ART-492 Settlement Finality Classifier carries three finality models: optimistic-rollup challenge windows, validity-proof settlement, and instant-BFT consensus (the same sub-second, no-reorg family Avalanche's Evergreen L1s sit in). Some ledger-consensus designs do not fit any of those three: a payment can enter a terminal failure branch it never recovers from (XRPL), or resolve bidirectionally where a quorum can still agree on a different outcome after apparent confirmation (a federated-BFT model, as used by Stellar). Those live on a separate sibling node rather than being folded into ART-492's enum, so a receipt from either page never mixes two settlement-model taxonomies.
For settlement-ops and reconciliation teams on XRPL- or Stellar-settled rails who need a documented, vendor-neutral basis for adjudicating a counterparty's finality claim, not a live network call.
RLUSD settles on XRPL under NYDFS supervision, with monthly Deloitte point-in-time attestations (not audits) and BNY Mellon as primary custodian since July 2025. The billable question for RLUSD/XRPL settlement-ops teams specifically is what happens when a counterparty asserts a payment "definitively failed" and re-sends it: a terminal-branch model can tell the difference between a genuine terminal failure and a claim that does not hold up, which is exactly the adjudication a FALSE_FINALITY_CLAIM or ABSENCE_PROOF_INCOMPLETE verdict is for.
USDC-on-Stellar settlement and tokenized-fund traffic on Stellar give the bidirectional, federated-BFT model its own billability anchor: a quorum-based consensus design has a different failure shape than a single-leader chain, and a counterparty dispute over "was this really final" needs a model that represents that shape rather than approximating it with a rollup-style challenge window.
Every receipt from this family, like every receipt across the suite, is described to readers as an OCG receipt; the internal shorthand used in build tracking is not reader-facing terminology.
Tendermint/Cosmos consensus finalizes a block in one round with no reorg window, which is the same terminal, non-probabilistic shape as the single_slot_bft outcome model already covered here, so a Tendermint/Cosmos deployment maps to single_slot_bft rather than needing its own model.
Aptos's Byzantine-fault-tolerant consensus likewise finalizes with no probabilistic reorg window, so it also maps to the existing single_slot_bft outcome model rather than requiring a new one.
This is a documentation mapping only: no kernel change ships for Tendermint/Cosmos or Aptos, and no chain-specific parameters are added.
Watch, not build: Sui and TON both have consensus designs distinct enough from the models covered here that they are being tracked, not implemented. No kernel work is planned for either in this wave.
No. ART-492 keeps its three models (optimistic-rollup, validity-proof, and instant-BFT) unchanged. This hub is for ledger-consensus designs with terminal-branch or bidirectional outcomes, which sit outside ART-492's enum by design.
No. The mapping states that traffic on those chains can be described with the existing single_slot_bft outcome model; it is a reader-facing note, not a new tool or a new kernel parameter.