Initial run - a single 1-hour window on mainnet, published raw and unedited. Continuous, multi-route benchmarks follow. The point of publishing early is the method: match by signature, separate ACK from first-seen, show the tail.

The experiment in one screen

FieldValue
MethodQUIC transaction relay
Window1 hour
Captured2026-06-30 · 09:14 UTC
RouteYellowstone TPU · Frankfurt
Submissions2,202
Observed2,121 (96.32%)
Deshred p50 / p90 / p99104 / 375 / 838 ms
Route ACK p500.01 ms

Methodology note: ACK latency is gateway health, not landing latency. Landing evidence uses deshred and processed observations matched by signature - so the number that matters is submit → first-seen, not the gateway ACK.

Full tables: benchmark report.

Percentiles (submit path)

PercentileClient ACK (ms)Route ACK (ms)Deshred (ms)Processed (ms)
p500.310.01104106
p750.370.01199200
p900.480.02375375
p950.580.02511511
p990.790.03838839
max1.620.1145,41245,417

Tail latency is real and shown, not hidden.

By leader region (deshred)

RegionSubmittedObservedRatep50 (ms)p90 (ms)p99 (ms)
Europe1,3111,26096.1%702601,022
Asia Pacific27124891.5%265650788
US East19419198.5%159360782
US Central14013898.6%111178413
US West4646100%167443467
Unknown24023899.2%109199614

Fastest leaders sat where the physics is best: Frankfurt validators landed at 20-48 ms p50, with Amsterdam and Stockholm close behind.

How to read without over-claiming

  1. Do not quote route ACK as landing latency.
  2. Read observation rate and tails together - a clean p50 with a 45 s max is still a real path.
  3. Region tables are leader geography, not client geography.
  4. One hour is the first public window, not a continuous SLA board.

Harness: solana-tx-bench.

Next

Open the full report and compare with later leader-paced A/B runs on /benchmarks.

To send on the co-located path in production, get started or start self-serve on app..

Frequently asked questions

Is gateway ACK the same as landing latency?
No. ACK latency is gateway health. Landing evidence uses deshred and processed observations matched by signature - submit to first-seen is the number that matters, not the gateway ACK.
Why publish tail latency?
Tail is real. Hiding p99 and worst cases is marketing. This run shows p99 deshred about 838 ms and a worst single transaction around 45 seconds - raw and unedited.
Is this a continuous multi-route board?
No. It is the first published 1-hour window. Continuous multi-route benchmarks are the next step; this report establishes the method and the first public numbers.