Decentralized Validation Network for Trustless Model Inference

The First ZKP-Powered Decentralized AI Verification Network, Ensuring On-Chain Authenticity for Local Models

Technical Architecture

Technical Architecture

Technical Architecture

Core Value

Single‑transaction verification, pluggable into any DApp/DAO

CLI, SDK and browser extension for one‑click proof generation, submission, query

3-Step Integration

1. Import SDK: pip install proofnet-core
2. Attach Prover: @verifier(model_type="llama2")
3. Generate Proof: model.prove(input_data)

Developer Portal

Create a task

Core Value

Single‑transaction verification, pluggable into any DApp/DAO

CLI, SDK and browser extension for one‑click proof generation, submission, query

3-Step Integration

1. Import SDK: pip install proofnet-core
2. Attach Prover: @verifier(model_type="llama2")
3. Generate Proof: model.prove(input_data)

Developer Portal

Create a task

Core Value

Single‑transaction verification, pluggable into any DApp/DAO

CLI, SDK and browser extension for one‑click proof generation, submission, query

3-Step Integration

1. Import SDK: pip install proofnet-core
2. Attach Prover: @verifier(model_type="llama2")
3. Generate Proof: model.prove(input_data)

Developer Portal

Create a task

Unleash Full Potential

Seamless Integration with Leading Technologies

Effortlessly connect with a wide range of external tools and technologies. Cryptographic Trust for AI Outputs

Unleash Full Potential

Seamless Integration with Leading Technologies

Effortlessly connect with a wide range of external tools and technologies. Cryptographic Trust for AI Outputs

Unleash Full Potential

Seamless Integration with Leading Technologies

Effortlessly connect with a wide range of external tools and technologies. Cryptographic Trust for AI Outputs

FAQ

FAQ

FAQ

Popular Questions

Which model architectures are supported right now?

How long does proof generation take and how much VRAM is needed?

I prefer to keep my weights private; can I submit only a commitment?

What happens if a proof is invalid or a validator acts maliciously?

Do users need a GPU to participate?

How does the network prevent collusion among validators?

Which model architectures are supported right now?

How long does proof generation take and how much VRAM is needed?

I prefer to keep my weights private; can I submit only a commitment?

What happens if a proof is invalid or a validator acts maliciously?

Do users need a GPU to participate?

How does the network prevent collusion among validators?

Which model architectures are supported right now?

How long does proof generation take and how much VRAM is needed?

I prefer to keep my weights private; can I submit only a commitment?

What happens if a proof is invalid or a validator acts maliciously?

Do users need a GPU to participate?

How does the network prevent collusion among validators?

Ready to build a verifiable AI future together?

Velink allows any AI model to complete its inference locally while generating a tiny cryptographic proof for instantaneous verification by light nodes on the chain.