Weight on Gonka is the output of a participant's proof-of-compute performance — how much verified GPU capacity it proved it controls in the most recently completed cycle. Weight is what everything else on the network is proportional to: consensus voting power, the amount of inference work routed to a participant, and how model assignments get made for the incoming epoch.

Confusingly, Gonka actually tracks weight in two places that can legitimately diverge mid-epoch:

• Staking-module power (consensus power) — what SetComputeValidators writes. It drives which validators produce blocks and how governance votes are weighted.
• Epoch-group power (internal network power) — lives in the inference module's epoch groups. It governs how much weight a participant carries when validating peers' proofs, how inference requests are allocated, and which models it's assigned to serve.

The two are synchronized only at epoch boundaries, when a successful participant's epoch-group results become the next staking-module power table. If two dashboards show different "weight" for the same participant, they are very likely reading from different systems — check which one before assuming either is wrong.

GNKScan's participants page shows weight from the most recently completed cycle, sorted highest to lowest. For the full mechanics of how weight is computed and reconciled, see How Gonka's Proof of Compute Works.