The RPC layer
your prod stack
actually deserves.

One endpoint, every chain we run. Caching, failover and per-tier rate limiting are the product — not an add-on you discover after your first outage.

—ms
Median latency
—
Requests · last 24h
—
Networks live
Recently shipped →
Aug 10shippedConfig regeneration hardened — no more deploy-time drift·
Jul 04shippedPublic chain dashboards with a live request map·
Jun 14shippedGateway config import / export as portable bundles·
Jun 14shippedIn-dashboard alert manager — channels + per-alert routing·
Jun 02shippedPer-key origin allowlists across every surface·
May 23shippedTron gateway — full, solidity and JSON-RPC surfaces·
May 20shippedMulti-upstream load balancing + gateway monitor·
Aug 10shippedConfig regeneration hardened — no more deploy-time drift·
Jul 04shippedPublic chain dashboards with a live request map·
Jun 14shippedGateway config import / export as portable bundles·
Jun 14shippedIn-dashboard alert manager — channels + per-alert routing·
Jun 02shippedPer-key origin allowlists across every surface·
May 23shippedTron gateway — full, solidity and JSON-RPC surfaces·
May 20shippedMulti-upstream load balancing + gateway monitor·
Requests · 24hlive
—
Requests / sec
—
5-minute rate, all chains
Median latency
—
p50 across upstreams
Lifetime requests
—
Why Hermod

Built for the boring part of going to production.

No surprise rate limits at 3am, no archive paywall, no per-method gotchas. The whole gateway is sized for one thing: keeping your reads fast and your writes deterministic.

Click any card → preview
Quickstart

No SDK. Just a URL your client already understands.

The gateway speaks plain JSON-RPC, so whatever library you already use keeps working. Swap the transport URL, keep the code.

Wire protocols
JSON-RPC · REST · WebSocket
Auth
Bearer API keys · origin allowlist
Clients
viem · ethers · web3.py · go-ethereum
Payment
Free tier · credits · x402
# Drop-in JSON-RPC — the public endpoint needs no key
curl https://infra.originstake.com/monad/evm \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc":"2.0",
    "id":1,
    "method":"eth_getBlockByNumber",
    "params":["latest", false]
  }'
Use cases

One stack, every shape of Web3 team.

The same gateway backs exchanges that can't drop a trade, indexers that can't miss a log, and agents that pay per call.

DeFi
DeFi & exchanges
Multi-upstream reads that fail over on their own, and trace access for vaults and liquidators on the chains that serve it.
Wallets
Wallets & on-ramps
Origin-pinned keys for browser bundles, batch requests, and one key across EVM, Tron and Solana.
Games
Web3 gaming
High sustained throughput and websockets that replay missed events after mobile network churn.
NFT
NFT marketplaces
Cached finalized reads, wide-enough log queries, and per-key usage so you can see what each service costs.
Indexers
Indexers & data
eth_getLogs, debug_* and trace_* routed to nodes that serve them, with finalized history served from cache.
Compliance
Compliance & RWA
Consistent multi-chain reads on a single key, with per-key and per-method usage records.
Bridges
Bridges & cross-chain
Several chains on one key and one auth shape, each backed by more than one upstream node.
AI agents
Autonomous agents
x402 pay-per-call with no signup, and an MCP server whose spending is capped per process.
From the engineering team

What we’re learning, written down.

Notes on the chains we onboard, the infra we rebuild, and the post-mortems we wish we'd written sooner.

EngEngineering · Jul 04, 2026Building a live request map without leaking a single IP
NetNetworks · May 23, 2026Tron RPC — three surfaces, one gateway
RelReliability · May 20, 2026Why we stopped circuit-breaking the hot path
ObsObservability · May 01, 2026A status code is not a diagnosis
Ready when you are

Ship to mainnet today.

Point your client at the public endpoint and start reading — no card, no call. Add a key when you need higher limits, per-key usage, and origin pinning.