onChain LLMs

Decentralized Large Language Models running directly on the blockchain, ensuring transparency, censorship resistance, and verifiable inference.

Key Features

Why run LLMs on-chain?

Cryptographic proofs ensure that the model output was generated by the specific model and input, preventing manipulation.

Distributed network of nodes executes model inference, eliminating single points of failure and censorship.

Seamlessly integrate AI logic directly into your smart contracts without relying on centralized oracles.

User or smart contract sends a prompt to the on-chain model registry.

Decentralized nodes pick up the task and run the model inference off-chain (or on-chain for smaller models).

ZK-proofs or optimistic challenges verify the correctness of the computation.

The verified result is returned to the blockchain, triggering downstream smart contract actions.

// Example Smart Contract Call

function askOracle(string memory prompt) public {

bytes32 requestId = onChainLLM.request(

"Llama-3-8b",

prompt

);

}