Read latency and burst reliability of the nodius Solana RPC versus the public Solana RPC, measured from the same vantage point against live production. No staging numbers, no cherry-picked windows.
50 concurrent workers each firing 10 getSlot calls (500 total). This is the test that decides whether your bot survives a volatility spike.
429 Too Many Requests. Nodius bills per call from a prepaid USDC balance, so a funded account's calls complete under burst instead of being throttled. For an autonomous strategy, that is the difference between executing and missing the trade.
Warm single-call latency, p50 / p99, from the same EU vantage. Lower is better. Both endpoints are fast on cheap reads; the difference is consistency under load (above).
| Method | Nodius p50 | Nodius p99 | Public RPC p50 | Public RPC p99 |
|---|---|---|---|---|
| getSlot | 17 ms | — | 14 ms | — |
| getHealth | 19 ms | — | 49 ms | 109 ms |
| getLatestBlockhash | 18 ms | — | 48 ms | 91 ms |
| getBalance | 16 ms | — | rate-limited during measurement | |
| getAccountInfo | 17 ms | — | rate-limited during measurement | |
Nodius latency is single-digit-to-low-tens of milliseconds from an EU vantage because requests terminate on bare metal in Frankfurt, not a distant shared pool. Where the public RPC could be measured gently (sequential, no concurrency), its warm latency was materially higher; under any concurrency it rate-limited before latency could be measured at all.
How these numbers were produced, so you can reproduce them.
· Target: live mainnet-beta production — https://rpc.nodius.xyz vs https://api.mainnet-beta.solana.com.
· Vantage: a single EU-based host, same network path for both endpoints, measured back-to-back in the same window.
· Latency: 3 warmup calls then 40 timed calls per method; p50/p99 of successful calls. Public-RPC latency was re-measured at a gentle sequential rate to avoid its rate limiter where possible.
· Burst: 50 concurrent worker threads × 10 getSlot calls each; success vs HTTP 429 counted across all 500 requests.
· Honesty: fields marked "rate-limited during measurement" are exactly that — the public endpoint throttled the measurement, which is itself the finding. No figures are interpolated or estimated.
· These are point-in-time measurements from one vantage. Latency varies with your location and load; the burst result is the structural difference, not a transient.
No signup, no card. A funded Solana wallet is the account — deposit USDC and make your first call in about a minute.
Get started