How are these benchmarks measured?

A black-box prober runs standard public RPC/DAS calls against each provider from multiple regions and records wall-clock latency, success and response size.

Why might my numbers differ?

Latency depends on region, plan tier and time of day. We record the tier used in the methodology and measure from fixed regions.

Which percentiles are shown?

p50/p95/p99 latency over successful calls, plus uptime and error rate over all calls.

Solana & EVM RPC benchmarks

Last updated: June 20, 2026

Provider latency & reliability (30d)
Provider	Probe	Region	p50 ms	p95 ms	p99 ms	Uptime %	Error %	Samples
No benchmark data for this window yet — the multi-region prober is collecting samples.

Methodology

A prober in each region runs a fixed set of standard public RPC/DAS/WS calls.
Each call is measured once (wall-clock); a timeout counts as a failed sample — retries never heal a measurement.
Samples roll up into p50/p95/p99 (successful calls only) plus uptime and error rate per window.
Comparable paid tiers are configured per provider and recorded here.

Download CSV

FAQ

How are these benchmarks measured?: A black-box prober runs standard public RPC/DAS calls against each provider from multiple regions and records wall-clock latency, success and response size.
Why might my numbers differ?: Latency depends on region, plan tier and time of day. We record the tier used in the methodology and measure from fixed regions.
Which percentiles are shown?: p50/p95/p99 latency over successful calls, plus uptime and error rate over all calls.