Quicknode vs Alchemy: 82 Days of Solana RPC Latency Data
Quicknode led the daily Solana RPC latency comparison vs Alchemy on all 82 scored days, with lower latency in 91.4% of 11,136 head-to-head measurements.
SOC 2 Type II Certified · ISO 27001
October 6, 2026 — 8 min read
Alchemy and Quicknode both run Solana RPC at scale. To compare them on neutral ground, we used the Solana Foundation's public RPC latency monitor: an open-source runner that probes six providers the same way from 29 locations.
On that monitor, Quicknode led the daily latency comparison on all 82 scored days, from July 16 through October 5, 2026. Across 11,136 head-to-head measurements, Quicknode recorded lower latency in 91.4%.
Quicknode, on 11 of the 12 Solana read methods the Solana Foundation measures. Here is the result in numbers:
Quicknode recorded lower latency in 10,182 of 11,136 head-to-head measurements (91.4%) across 12 read methods, 29 locations, four latency measures, and eight time windows.
Quicknode led the daily comparison on all 82 scored days, July 16 through October 5, 2026.
Light calls ran about 11x faster in the 24 hours to October 5, 2026: getSlot P50 was 1.9ms for Quicknode and 20.5ms for Alchemy.
Heavy reads showed the largest absolute gaps: recent getBlock P95 was 78.0ms vs 471ms, and getProgramAccounts P95 was 108ms vs 500ms.
The one close call is archival getTransaction, where Alchemy posted the lower 24-hour medians.
The Solana Foundation runs rpc-latency-monitor as a neutral measurement party for its VIP Trading Program. The Foundation doesn't run RPC infrastructure. It probes every provider the same way from the same vantage points, and the code is open source under Apache 2.0. Results are public on the Solana Foundation RPC dashboard.
The dashboard covers 29 locations on four infrastructure types: Teraswitch, Latitude, AWS, and Google Cloud. Each location probes 12 read methods and reports average, P50, P95, and P99 latency, plus error rate, for each provider.
The probes are built so a provider can't win on cached or empty answers:
No-cache requests. Every request sets no-cache headers, and targets move each cycle. Account reads rotate over real accounts seen in recent blocks rather than one fixed address.
Real work. Recent getBlock fetches full transaction details, and getProgramAccounts returns real account data across a rotating set of query shapes.
Empty answers fail. A 200 response with an empty, null, or truncated payload counts as a failure, not a fast success.
Archival reads are randomized. Every provider gets the same random, never-repeated block from roughly 40 million slots back, so nothing can be pre-cached.
We pulled every location, method, and time window the dashboard publishes on October 5, 2026, between 18:10 and 18:19 UTC. The data runs through October 5, 2026 at 18:10 UTC, and the monitor's history starts on July 16, 2026.
For each combination, we compared Quicknode's and Alchemy's reported values for average latency, P50, P95, and P99. Each value is the mean of the dashboard's data points for that series across the time window. That gives 11,136 head-to-head measurements: 29 locations, 12 methods, 4 latency measures, and 8 time windows.
For the daily view, we scored each day on its own. A day goes to whichever provider recorded lower latency in more of that day's measurements, up to 1,392 per day (29 locations, 12 methods, and 4 latency measures). Days where fewer than 80% of series reported are left out.
In the 24 hours to October 5, 2026, Quicknode had the lower P95 on 11 of 12 methods. The gap ranges from about 2x on archival getBlock to more than 10x on light calls.
Each figure is the median across 29 locations of that provider's reported value for the 24-hour window ending October 5, 2026 at 18:10 UTC. Lower is faster. The last column counts the days, out of 82, on which Quicknode led on that method.
Method | Quicknode P50 | Alchemy P50 | Quicknode P95 | Alchemy P95 | Days Quicknode faster (of 82) |
|---|---|---|---|---|---|
getHealth | 1.9ms | 20.4ms | 3.5ms | 33.8ms | 82 |
getSlot | 1.9ms | 20.5ms | 3.0ms | 34.7ms | 82 |
getLatestBlockhash | 1.9ms | 20.4ms | 3.1ms | 35.7ms | 82 |
getAccountInfo | 3.3ms | 20.5ms | 7.9ms | 35.8ms | 81 |
getMultipleAccounts | 4.0ms | 20.6ms | 8.7ms | 41.0ms | 81 |
getTransaction (recent) | 3.3ms | 29.8ms | 8.1ms | 50.1ms | 81 |
getTransaction (archival) | 21.5ms | 20.8ms | 53.8ms | 41.1ms | 75 |
getTokenAccountsByOwner | 16.7ms | 76.0ms | 48.1ms | 235ms | 82 |
getSignaturesForAddress | 24.0ms | 156ms | 80.0ms | 241ms | 82 |
getProgramAccounts | 27.0ms | 102ms | 108ms | 500ms | 82 |
getBlock (recent) | 28.6ms | 271ms | 78.0ms | 471ms | 82 |
getBlock (archival) | 167ms | 357ms | 296ms | 558ms | 82 |
getHealth, getSlot, and getLatestBlockhash returned in 1.9ms at the median location for Quicknode and about 20ms for Alchemy. Account reads follow the same pattern: getAccountInfo at 3.3ms vs 20.5ms and getMultipleAccounts at 4.0ms vs 20.6ms. These are the calls a trading bot or wallet makes most often, so small per-call differences add up across a session.
Recent getBlock came back in 28.6ms vs 271ms at P50. getSignaturesForAddress ran 24.0ms vs 156ms, getTokenAccountsByOwner 16.7ms vs 76.0ms, and getProgramAccounts 27.0ms vs 102ms. The tails spread further: getProgramAccounts P95 was 108ms for Quicknode and 500ms for Alchemy.
This is the one method where the two providers are near even. In the 24 hours to October 5, 2026, Alchemy's median P50 was 20.8ms against Quicknode's 21.5ms, and its P95 was 41.1ms against 53.8ms. Across all time windows, Quicknode was faster in 68.5% of comparisons on this method and on 75 of 82 days, so the lead holds over time but it is narrow. Teams with archival-heavy workloads should test this call on their own traffic.
A single snapshot can reflect one good hour. The more useful question is whether the result repeats. It does: Quicknode led the daily comparison on all 82 scored days, and its daily share never fell below 73%.
On 11 of the 12 methods, Quicknode was faster on at least 81 of the 82 days.
Across the eight time windows, from 30 minutes to one year, Quicknode's share ranged from 88.6% to 94.3%.
Quicknode was faster in at least 71% of comparisons at every one of the 29 locations. The closest were Teraswitch Amsterdam (71.4%), Teraswitch Pittsburgh (72.7%), and Latitude Singapore (73.4%).
Broken down by method and location, Quicknode won every one of the 32 comparisons in 208 of 348 pairs. Alchemy led in 11, and 9 of those were archival getTransaction. Here is every location:
The result also holds on every infrastructure type the monitor uses:
Infrastructure | Comparisons | Quicknode faster |
|---|---|---|
AWS | 2,688 | 94.2% |
Google Cloud | 2,688 | 92.6% |
Latitude | 2,688 | 91.0% |
Teraswitch | 3,072 | 88.4% |
Latency only matters if the request succeeds. In the 24 hours to October 5, 2026, the median error rate was 0% for both providers on 10 of the 12 methods. Alchemy's median was 0.34% on getProgramAccounts, and Quicknode's was 0.52% on getSignaturesForAddress.
Writes. The Foundation monitor measures reads only, so sendTransaction and transaction landing are out of scope.
Pricing, rate limits, and enhanced APIs. Those are covered in our Infura vs Alchemy vs Quicknode comparison.
Other providers. The Foundation monitor tracks six Solana providers and this post compares two. The full field is on the Solana Foundation RPC dashboard.
Public data narrows the shortlist. Your own traffic makes the decision. Point the methods your app calls most at both providers, from the region closest to your users, and compare P95 rather than averages. See Solana RPC on Quicknode for endpoints, supported methods, and archive access.
On the Solana Foundation's public RPC monitor, Quicknode led the daily latency comparison on all 82 scored days from July 16 through October 5, 2026. Across 11,136 head-to-head measurements covering 12 read methods and 29 locations, Quicknode recorded lower latency in 91.4%.
Heavy reads. In the 24 hours to October 5, 2026, recent getBlock P95 was 78.0ms on Quicknode vs 471ms on Alchemy, and getProgramAccounts P95 was 108ms vs 500ms. Light calls such as getSlot ran about 11x faster on Quicknode.
Archival getTransaction. Alchemy posted lower 24-hour medians on that method (20.8ms vs 21.5ms at P50), though Quicknode was faster on 75 of 82 days.
We pulled every location, method, and time window from the Solana Foundation's public RPC monitor on October 5, 2026 and compared the two providers' reported average, P50, P95, and P99 latency for each combination, 11,136 measurements in all. Each day was also scored on its own across up to 1,392 measurements.
P95 is the response time that 95% of requests come in under. It captures the slow tail that an average hides, which is what a trading bot or wallet feels during busy periods. A provider with a low average but a high P95 still causes timeouts.
The Solana Foundation runs it as an independent measurement party for its VIP Trading Program. The Foundation does not operate RPC infrastructure, the runner is open source under Apache 2.0, and results are published at solana.com/data.
Ready to build on Solana? Create a free Quicknode account or see Solana RPC on Quicknode.
Founded in 2017, Quicknode deploys institutional-grade blockchain infrastructure for developers and enterprises. With 99.99% uptime and support for 75+ chains, teams build and scale onchain applications without compromise.
The latest engineering insights, product updates, and web3 news delivered straight to your inbox.
🤖 NEW: Quicknode AI Cookbook 10 tested recipes for AI coding agents.
See recipes    Â