Usage you can check, and what isn't used burns

Every bell at 20:00 UTC counts what each seat's key spent on inference. What the seats did not spend buys $QUOTUM on the pair and burns it. Each row below is one bell, with its tx. No wallet needed. The same rows are in JSON.

bought back and burned· 
Spent on inference· 

Separately, the token's deployer burned 28,604,787 QUOTUM on Sep 26 at 12:11 UTC. It was not a bell and is not counted above. Tx 0xb7b3…110c

Each bell, the cap spent on inference bought back and burned no burn
··
··
··
··
··
··

Bell by bellreading the receipts

BellSeats
live, active
Spent on inferenceSpentRequests, tokens in and outBurnedTx
······
······
······
······
······
······

A seat is live if someone holds it at the bell, and active if its key spent more than $0 that session. Requests and tokens come from Venice's billing history and the calls that pass through quotum. A partial count shows only what we saw: a key copied into another app may be missed. A dot is a number we do not have, not a zero.

Check it yourselfthree steps, one bell

  1. 1

    The seat's receipt

    Open the session's receipts: one row a seat, with what its key spent at the bell. The row's hash is the keccak256 of that row's JSON.

    ·
  2. 2

    The hash on chain

    The keeper reports each receipt to the TaxSink contract: a Reported(session, seat, USD, receiptHash) event carries the same hash, before the bell.

    ·
  3. 3

    The bell and the burn

    The bell tx buys $QUOTUM with what the seats did not spend and burns it: the Bell event, the buy, and a transfer to 0x0. The token's total supply drops by the same amount.

    ·

JSON: /api/usage.json. The receipts: /api/receipts?state=1. How the cap and the bell work is in the docs.

PAYG, separately

Seat bell: contract-reported. PAYG: operator-reported margin, chain-verified burn. PAYG never changes seat receipts, record roots or seals.

Checking PAYG accounting.

Prepaid USDG inference · Aggregate JSON and statement hash