RPC Plane
Install
Solana · Mainnet · Devnet · Testnet Bring your own providers

Built for teams that can't
afford an RPC outage.

RPC Plane sits in front of the Solana providers you already pay for and makes them reliable together — routing, failover, and slot-aware health scoring with your keys, on your machine. Here's what that looks like for the four teams that lean on it hardest.

Trading & MEV Wallet & dApp backends Indexers & analytics Infra teams
Trading & MEV

A transaction that lands
one slot late is a missed fill.

At ~400ms slot times, an RPC blip during a volatile minute is real PnL. A stale getAccountInfo misprices a trade; a sendTransaction that returns OK but never reaches a leader is a fill you didn't get. Single-provider setups have no recourse when it happens.

  • →Write broadcast. sendTransaction fans out to every healthy provider in parallel — the transaction reaches a leader through whichever path is fastest, maximizing landing probability.
  • →Slot-drift demotion. A provider 14 slots behind can still return HTTP 200. RPC Plane sees the drift and stops quoting off it before your strategy does.
  • →parallel_race reads. Send to every healthy provider and take the fastest valid success — lowest achievable read latency when it matters most.
  • →Sub-millisecond overhead. Health scores are pre-computed; there is no extra I/O in the hot path.
Binary features → Routing strategies →
Wallet & dApp backends

One provider outage,
every user's app looks broken.

The balance APIs, transaction-history services, and notification workers behind your wallet or dApp all read from Solana RPC. When your primary provider degrades, it isn't one screen that breaks — it's every user at once, a support queue that floods, and trust that erodes on public timelines.

  • →Automatic failover. A per-provider circuit breaker opens on failure in under a second, probes for recovery, and resumes traffic on its own. A provider outage becomes a non-event your users never see.
  • →Health-scored routing. Every read goes to the healthiest provider, ranked live on latency, error rate, and slot freshness — so a degrading provider is demoted before it shows a wrong balance.
  • →One URL change. Point your client at http://localhost:9400. solana-web3.js, Anchor, and mobile SDKs work unchanged.
  • →No data custody. The proxy runs on your infrastructure with your keys. Nothing about your users leaves your machines.

RPC Plane proxies HTTP JSON-RPC today. WebSocket subscriptions (accountSubscribe, logsSubscribe) are on the roadmap — until then, notification workers can subscribe to a provider directly while their reads and writes route through the plane.

Self-hosted binary → Quick start →
Indexers & analytics

Your dataset is only as correct
as the RPC behind it.

A provider that lags 20,000 blocks behind tip, or silently returns incomplete historical data, quietly corrupts everything downstream — and the query still returned HTTP 200. Backfills and streaming ingest need correctness, not just uptime, and per-call billing on getTransaction and getBlock adds up fast.

  • →Slot-aware health scoring. Catches the stale-but-200 case — a provider behind tip is deprioritized before it feeds your indexer bad data.
  • →Failover that keeps ingest flowing. When one provider fails, the pipeline reroutes automatically instead of stalling or dropping blocks.
  • →Cross-provider validation coming soon Detect when providers disagree on the same request and flag the one returning divergent state.
  • →Cost analytics (dashboard). Track credit burn per method and provider so you can cut spend without cutting reliability.
Binary features → Compare provider rates →
Infra & platform teams

You run RPC for everyone else.
Don't build the failover layer too.

Provider contracts, rate limits, and outages all land on your desk. Rolling your own retry-and-failover script means a stale-data detector, slot tracking, circuit breakers, and metrics you have to maintain forever. RPC Plane is that layer, already built and shipped as one binary.

  • →Zero infrastructure. A single statically linked binary — no Postgres, no Redis, no queues. Drop it in a sidecar container next to each service.
  • →weighted_random load spread. Distribute traffic across keys and providers so no single key trips its rate limit — stack free tiers into production-grade redundancy.
  • →Prometheus metrics + hot reload. Scrape :9401/metrics straight into Grafana; retune routing or add a provider with no restart.
  • →Fleet-wide visibility when you want it. The hosted dashboard adds provider health history, cost analytics, and alerts on top of the same engine — free during open beta.
Self-hosted binary → Cloud dashboard →
Works with what you have

Any HTTP JSON-RPC Solana endpoint.

Helius, QuickNode, Triton, Alchemy, Chainstack, the public mainnet endpoint, and your own validators. You bring the keys; RPC Plane makes them reliable together. No matter the workload above, the integration is the same one URL change.

One URL change.
Every workload gets more reliable.

Free, source-available, drop-in. Point your app at the proxy and keep the providers you already pay for.

Quick start Explore the binary