Services
Availability is written next to every heading on this page. We would rather lose a deal on the first call than on the third invoice.
The only service we can start this week, because the capability already exists inside Phantrium. Everything else on this page is being built up to that standard.
A review that produces a hundred findings and no priority is a document, not a defence. Ours ends with the three things to do this month.
One endpoint per chain, with metering that reflects what a call actually costs us to serve. Live now on four chains and free while the beta runs — you can create a key yourself and be calling in under a minute. We will tell you before we charge anyone.
Most providers bill "requests". An eth_getLogs across a wide range is not the same work as an eth_blockNumber, and pricing that pretends otherwise ends in a surprise invoice.
At launch we route to vetted upstream providers behind our own gateway, with failover and a block-height check so a lagging backend never serves you. As our own nodes come online, traffic moves to them chain by chain. You keep the same endpoint and the same key throughout — the change is ours to manage, not yours to migrate.
We would rather say this out loud than let you discover it. It is also why our pricing does not undercut the market on day one.
A full node on dedicated hardware, run as infrastructure rather than as a hobby. Provisioned per engagement, because the machine is bought for you.
| What we run | How it is kept alive | What you can see |
|---|---|---|
| Execution client | Dedicated NVMe, automatic restart, disk-headroom alerting before it bites | Block height, peer count, sync state |
| Consensus client | Separated from the execution client so one failure is not both | Head slot, participation, lag against the network |
| Restore path | Snapshot restore rehearsed before production, not during an incident | Date of the last successful rehearsal |
| Upgrades | Client releases tracked; hard forks scheduled with you, in writing | Planned maintenance on the status page |
One node is not redundancy. Until a second machine exists for your chain, we say so in the contract and keep a fallback endpoint in place.
This is the one service where our mistake costs your money instead of your uptime. So it opens last, and only with evidence.
A full month of validator operation with no missed duties, measured by the same monitoring that feeds the public status page.
Mainnet with our stake before anyone else's, so the first slashing risk we take is our own.
Signing keys separated from the node, a written 03:00 procedure someone else can execute, and a published incident history.
If you need staking today, we will say so and point you elsewhere. We would rather be the provider you call next year than the one you write about this year.
Oracle operation — running a node that feeds price data to a live decentralised feed — is something we are working toward, not something we sell. It requires admission to a network, which is earned over years and cannot be bought or built. When that changes, it gets its own section here with the confirmation to back it.