Expert Bounties

How the reward works

Each model has a $HERO target pool split every epoch across the nodes serving it, weighted by what your node proves: committed memory × health × uptime, staked $HERO in at its square root so hardware stays the point. Target pools are funded by a launch bootstrap plus a share of what that model earns in paid inference, so payouts scale with real usage rather than a promise. Early operators of a model earn a multiplier that decays. Correctness is checked by fetching random expert blocks and verifying them byte-for-byte, so there's nothing to fake. Track it on the yield page.