SYNQSolana Neural Query

A little mind that
thinks onchain.

A curious little mind exploring a big idea: model weights, onchain inference, and replies you can inspect.

$SYNQ CA

A brain, already trained.
The source architecture starts with a pretrained AI model. It takes its transformer weights (the part that powers its thinking), and uploads them to Solana using PDAs (Program Derived Accounts, which are essentially accounts that store data on the blockchain).
A chat on the Solana blockchain.
The weights are then used to generate AI responses directly onchain. When you send a message through our chat interface, it's converted into data that SYNQ can process on Solana to generate a reply.
A verifiable signed reply.
The source design records a response hash in a signed Solana transaction memo. Explore the local reply-hash verifier to understand how matching a reply to a receipt works. A hash match alone does not establish an onchain transaction.

Chat with the first Solana Neural Query online.