Prismrelay
About Prismrelay

Useful work.
Open infrastructure.

Prismrelay explores a compute network built around independent hardware, verifiable work, and shared economics.

From request to result

A request needs an available machine with the right model and memory. Routing considers capability, queue depth, and latency. Processing produces an output; verification determines whether it meets the job requirements. A receipt can make the worker and the work easier to examine. A receipt alone does not prove the answer is correct.

Capacity is a starting point

GPU ownership creates potential supply. Useful demand, reliable execution, and verification turn that potential into a service. The setup and network explorers let you change inputs and inspect these relationships without claiming live throughput or earnings.

How the proposed economy fits

The concept connects provider compensation to useful work. The network-margin rule shown here allocates half to stakers and half to token supply reduction. It is a mechanism to examine, not a record of payouts or an assurance of returns. Prismrelay uses $PRLY as its unissued preset identity; no token contract or purchase is offered.

Explore the split

Prompts and permissions

Your request drafts stay in this browser. Submitting a draft here does not send it to an inference provider. Wallet access begins only when you choose a wallet and approve its account request. The site reads your account and current network, with no balance lookup, signing, asset approval, or transaction.

Questions worth asking

What can I do without connecting?

Explore routing, adjust capacity inputs, prepare request drafts, and read the mechanism. SKIP closes the wallet selector, and you can connect later.

Which network does the wallet use?

Robinhood Chain mainnet is the target. The address menu shows the network actually returned by your wallet. A network change requires your explicit choice; a wallet that does not support it keeps its real current-network label.

Does a saved draft run inference?

No. It is stored locally so you can revise the request. Starting a worker or sending a workload reaches the availability boundary with no charge and no claim of completed compute.

Does the revenue graphic report performance?

No. It calculates a proposed allocation from your input. Its values are not live revenue, payouts, a token burn, or a forecast.