The Hidden Cost of "Free" RPC: How Compute Units Become Surprise Bills

Introduction

You wire up a free RPC endpoint, point your app at it, and for weeks billing never crosses your mind. Then a query pattern changes or traffic doubles, and one of two things happens: the requests start failing, or an overage charge lands that you never watched building. Neither is bad luck. Both come down to a detail most pricing pages leave out, which is what a single request costs to serve.

That cost is the real variable behind an RPC bill, and it moves far more than the headline price per unit suggests. Two providers can quote the same rate per million compute units and still charge you very differently for the same app, because what you pay is driven by the methods you call, not the number on the pricing page.

How compute units are calculated

RPC platforms bill in compute units rather than flat request counts, because the work behind requests is not uniform. Returning one account balance is cheap. Scanning a wide block range or replaying a transaction is not. A compute unit is the weight a provider puts on each method to reflect that difference.

Luganodes publishes the formula, which makes the mechanic easy to see: the units for a request are the chain's base rate times the method's weight. On Ethereum the base rate is 20, so a routine eth_getBalance costs 20 units. A log query like eth_getLogs weighs 2x and costs 40. A deep trace such as trace_replayTransaction weighs 4x and costs 80. Bitcoin sits at a base rate of 5, so a plain read is 5 units and a wide scan like getrawtransaction is 20. Batch calls cost the sum of their parts. Read the weights once and the invoice stops surprising you.

Why a low price per unit can still cost more

Method weights are not standardized across the market. A call that one provider prices at a single unit can cost a couple of hundred somewhere else, so the advertised rate per million units tells you almost nothing by itself. A cheap-looking sticker price attached to heavy weights can work out dearer, per real request, than a rate that looks higher on paper.

What you want to compare is the effective cost of your own traffic: take the methods your app leans on most and price those, not the marketing number. A wallet that mostly reads balances and an indexer that lives on log queries will get very different mileage from an identical price sheet.

The methods that quietly run up the bill

Overages almost always come from the same short list. Log queries, traces, debug calls and wide state scans carry the heaviest weights, and the workloads that depend on them, indexers, dashboards and DeFi backends, hit them thousands of times a day. An eth_getLogs call left open across a large block range is the textbook offender: it works, it returns what you asked for, and it burns through an allowance faster than anything else in your code.

On a pay-as-you-go plan the meter keeps running past your included quota and settles up at the end of the month. That is where the invoice nobody budgeted for comes from. The metering is not the villain. The damage comes from a meter with no ceiling, sitting behind a pricing page that never showed you the weights.

Keeping your usage predictable

Most of the fix is habit. Learn which methods in your stack are the expensive ones, and bound the block ranges on log queries instead of scanning wide. Read per-method usage in your dashboard rather than waiting for the bill, and put alerts on usage and error spikes so a runaway loop shows up in minutes, not at cycle close.

A good provider makes that the default. Luganodes meters every method in units you can audit, and its free tier is hard-capped at 25 million compute units a month with no card: when you hit the limit, traffic pauses instead of converting into a charge, so a mistake costs you a stalled request, not a bill. Paid tiers raise the ceiling with overage you can predict, and AI Insights points to the exact methods eating your budget and the cheaper patterns that would replace them.

Priced this way, an entry tier does what it claims, and the number at the end of the month is one you chose rather than one you discovered. Spin up a free endpoint at rpc.luganodes.com and watch where your compute units go before they add up.

About Luganodes

Luganodes is a world-class, non-custodial blockchain infrastructure provider that has rapidly gained recognition in the industry for offering institutional-grade services. It was born out of the Lugano Plan B Program, an initiative driven by Tether and the City of Lugano. Luganodes maintains an exceptional 99.9% uptime with round-the-clock monitoring by SRE experts. With support for 40+ PoS networks and serving 60+ institutional cliens, it ranks among the top validators on Polygon, Polkadot, Sui, and Tron. Luganodes prioritizes security and compliance, holding the distinction of being one of the first staking providers to adhere to all SOC 2 Type II, GDPR, and ISO 27001 standards as well as offering Chainproof insurance to institutional clients.

The information herein is for general informational purposes only and does not constitute legal, business, tax, professional, financial, or investment advice. No warranties are made regarding its accuracy, correctness, completeness, or reliability. Luganodes and its affiliates disclaim all liability for any losses or damages arising from reliance on this information. Luganodes is not obligated to update or amend any content. Use at your own risk. For specific guidance, please consult a qualified professional.