monerod, the wallets and the CLI arrive from the official getmonero downloads, hashes checked before the first run. Builds from anywhere else convert a privacy instrument into a listening post — the check costs you a minute and retires an entire class of supply-chain risk before it ever reaches the disk.
Sockets come second: RPC bound to localhost or a WireGuard interface, unrestricted mode never exposed, and 18089 in restricted form behind TLS for the cases where you deliberately serve other wallets. A remote node lends wallets a hand; it assumes none of the wallet-side hygiene on which your spends actually depend — that work stays with you forever.
Upgrades come third, as ritual: snapshot, apply, confirm height and peer count, retire the old image only after the numbers agree. Walk disk growth monthly, keep a fifth of the volume free for the database map, and delegate the kernel to unattended upgrades while you keep the chain. Nothing here is clever; all of it separates operating from improvising.
The same habits scale to metal when the archive grows — verify, restore from rehearsed images, treat unsigned binaries as hostile until proven otherwise. The project’s guidance covers each step, and the citation block below points at it directly rather than paraphrasing it from memory.
Bind the whole section together with a runbook written as you learn: which wallets point where, which port serves whom, when the last snapshot ran. Half a year later the note answers what memory cannot, and writing it cost you nothing but an evening’s honesty.
The case for running a public node
Every public node thickens the network’s decentralization, so hosting one is a contribution with a measurable shape: more independent validation, fewer wallets compelled to trust somebody else’s server. House it on a machine settled in XMR with no identification on file, and nothing anywhere ties the box to its operator — the exact property the network was engineered to protect and reward. Outbound RPC from your own box keeps trust in your own hands.
Stating the limits with equal clarity: the node’s address is public because usefulness requires it. Treat the machine accordingly — restricted RPC, no wallet files resident, no experiments you would decline to publish. And operating a node anonymizes nothing about your spending; that work belongs to wallet practice alone, and no port forwarding changes it.
A node is a utility, not a laundering service: it relays, verifies and serves blocks, full stop. Anyone marketing it to you as an anonymity appliance is selling folklore with a dashboard attached. The honest value proposition is narrower and better — your wallet stops interrogating strangers about your balances, and the network gains a peer it cannot quietly censor.
To give more than the minimum: leave the port open, raise the connection ceiling, and let the mesh spend bandwidth you never meter. If the same box backs your own wallet, split the sockets with intent — restricted RPC for the public, unrestricted mode confined to the tunnel. Service and privacy coexist on one machine exactly as long as the boundaries stay deliberate.
Disqualifiers, listed and priced
Quotas on a node are a design error before they are a billing error — 18080’s murmur is constant, honest, and exactly what you signed up to host, so metering it punishes the service itself. A close second: any AUP that treats popularity as suspicion, where a rising peer count becomes an abuse ticket and your public service quietly becomes supervised parole.
Platform disqualifiers follow: burst memory that fractures the database under neighbor pressure, oversold volumes that fold during sustained reads, virtualization without genuine isolation. Identity disqualifiers finish the sweep — identification at signup welds the node to your name forever, and payload logging relocates the ledger from your disk to theirs. The bar here: token or email-plus-password entry, no payload or guest traffic logs, metadata held at most 24 hours.
Half the disqualifiers are self-inflicted: unrestricted RPC facing the scanners, builds from unknown sources, upgrades without snapshots, wallet caches left on a public machine. The host repairs its half; only you can repair yours. The durable pairing is an operator who reads documentation and a host that bothers to write it down properly.
And one economic trap hides in plain sight: the rock-bottom node VPS whose specifications flatter themselves until sync week, when throttled reads stretch the initial download across a month. Price nodes in dollars-plus-downtime like everything else on this site — Titan’s $19.20 annual-eff with 320 GB of NVMe is the calm public-node floor, and bargains below it usually rent you the incident instead of the service.
The build we endorse, plus the map around it
Endorsed configuration: Prime at $12 ($9.60 annual-eff) — four 9950X cores, eight gigabytes of ECC, 160 GB of NVMe, an unmetered 2.5 Gbps port — running monerod pruned with validation complete, positioned in Zurich under Swiss FADP process with dependable EU transit. Archives without trimming go to Summit: 2×2 TB of NVMe from $89 list ($71.20 annual-eff), same account, same absence of identification anywhere in the flow.
The checklist that travels with it: binaries from the official source with hashes verified, RPC on localhost or WireGuard, 18089 restricted and TLS-wrapped if wallets are served, 18080 open to the world, snapshots preceding every upgrade, disk walked monthly. XMR on the invoice aligns the payment rail with the network’s posture; the public address stays public by design, and nobody pretends otherwise.
For the mechanics, How to pay a VPS with Monero details the settlement rail end to end, and Best Monero VPS 2026 grades hosts on their native XMR handling if the checkout itself is the deciding axis for you.
The wider map: Best VPS for a Bitcoin node for the other chain, Best crypto-only VPS for the market view, Bare metal with Monero (XMR) for the metal exit, Best privacy jurisdictions 2026 for the legal atlas — with the standing shelf: Buy a VPS with crypto, Offshore VPS, VPS vs dedicated bare-metal, Best no-KYC VPS 2026, FAQ: crypto VPS.
Growth, written in advance: Prime holds the personal role, Titan inherits the public one when strangers depend on the service, Summit takes the never-trim archive. The runbook, the coin and the checkout never change along the way — which is the entire point of buying infrastructure from a catalog instead of assembling it from exceptions.
Can other people point wallets at my node?
They can, if you let them: publish 18089 as restricted, TLS-protected RPC for the public, or keep the node inside a WireGuard tunnel for yourself. The first shares the service; the second reserves it for your own devices.
Does a node need a public address?
For P2P participation it does — peers must reach 18080 to find you. Every VPSbit server carries a dedicated IPv4 with a /64, so reachability never rides on an address some previous tenant burned.
Which protects privacy more — pruned or full?
Neither, and for the same reason: both validate the entire chain, and pruning only narrows the history your node re-serves to others. Your wallet receives identical verification either way, so decide by disk capacity, not by doctrine.
Is there an official guide to follow?
Yes — getmonero.org maintains the node guidance and the signed downloads, both linked from the citations on this page. Verifying hashes before the first run belongs to the same discipline, not to paranoia.
Can Core run monerod?
For one wallet’s pruned store, yes, tightly — four gigabytes holds a quiet personal node. Prime is the realistic minimum once RPC callers, wallet cache and a livelier peer set join, and Titan is the calm public tier.
What does the initial sync cost in time?
Hours on NVMe rather than days, shorter still with a prune window. Attach upgrades to attended windows with snapshots taken first, and the chain stops owning your evenings altogether.
Is a public node a privacy risk for me?
The address is public by design; the operator is not identifiable, because the account was opened without identification and settled in XMR. Keep wallet files off the box and RPC restricted, and exposure stays exactly where you chose to put it.
Primary sources
Ready to launch?
Build the box — VPS or bare metal — create the password, pay the invoice that follows.