Arbitrum node
A dedicated Arbitrum node. Reads that never wait in line.
An Arbitrum node you rent for yourself answers one customer: you. Liquidation bots, dApps and indexers get steady reads from a server sized for Arbitrum, and no neighbour can eat your request budget.
- Full
- $139/mo
- Archive
- $289/mo
- Ready
- 24 h
02Full or archive
Arbitrum full node or Arbitrum archive node?
Both speak the same JSON-RPC. They differ in how much past state can still be asked for.
Full node
$139/mo
- Keeps
- Current state, pruned, with enough recent history for live apps
- Query
- Contract reads, balances, logs, fee estimates and sending transactions
- Cannot
- Return state for blocks the node has pruned away
- Disk
- Arbitrum lists a pruned snapshot of about 2.3 TB (HashDB) or about 2.4 TB (PathDB)
Archive node
$289/mo
- Keeps
- Every state since the Nitro upgrade
- Query
- Reads and balances at any Nitro era block
- Watch out
- Blocks from before Nitro (Classic) sit in a separate system, covered below
- Disk
- Arbitrum lists an archive snapshot of about 3.7 TB
A fast way to choose
Does your code ask what is true right now? Full node. Does it ask what was true at an old block? Archive node.
Sources: Arbitrum docs, run a full node, checked October 2026; Arbitrum docs, run an archive node, checked October 2026
03How it works
How an Arbitrum node works: one program, two jobs, one eye on Ethereum.
Arbitrum is built with Nitro, a single node program. It reads Ethereum for the truth and the sequencer feed for speed.
Ethereum L1
Supplies the posted batches and finality. Nitro needs an Ethereum RPC and a beacon RPC.
Batch reader
Reads batches from Ethereum and fresh blocks from the sequencer feed.
Execution engine
Runs the EVM and stores state, in HashDB or PathDB.
Your endpoint
JSON-RPC on nodes.herorpc.com. The node itself uses port 8547.
An Arbitrum node cannot sync from peers. It takes its blocks from Ethereum batches and the sequencer feed.
That is why Arbitrum docs open with why you need an L1 node at all: it is the source of truth.
Your app never touches either feed. It sends JSON-RPC to one URL and gets answers.
04Know the field
HashDB or PathDB: the choice you cannot undo.
Nitro stores state in one of two ways. You pick when the database is created, and Arbitrum says an existing database cannot be switched.
Both docs pages are undated, so treat the sizes as guides and expect them to grow.
HashDB (default)
- Full
- Pruned snapshot about 2.3 TB. Pruning runs offline and can take days
- Archive
- No archive snapshot listed
- Source
- Arbitrum docs, full node page
PathDB (v3.9.x or later)
- Full
- Full-path snapshot about 2.4 TB. Pruning runs online
- Archive
- Archive-path snapshot about 3.7 TB
- Source
- Arbitrum docs, full and archive pages
Nitro is the only node software for Arbitrum, so the real choice is the database. PathDB cannot validate. This describes the chain, not a promise about how a given node is configured.
Sources: Arbitrum docs, run a full node, checked October 2026; Arbitrum docs, run an archive node, checked October 2026
05Arbitrum, specifically
Six Arbitrum facts the dashboards never mention.
Small details from the official docs that decide whether a node stays up.
No peers to find
Arbitrum has no peer to peer sync. A node follows Ethereum batches and the sequencer feed instead.
Pruning can stop the RPC
Pruning HashDB is offline work. On Arbitrum One it can take days, and the node serves no RPC meanwhile.
Classic is another era
Arbitrum One ran Classic before Nitro. Nitro still answers most methods for those blocks through a classic redirect, on Arbitrum One only.
Three ports
HTTP RPC on 8547, WebSocket on 8548 and the sequencer feed on 9642.
A snapshot starts it
The first run on Arbitrum One needs a snapshot passed in with the init URL flag.
Stop it slowly
Arbitrum advises a stop timeout of about half an hour (1800 seconds in its example), so the node can finish writing its state.
Sources: Arbitrum docs, run a full node, checked October 2026; Arbitrum docs, node FAQ, checked October 2026; Arbitrum docs, node troubleshooting, checked October 2026
06Who rents one
Who rents a dedicated Arbitrum node.
Teams whose Arbitrum calls are worth more than the cost of a private lane.
DeFi bots and keepers
Liquidations and arbitrage punish a slow read. A private node answers without sharing a rate limit.
Indexers and analytics
History queries across Nitro era blocks call for the archive node.
Orbit and L3 builders
Chains built on Arbitrum One read their parent chain all day. An endpoint of your own keeps that dependable.
Wallet and exchange backends
Balance checks and deposit scans at volume, away from the daily caps of a free endpoint.
07Price check
An Arbitrum node next to the usual price tags.
Allnodes lists a price per node type, which makes the bars fair. The other two providers hide the final number.
Full node
Archive node
Allnodes: the cheapest dedicated plan on its Arbitrum pricing page.
Source: Allnodes Arbitrum pricing, checked October 2026
Chainstack
Plan from $199 a month
Dedicated nodes need a Pro, Business or Enterprise plan, and the Pro plan alone starts there.
Chainstack pricing, checked October 2026
QuickNode
Price on request
A cluster starts at 2 nodes, and the price comes as a quote.
QuickNode clusters page, checked October 2026
You save each month
$181 full, $351 archive
against the Allnodes list price above
08What you get
Your Arbitrum RPC node, billed the way you choose.
Pick the node type and how long to prepay. The price updates as you choose.
Your Arbitrum node
Node type
Term
- A dedicated server sized for Arbitrum
- Nitro with its Ethereum feed already attached, one endpoint
- A backup node, included
- A private endpoint on nodes.herorpc.com
- Your server logins
- No setup fee, ready within 24 hours
$139
a month, full node
Billed monthly
Pay in BTC, ETH, USDT, USDC, BNB, SOL, TRX, LTC, DOGE or XMR.
10Questions
Arbitrum node questions
Real questions from node operators, answered plainly.
Why do I need an L1 node to run an Arbitrum node?
Because Arbitrum takes its transaction batches and finality from Ethereum. Nitro needs an Ethereum RPC and a beacon RPC to know what has been posted. Both come from our own Ethereum node with your Arbitrum node, so you bring nothing.
Can I run an Arbitrum node in p2p mode?
No. Arbitrum nodes do not sync from peers. They follow batches on Ethereum and the sequencer feed.
Is there a way to get pre-Nitro archive data from a Nitro node?
Partly. Nitro serves most methods for blocks before the upgrade through a classic redirect, on Arbitrum One only. Full Classic archive access needs a Classic node as well.
How can I tell that my Arbitrum node is fully synced?
Call eth_syncing, which returns false once the node has caught up, then compare eth_blockNumber with an Arbitrum block explorer.
What are the minimum hardware requirements for a full node?
The official run page gives an example setup rather than a fixed minimum. Your rented node arrives on a server sized for Arbitrum, so there is nothing to match by hand.
What does "Head state missing, repairing" mean?
The node cannot find its latest state and is rebuilding it. It tends to show up after a node was stopped before it finished writing, which is why Arbitrum advises a very long stop timeout.
Can I switch an Arbitrum node from HashDB to PathDB?
Not in place. Arbitrum says an existing database cannot be switched, so you start a fresh one with the other scheme.
Why does pruning take my Arbitrum RPC offline?
Pruning HashDB runs while the node is stopped, and on Arbitrum One that can take days. PathDB prunes while it keeps serving.
How much is an Arbitrum node each month?
Full is $139 and archive is $289. Nothing is charged for setup, and paying 3, 6 or 12 months up front takes 10, 15 or 25 percent off.
When does my Arbitrum endpoint go live?
Inside 24 hours of your payment, with the endpoint and logins waiting in your dashboard.
Put your Arbitrum reads in the fast lane.
From $139 a month for full, $289 for archive. Your backup node sits alongside, and the endpoint arrives within 24 hours.