Average latency by method
How fast each provider handles a specific request type across all regions. Pick a method to compare.
8.52 ms
17.39 ms
20.08 ms
107.18 ms
- Alchemy
- dRPC
- QuickNode
- Infura
Compare RPC performance across popular chains
See how different providers compare on latency, reliability and failed requests, all tested under identical conditions.
Global Average Response Time
LiveLast updated: Jun 26, 2026, 16:51 UTC
Alchemy0.00 ms
QuickNode0.00 ms
Infura0.00 ms
dRPC0.00 ms
Average response time across all methods in the current global benchmark window.
Chain
Region
Average Latency
0.00ms
Alchemy
P50 Latency
0.00ms
Alchemy
P95 Latency
0.00ms
Alchemy
Success Rate
0%
Alchemy
| Provider | Avg latency | P50 latency | P95 latency | Success rate |
|---|---|---|---|---|
| AlchemyFastest | 15.87 ms | 5.95 ms | 30.61 ms | 100% |
| QuickNode | 45.19 ms | 15.42 ms | 180.14 ms | 100% |
| Infura | 113.97 ms | 93.34 ms | 276.08 ms | 100% |
| dRPC | 130.20 ms | 25.06 ms | 661.06 ms | 99.46% |
Failed requests by method
Which request types each provider struggles with across all regions. Pick a method to see where errors pile up.
0
9
47
72
- Alchemy
- QuickNode
- Infura
- dRPC
Methodology
Same test, same rules, every provider
Every provider is measured under identical conditions, so the numbers stay comparable and hard to game.
Tested every 10 seconds
Common EVM RPC methods run against each provider on the same fixed cadence, around the clock.
Four global regions
Latency and success rate are measured from US East, US West, EU Central, and AP Southeast.
Standard paid accounts
Every provider is tested on the same standard paid RPC service tier, with no special treatment.
Same method set
Every provider is hit with the same mix of common EVM calls, from simple reads to heavy getLogs queries.
Clear failure rules
Non-200 HTTP responses, non-zero RPC statuses, and requests over 8 seconds all count as failures.
Transparent results
Raw benchmark data is published as machine-readable tables, so anyone can audit and reproduce the numbers.
Benchmarks for agents
The companion Markdown page returns this data as text tables with field definitions and source metadata, built for agents, crawlers, and anyone who wants the data without the interface.
Open Markdown data
Latency is a tax on every request
Slow RPC makes everything slower. Slower reads, slower UIs, slower confirmations, until users give up. Build on the infrastructure that leads these benchmarks.