Model Registration (One-Time)

Before an AI model can be used within the Hubic network, it must be registered through a one-time process that cryptographically anchors its identity, verification logic, and metadata to Ethereum smart contracts.

This ensures:

  • Immutable model identity

  • Proof compatibility enforcement

  • Optional RWA token linkage


🧾 Registration Workflow:

  1. Model developer compiles the model into a zero-knowledge circuit.

  2. A unique hash (e.g. SHA-256 of the zk-circuit) is generated.

  3. Metadata is prepared (e.g. zk format, version, license terms).

  4. Developer submits to the zkRegistry via ModelSubmission contract.


📦 On-Chain Structure (Simplified):

struct ModelSubmission {
  bytes32 model_hash;        // Unique zk-circuit hash
  address verifier_contract; // Ethereum contract used for zk verification
  address owner;             // Wallet entitled to inference royalties
  uint64 timestamp;          // Registration time
  string metadata_uri;       // Off-chain descriptor (IPFS, Arweave)
}

🧠 Off-Chain Metadata Example:

{
  "name": "RWA-Yield-Optimizer-v1",
  "zk_format": "Halo2",
  "input_spec": "[APR, stake_volume]",
  "output_spec": "[reallocation_target]",
  "accuracy": 0.921,
  "rwa_token": "0xTokenizedModelERC20",
  "description": "Optimizes validator restaking using zkML."
}

🌍 RWA Integration:

  • Ownership Tokenization: The owner address can be a smart contract that manages revenue splits to ERC-20 or NFT holders.

  • Governance-Scoped Models: DAOs can submit models and bind them to governance logic or yield policy.

  • Usage Metering: All inference calls are logged and traceable for performance-based monetization.

Once registered, the model becomes a verifiable, composable, and revenue-capable digital asset — optionally wrapped into an RWA token.

Last updated