Transparent vs Compute-Unit RPC Pricing: What It Actually Costs at Scale

Most RPC providers price usage in compute units, credits, or request units. Every method call costs a different, provider-set amount, so two teams sending the same request volume can land on very different bills. Instanodes prices Nodes-as-a-Service in flat monthly tiers with fixed request ceilings instead, across 50+ blockchains, backed by a 99.95% uptime SLA. The difference only shows up once your usage pattern changes.

A team building a dashboard and a team running an indexer send very different request mixes to the same chain. Under compute-unit pricing, those two teams pay dramatically different amounts for what looks like similar volume on a usage graph. Under flat, request-based pricing, both know their ceiling before they sign up.

Why the bill is hard to predict in the first place

Alchemy and QuickNode both meter usage by request type. A balance check costs a small number of units. A heavier call, an eth_getLogs, a trace method, or an archival read, costs far more, sometimes ten to twenty times as much. Alchemy's own documentation prices eth_getTransactionReceipt at 200 compute units and eth_getLogs at 75 units per log returned. One event-heavy contract during a busy hour can generate a bill nobody budgeted for.

Infura took a different route in 2026, moving from monthly quotas to daily credit caps. The free tier now allows 3 million credits a day at 500 credits per second. The Developer tier raises the daily cap to 15 million. The Team tier goes to 75 million a day. Daily caps solve one problem, runaway monthly overages, and introduce another: a single traffic spike can knock an account offline for the rest of the day once the cap resets at midnight rather than continuously.

Chainstack prices in request units. Its Business plan, launched in April 2026, runs $499 a month for 200 million request units and 600 requests per second, with overage priced at $10 per million units above the cap. That's predictable, as long as usage stays under the line. Cross the line regularly and the plan below stops making sense while the plan above feels like overkill.

None of this is a design flaw. Metered pricing lets a provider charge close to the real compute cost of each request. The trade-off lands on the customer, who has to model request mix, not request volume alone, to know what next month's invoice looks like.

What flat, usage-based RPC pricing looks like instead

Instanodes runs pricing as a small set of tiers, each with a fixed daily and monthly request ceiling rather than a per-method meter:

Six-month and annual billing bring the price down further, with annual running 14% below the monthly rate. Above Advanced, Instanodes builds custom Enterprise plans around actual workload rather than a fixed tier.

The practical effect: an eth_getLogs call and a balance check count the same against a quota. A team doesn't need to audit which RPC methods its application calls most often just to predict next month's cost. Instanodes states this model saves teams 30 to 50% at high volumes compared with QuickNode's credit jumps, a figure published on Instanodes' own pricing page rather than one calculated for this comparison.

If you're evaluating a provider for a new project, the free tier is enough to test your request pattern before committing to a paid plan.

What the numbers look like at real volume

Take a team sending roughly 10 million requests a month with a moderate mix of standard and heavier calls.

• Alchemy or QuickNode: this volume might sit comfortably inside a mid-tier plan, until a spike in eth_getLogs or trace calls during a busy week pushes the account into overage pricing for the rest of the billing cycle.
• Chainstack: 10 million request units a month fits well inside the Business plan's 200 million allotment, so cost stays flat unless the team runs several products on one account.
• Instanodes: the same volume sits inside the Basic tier at $79 a month, with the ceiling and the SLA both fixed regardless of which methods get called.

The gap widens as volume grows. Heavier archival and analytics workloads are exactly where compute-unit and request-unit pricing charges the most per request, and exactly where flat, request-based pricing stays the same per request.

Questions worth asking before you sign an RPC pricing contract

1. What does a single heavy method cost relative to a basic balance check, an archival read or trace call specifically, on the plan you're evaluating?
2. Do overage charges apply per request, per day, or per month, and what happens the moment usage crosses the line mid-cycle?
3. Does the provider publish real numbers for its cost model, or only a marketing page with no specifics?
4. What does your last full billing cycle actually look like, pulled from your own request logs, not estimated volume?

Talk to us with your numbers and we'll tell you which tier fits before you commit to anything.

FAQ

What is compute-unit pricing for RPC providers?

Compute-unit pricing charges usage based on the type of request made rather than a flat request count. Alchemy and QuickNode both use this model. A heavier method such as eth_getLogs or a trace call consumes far more units than a simple balance check.

Why does RPC pricing get expensive at scale?

Costs rise unevenly because heavier methods, archival reads, trace calls, and log queries, consume many more units per request than basic reads. A team's bill depends on which methods get called, not on total request volume alone, which makes forecasting difficult under a metered model.

How does Instanodes price RPC usage differently?

Instanodes charges flat monthly tiers with a fixed daily and monthly request ceiling, starting at $0 for the Free tier and running through Build, Basic, and Advanced plans up to $169 a month, plus custom Enterprise pricing above Advanced. Every request counts the same against the quota regardless of the method called.

How does a self-hosted node compare on cost with a managed RPC provider?

Self-hosting means paying for fixed hardware and taking on monitoring, upgrades, and failover yourself, with cost staying flat regardless of request volume. A managed provider like Instanodes charges a fixed monthly fee per tier that covers infrastructure, monitoring, and the uptime SLA, without hardware or operational overhead.

How do I choose an RPC provider based on price?

Compare providers on the request mix you actually send, not the headline price. Compute-unit and request-unit providers charge widely different amounts for the same volume depending on which methods dominate your traffic. Ask for a cost estimate based on your real logs before committing to any plan.

Does Instanodes charge more for archival or heavy RPC methods?

No. Archival access is included in Instanodes' paid tiers and priced by plan, not metered per method. An archival read costs the same against a quota as a standard call.

If your current provider's invoice doesn't match what you'd expect from your traffic, that's usually a request-mix problem, not a volume problem, and it's worth checking before your next renewal.