> For the complete documentation index, see [llms.txt](https://docs.verio.network/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.verio.network/verio.md).

# Verio

A public evidence layer for model provenance, execution verification, and attribution.

Verio is a public evidence layer for AI. It works on top of the model repositories and inference providers people already use, and turns the claims behind an AI output (which model version ran, what data it learned from, whether its reported results reproduce) into evidence that anyone can check.

Today those claims are declarative. A provider says which model it served, a model page reports a benchmark score, a derivative declares its ancestry. Existing infrastructure records these statements but offers no independent way to verify them. As AI outputs begin to drive payments, agent actions, and regulated decisions, that is no longer enough.

## How Verio works

Three primitives give every claim an identity, a record, and a lineage:

* [**Artifact Passports**](/protocol/artifact-passports.md) give models, datasets, adapters, evaluation suites, and execution profiles immutable, content-committed identities with explicit lineage.
* [**Inference Receipts**](/protocol/inference-receipts.md) bind each output to an artifact version, a canonical execution profile, and an accountable provider.
* The [**Provenance Graph**](/protocol/provenance-graph.md) connects derivatives to the upstream models and training data they depend on.

Any provider can become verifiable without migrating models or workloads: it emits receipts and accepts replay. Model weights, datasets, and computation stay off-chain; the chain holds only commitments, identities, collateral, disputes, and settlement.

## How the network is secured

Four mechanisms secure the network: canonical execution profiles that make replay deterministic, batch-anchored receipts, randomized audits seeded by public randomness, and bonded challenges with slashing. They combine into four cumulative [verification classes](/protocol/verification-classes.md), from **signed** to **audited**, **replayed**, and **redundant**, so applications buy assurance in proportion to the value at risk. Audited receipts add roughly one percent to the cost of a request.

## $VERIO

[$VERIO](/token/overview.md) is the unit of settlement and security. Applications pay for execution and verification in $VERIO; 90% of each fee flows along the provenance graph to the providers, verifiers, and upstream model and data contributors that did the work; and providers and verifiers stake $VERIO as slashable collateral.

## Where to start

<table data-view="cards"><thead><tr><th></th><th></th></tr></thead><tbody><tr><td><a href="/vision.md"><strong>Vision</strong></a></td><td>Why AI needs evidence, not more documentation.</td></tr><tr><td><a href="/protocol/architecture.md"><strong>Protocol</strong></a></td><td>Passports, receipts, the provenance graph, and verification classes.</td></tr><tr><td><a href="/network/request-lifecycle.md"><strong>Request lifecycle</strong></a></td><td>How a request moves from submission through audit to settlement.</td></tr><tr><td><a href="/token/overview.md"><strong>$VERIO</strong></a></td><td>Utility, staking, and fee flows.</td></tr><tr><td><a href="/roadmap.md"><strong>Roadmap</strong></a></td><td>Current progress and the development phases ahead.</td></tr><tr><td><a href="/whitepaper.md"><strong>Whitepaper</strong></a></td><td>The full design, with sources.</td></tr></tbody></table>

## Join our community

* Official X: <https://x.com/verio_network>
* Application: [https://verio.network](https://verio.network/)
* Community Discord: <https://discord.gg/verio>