root
[home][gpus][pilot][tiers][api keys][docs][about]
ca [copy]
[pump.fun] [connect wallet]

Root on a real GPU, funded by wallet inflows

Observed SOL inflows fund a cost-capped Vast pilot.
Public capacity appears only when Vast confirms a live machine; keys can join its controlled session queue.

[get an api key] [how it works] [see the fleet]
treasury: SOL inflows: SOL gpus online: gpu hours given: free sessions:
live data: connecting…
provider telemetry unavailable
[see provider fleet]

the ledger

on-chain wallet inflows · provider data when connected
activity
gpu hours
sessions
wallet inflows
gpu cost
wallet inflows (SOL) · historical on-chain transfers

how it works

01
Every trade pays a fee
$ROOT earns a creator fee on every trade, forever. Those fees do one thing: they pay the hourly bill on the machines in the fleet.
02
Holding is the subscription
Hold the tokens for a tier and your key works. Sell below it and it stops on the next check. Nothing to top up and no card on file.
03
Then it is a real machine
Not an endpoint with a model behind it. SSH, root, your own filesystem. Install what you like and run whatever you want on it.

tiers

checked live against your wallet. free needs nothing at all.
Live provider / Vast.ai

GPU capacity, without the guesswork.

Only machines confirmed by the provider appear here. Rates, capacity, and runway are calculated from fresh telemetry; marketplace listings never count as a running fleet.

Confirmed instances

Tier hardware targets

configured access goals · live capacity shown above
These are configured tier targets, not rented machines. Live capacity appears above only after Vast confirms a verified node.
Controlled spend / one machine

A real GPU, funded by observed inflows.

The pilot is deliberately small: one verified on-demand offer, a hard $0.50/hour ceiling, and an operator confirmation before any paid launch.

public pilot state

loading pilot state…

The public site can show a Vast-confirmed GPU and its real hourly rate. It never publishes SSH hostnames, ports, or provider credentials.

checking session readiness…

operator controls

Paste the private operator token only in your own browser session to inspect offers, launch one approved pilot, or stop the tracked instance.

operator controls are locked
A launch always re-checks the live Vast offer and refuses any price above $0.50/hour. Stopping ends GPU runtime; check Vast for any separate storage terms.
Wallet access / live balance

Your balance decides your runway.

There are no subscriptions to manage. A wallet’s current $ROOT balance determines the access tier checked by every key and session request.

Access ladder

tier rules from the live application

Full tier detail

balance is read from chain, not from an account you make here

The tier is read from the wallet at the time a key or pilot-session request is checked. Queue placement is not an active GPU session and does not expose provider access.

Buy more and your next balance check can change the tier. A future session relay will re-check the balance before activating a controlled session.

Wallet-gated API access

Bring your wallet. Leave the account form.

Your key is derived from a live on-chain balance, bound to one wallet, and rechecked before a session request. It is not a promise of a GPU—only a safe way to enter the queue.

Get your wallet key

Paste a wallet address or connect a Solana wallet. We read its $ROOT balance on chain, work out the tier, and hand you a key. The key stops working when the balance drops below the line.

How access works

01 · Verify
We check the wallet balance directly on chain.
02 · Issue
Your key is tied to that wallet, not a shared account.
03 · Queue
Requests wait for a confirmed pilot and fresh capacity.
Read the contract

Small promises. Real machines.

ROOT connects observed wallet inflows to a tightly controlled GPU pilot. The rules below describe what is live today and what is deliberately not promised.

What this is

A fleet of rented GPUs paid for by observed wallet inflows on a token. The first Vast pilot is operator-confirmed and cost-capped; public capacity appears only after Vast reports a live instance.

It is not a model API. When a session relay is activated, it will provide a container with root and a GPU. Until then, keys can join the controlled pilot queue but do not expose machine access.

Getting a session

# request a session after receiving a key
curl -X POST https://root.rent/api/session \
  -H "content-type: application/json" \
  -d '{"key":"root_xxxxxxxxxxxxxxxx"}'

A request is bound to its key and checked against the wallet’s live token balance. It receives a queue or readiness status only; SSH credentials are not published by this site.

Checking your tier from code

curl -s https://root.rent/api/tier?wallet=$WALLET

{ "balance": 128400,
  "tier": { "name": "taproot", "hours": 4, "gpu": "A100 80GB" },
  "next": { "name": "heartwood", "hold": 1000000, "need": 871600 } }

What you get

  • an operator-confirmed pilot only after Vast reports the instance live
  • a wallet- and tier-gated session request with a daily time allowance
  • no provider credential, direct SSH host, or port exposed in the browser
  • real provider rate and fleet state while the pilot is running
  • no automatic treasury transfers, credit top-ups, or unattended GPU launches

Limits, honestly

The pilot currently supports one explicitly approved Vast instance, capped at $0.50/hour. Session relay activation remains operator-controlled, so a queue entry is not a promise of immediate shell access.

Fleet size should follow revenue. The dashboard labels observed wallet inflows rather than guaranteed creator fees, and it counts cost only when Vast confirms a running instance.

Where the numbers come from

  • treasury — live SOL balance of the fee wallet, read from chain
  • wallet inflows — inbound SOL transfers observed on the configured dev wallet
  • gpu cost — the actual hourly rate of every machine currently up, summed
  • runway — treasury divided by that daily burn
  • tier — your token balance, checked with getTokenAccountsByOwner

The honest part

The treasury is an operator wallet. We pay the GPU bills from it and the balance, the claims and the spend are all on this page so you can check the arithmetic. It is transparent, not trustless, and it would be dishonest to describe it any other way.

Not investment advice.

About root

Infrastructure with receipts.

A creator-fee-funded experiment where the ledger, provider state, and access rules stay visible.

Why

A GPU is the one thing in this industry that everybody needs and almost nobody can casually afford. An H100 is somewhere around two and a half dollars an hour, which is nothing to a company and a real decision for a student, a hobbyist, or anyone trying something for the first time.

Meanwhile a token on pump.fun throws off a creator fee on every single trade, forever, and in almost every case that money goes into a wallet and stays there.

So point it at the bill. Traders pay the fee because they are trading. The fee rents machines. The machines are free to use. Nobody is ever asked to pay for the thing they are using, which is the only reason the word free is honest here.

Why free at all

Because a fleet that only serves holders is just a discount, and a discount is not a story. Somebody opening their first shell on an H100 without a credit card is a story, and the ledger above proves it happened.

What it costs to run

The currently confirmed provider nodes cost about a day. Every hour those cards are up shows on the ledger whether anybody used them or not, because hiding idle cost would make the runway number a lie.

root · gpu hours paid for by trading fees [docs] [about] [pump.fun]