Composite score
77.7
Rank 9 of 12
326 samples
Providers
Ethereum RPC and transaction services. The public endpoint implements a reduced method set.
commercial · rank 9 of 12
Website (opens in a new tab)Documentation (opens in a new tab)
1h window · 60 minutes observed · 1 chain
77.7
Rank 9 of 12
326 samples
100.00%
Rate limits and unsupported methods excluded
139.2ms
p50 113.9 · p99 139.2
100.00%
5 responses judged
98% of requests were rate limited (286 of 291); excluded from availability
12 providers scored against each other.
| Component | Weight | Measured | Score |
|---|---|---|---|
| Availability | 40% | 100.00 % | 100.0 |
| Latency | 30% | 139.22 ms | 33.1 |
| Correctness | 20% | 100.00 % | 100.0 |
| Head freshness | 0% | not applicable | — |
Failures are classified rather than collapsed into one count. Our own failures are shown for transparency but are never counted against the provider.
| Outcome | Count | Share |
|---|---|---|
| Successful | 5 | 1.53% |
| Provider failure | 286 | 87.73% |
| Network failure | 0 | 0.00% |
| Rate limited | 286 | 87.73% |
| Unsupported method | 0 | 0.00% |
| Agent error | 0 | 0.00% |
| Measurement error | 35 | 10.74% |
The only figures on this site that are not measurements. A limit is a claim the operator makes, so it is quoted with the page it came from and the day it was read — never inferred from a refused request, which would put a number against a company that never stated it.
Requestson Ethereum25 / second
per account
“Free up to 25 RPS. Contact us for plans if more RPS is required.”
Quoted, not measured. Read 2026-08-27 from www.merkle.io— a provider can change this without telling anybody, so the date is part of the figure.
The status page judges every provider against an objective this site publishes. That is not an SLA, and a provider that misses it has not broken a promise — so what it actually promised, where it promised anything, belongs here in its own words. The objective
Merkle does not publish an availability commitment that could be found in its own documentation. That is the common case for a public endpoint and is worth knowing before depending on one: the availability figures on this site are measurements against a bar this platform set, not against anything Merkle agreed to.
An availability percentage has no shape. This is the shape: what departed from this provider’s own normal behaviour, and what its normal behaviour already is. How an incident is decided
None. Across the last 30 days, Merkle did not depart from its own usual behaviour on any measured call for long enough to count as an incident.
Calls that fail, or are refused, at about the same rate every day. These are not events — they are what this provider does — and none of them will be different tomorrow.
The same measurements, grouped by the vantage point they were taken from. Latency is a property of a network path, so a provider fast from one region is not necessarily fast from another — and the composite score above is an average across whichever regions are reporting.
A live badge for Merkle on Ethereum, free to use. It renders from the current measurement each time it loads, so it cannot be frozen at a good day — which is what makes it worth putting in a README. Every badge and parameter
The composite score, coloured by band.
Markdown
[](https://rpcnode.io/providers/merkle)HTML
<a href="https://rpcnode.io/providers/merkle"><img src="https://rpcnode.io/badge/provider/merkle.svg?chain=ethereum" alt="Merkle on RPCNode.IO"></a>