Inference DA

A key technology that makes OpenGradient's inference capabilities end-to-end verifiable and secure is Inference Data Availability (Inference DA). As inference verification explains, we use configurable security techniques for securing inference, ranging from ZKML to TEEs. Using our in-house Inference DA technology, we embed these cryptographic proofs or attestations into OpenGradient's native blocks - making inferences end-to-end verifiable and reproducible by all validator nodes on the network.

By including EVM transactions and their associated inference proofs in every OpenGradient block, we make transactions that use AI models end-to-end verifiable by our entire network, resulting in extremely strong security and consensus. This means that all inferences are fully verified and secured by every single full node on our network, which has the same level of security as regular EVM state transitions.

NOTE

Inference proof verification is part of our state transition function, ensuring end-to-end verifiability and consensus on transaction and inference execution.

Individual users and light clients can also use our Inference DA to verify specific inferences they are interested in, allowing off-chain and cross-chain users to gain full confidence in their results.