QMATRIX
DocsFollow on X
litepaper · v0.1

QMATRIX protocol docs

on-chain specialist-attestation for scientific training data

§ Introduction

Introduction

QMATRIX is the on-chain specialist-attestation layer for scientific training data. It is a decentralized marketplace where domain-expert labelers, including chemists, physicists, and materials scientists, produce and verify the molecular, materials, and optimization datasets that quantum-AI models train on. Launching on Virtuals (Base).

§ The problem

The problem

The frontier-AI bottleneck moved off GPUs and onto expert human data. The open web is exhausted, with under 5% of available text meeting frontier standards. Labs now pay domain experts $500+/hr versus roughly $15/hr for generic annotators.

The hardest class is scientific. Reliable molecular and materials labels need costly DFT calculations, hours of compute per molecule, and PhD-grade judgment that cannot be crowdsourced. A single curation campaign runs $200K to $1.5M.

Meta took a $14.3B stake in Scale AI. That is the strongest argument for a neutral, protocol-owned network rather than a closed vendor sitting between labs and the data they depend on.

§ How QMATRIX works

How QMATRIX works

  • Specialist-labeler marketplace. Experts earn $MATRIX to label scientific datasets, trainers buy access, the protocol mediates.
  • Domain-expert attestation. On-chain quality consensus that needs PhD-grade expertise in both directions, producer and verifier.
  • Scientific data-class focus. Molecular, materials, and optimization datasets.
  • A quality-attestation primitive. A standard schema for a data class that today moves through one-off bilateral contracts.
Anchor on the concrete data bottleneck, not quantum as buzz. The QML-needs-this-data thesis is forward-looking, not shipped revenue.
§ Where QMATRIX fits

Where QMATRIX fits

Centralized expert players proved the model but are off-chain black boxes with no token. Decentralized players target broad data. The one decentralized scientific play does compute, not attestation.

QMATRIX is the only one pairing domain-expert attestation with the scientific data class: the on-chain Mercor or Surge for quantum-AI data. We complement the decentralized-compute plays rather than compete with them.

§ $MATRIX

$MATRIX

  • Label and attest. Experts earn $MATRIX.
  • Consume. Trainers pay $MATRIX for verified-dataset access.
  • Quality consensus. Stake to attest; bad-faith attestations are slashed.
  • Govern. Direction of the network sits with token holders.

Launching on Virtuals (Base). No wallet connect. No tokenomics on this page.

$MATRIX is a utility token for the QMATRIX network. Nothing here is financial advice.

§ Roadmap

Roadmap

  • NowLaunch $MATRIX on Virtuals. Docs live.
  • First domainsMolecular and materials live, attestation schema in production.
  • Verified flowDatasets flow to trainers under on-chain quality consensus.
  • More domainsOptimization and adjacent scientific classes onboarded.
  • The standardQMATRIX as the default attestation layer for scientific training data.

no hard dates

§ FAQ

FAQ

What is QMATRIX?
The on-chain specialist-attestation layer for scientific training data. A marketplace where domain experts produce and verify molecular, materials, and optimization datasets for quantum-AI training.
Why can't this data be crowdsourced?
Reliable labels need DFT calculations and PhD-grade chemistry, physics, or materials judgment. Generic annotators cannot produce or verify it, and the cost of getting it wrong is a model that learns the wrong physics.
How is it different from generic labeling networks?
Two things: domain-expert attestation as the quality primitive, and a deliberate focus on the scientific data class rather than broad web text or image labels.
Is there a product yet?
Pre-launch. The protocol, schema, and token launch on Virtuals (Base). First domains are molecular and materials.
When does $MATRIX launch?
On Virtuals (Base). Follow the X account for the launch window.