NeuroShardRun it. Verify it. Improve it.

One public project for native consensus, verified neural training, and an open ledger.

Locally controlled identity
Generate keys on your machine. Follow the chain and contribute computation without website registration.
Explicit execution rules
The LLM testnet trains an adapter on a frozen pretrained model and verifies paid inference through full replay. Its source and numerical profile are pinned by genesis.
Measured progress
Inspect accepted updates, reproduce experiments, and help resolve verification costs, independent ownership, and sustained operation.

This is an experimental native testnet. The initial validators are operated by one organization across two machines. Test balances have no redemption promise. Consult the model card, release notes, and public repository before participating.

Continual model evolution documents the separate full-model training, data, growth and native-dispute experiments. It includes measured failures and reproduction instructions; these experiments have not replaced the public testnet's model.

NeuroShardRun it. Verify it. Improve it.

Locally controlled identity

Explicit execution rules

Measured progress