Toric

What is Toric and what is it for?

Toric is a Holochain app for judging the integrity of AI models: instead of one lab, host or leaderboard deciding which model record can be trusted, a network of independent participants checks it and puts its own standing behind the answer. Each check leaves a trust score that is kept permanently and can be traced back through the validators who produced it.

Toric is a Holochain app for judging the integrity of AI models. Its maintainers describe the aim plainly: rather than taking a model's published description on faith because a single publisher or leaderboard says so, a network of independent participants checks the record and backs its verdict with its own standing. What is being checked is a model's manifest — the claims it makes about itself — and the verdict is kept where others can inspect it.

Validation works as a commit–reveal round. A validator first commits an assessment without showing it, then reveals it, and the revealed assessments are counted against a quorum; how much a validator's assessment weighs depends on its record of earlier validation work, so a participant with a demonstrated history counts for more than a newcomer. The result is a trust score that is stored permanently on the network and can be traced back through the participants who produced it.

A concrete example of how the project is put together is the validator client, which connects directly to its own local Holochain conductor with no intermediary. Underneath, the app is assembled from four separate Holochain networks handling the model registry, the validation protocol, participant identity and internal credit accounting, alongside an API for the interface and a desktop wrapper. The repository README is the place to look for setup steps and the component list.

From 2 public sources, checked 23 September 2026.