Leaderboard — cumulative across all measurements (on-chain confirmed)
Loading…
How we measure
Every result on this page comes from real mainnet transactions and is verifiable on-chain. No simulations, no testnet.
- Same transaction, raced. For each run we build identical transactions and send them to every endpoint at the same instant, so providers are compared on equal footing — the only variable is how fast each one gets the tx onto the chain.
- Shred-resolution landing detection. We don't poll an RPC and ask "did it land yet?" — we tap the validator Turbine stream directly and timestamp the moment the first shred carrying your transaction arrives. That's the earliest physically observable signal that a block producer included the tx, with single-shred timing precision rather than the hundreds-of-milliseconds blur of polling.
- Location-neutral latency. From the raw send→observe time we subtract network transit on both ends — a TCP-connect handshake to the endpoint on the send side, and the leader→our-node propagation on the receive side — so the reported milliseconds reflect the provider's relay + inclusion speed, not how far our measurement box happens to sit. RPC processing time stays in the number, because that's the provider's job.
- Slot distance. Alongside milliseconds we report how many slots after send the tx landed. Slots are integer and identical for everyone on the network, so this metric is immune to clock and distance effects — the most robust way to compare providers. We deliberately use shred arrival and slot distance rather than ticks: ticks are the usual unit for sub-block resolution, but they aren't tied to real time — they advance at each validator's own scheduler pace, so they aren't comparable across leaders.
- On-chain confirmed. Landing is cross-checked against the chain (signature status + confirmed slot); every confirmed tx is verifiable in any explorer. Latency percentiles shown are p50 / p95 / p99.
Measurement queue
Payment and execution status of recent measurement orders.
Loading…
Request a measurement
Benchmark your endpoint(s) against the field. A run is queued once payment is confirmed on-chain — price is a small service fee plus the real transaction fees the run will incur.
My measurements
Enter your access token to see your orders and per-slot landing results.
No token loaded.