Blend vs XOXNO: Lending on Stellar Compared

Blend vs XOXNO at a glance

	Blend V2	XOXNO Lending
Architecture	Isolated pools, anyone can deploy one	Three hubs hold liquidity; nine spokes set risk
TVL / borrowed (8 Oct 2026)	$144.3M / $43.8M	$130K / $48K
Main assets	XLM, USDC, EURC (Fixed pool)	31 markets incl. XLM, USDC, RWAs, Aquarius LPs
Max LTV, USDC collateral	95% collateral factor (Fixed pool)	76% Blue Chip, 88% Stables & FX
Flash loans	Flash borrow as debt, no separate fee	Cash flash loans at 9 bps, fee-free flash positions
Oracles	One SEP-40 oracle per pool, fixed at creation	Reflector TWAP and RedStone must agree
Liquidations	Dutch auctions; bad debt to backstop	Health-targeted repayment with a sliding bonus
Audits	Certora, Code4rena (V2)	None published; Certora specs in repo

Blend and XOXNO by the numbers (October 2026)

Measure	Blend V2	XOXNO (Stellar)
TVL	$144.3M	$130.4K
Borrowed	$43.8M	$48.0K
Peak TVL	$173.1M (1 Oct 2026)	$133.9K (5 Oct 2026)
Largest holdings	XLM $132.6M, USDC $11.5M	XLM $70.7K, USDT0 $28.0K, EURC $23.5K
USDC supply rate	6.68% (Fixed pool)	2.48% (Core hub)
Fees, last 30 days	$324K	n/a
Tracked since	May 2025	2 September 2026

All figures are from DefiLlama on 8 October 2026. TVL means supplied minus borrowed. Blend's TVL fell about $29M in the week after its 1 October peak; DefiLlama's token breakdown shows almost all of it is XLM, so XLM's price moves the total. Network context: the best lending protocols on Stellar.

How to choose

• Supplying USDC for yield: Blend's Fixed pool paid 6.68% against XOXNO's 2.48%, with far more depth. Check the pool's status and backstop first.
• Borrowing against an Aquarius LP token or an RWA: XOXNO; Blend's main pool does not list them. See Aquarius LP tokens as collateral.
• Looping or leverage in one transaction: XOXNO's multiply packages it; on Blend you build it from a flash borrow and requests.
• Large positions: Blend, simply because XOXNO's caps and liquidity are small.

Architecture — isolated pools vs hubs and spokes

Blend is a set of isolated lending pools. Each has its own assets, risk parameters, oracle and backstop, and anyone can deploy one. XOXNO keeps liquidity in shared hubs and isolates risk differently: an account is bound for life to one spoke, and the spoke decides what it may post and borrow.

Blend's docs put the trade-off plainly: "lenders and borrowers are only exposed to the risk of the pool they're using." The cost is fragmentation; liquidity in one pool cannot serve another. Each pool also has an admin who can change most settings, unless the admin key is disabled, and Blend's docs tell users to check the admin. The Fixed V2 pool caps each user at six positions and pays 20% of borrower interest to its backstop.

XOXNO's three hubs (Core, RWA, AMM) hold the cash and interest indexes; its nine spokes, such as Blue Chip, Stables & FX and AMM Collateral, set LTVs, caps and liquidation curves per asset. Liquidity is shared, but a bad asset in one spoke cannot back a loan in another. Each account is an NFT, so a whole position can be transferred at once.

Assets and LTVs

Blend's Fixed V2 pool lists three assets: XLM with a 75% collateral factor, and USDC and EURC at 95%. XOXNO lists 31 markets across its spokes, from 50% for Aquarius LP tokens to 88% for USDC in its stablecoin spoke, and keeps RWAs, AQUA and LP tokens collateral-only.

Blend's numbers are not directly comparable to XOXNO's LTVs: Blend also applies a liability factor to borrowed assets (75% for XLM, 95% for USDC and EURC in the Fixed pool), and the borrow limit combines both. Blend also caps utilisation per asset, at 70% for XLM and 90% for USDC in that pool, which limits how much an oracle failure could drain. XOXNO's Blue Chip spoke allows 75% on XLM and 76% on USDC; its Etherfuse, Spiko and Centrifuge spokes allow 60% on tokenised Treasuries.

Flash loans

Blend V2's flash loan adds the borrowed amount to the user's debt, sends the asset to a receiver contract, then processes the user's other requests and checks health at the end, so it carries no separate fee. XOXNO has two kinds: cash flash loans that must be repaid with a 9 basis-point fee, and fee-free flash positions.

The practical difference is who the feature serves. Blend's version suits a user building leverage or refinancing inside their own position; because it is a borrow, it is blocked when borrowing is. XOXNO's cash flash loan serves arbitrageurs and liquidators, while its multiply function wraps a flash position, a swap and a supply into one call.

Oracles

Every Blend pool uses one SEP-40 oracle contract chosen at creation, and Blend's docs warn that it "CANNOT be changed after a pool is created". XOXNO requires two independent sources, a Reflector time-weighted average and a RedStone feed, to agree within a band (about ±10% for XLM) before it accepts a price.

Oracle choice is where Blend's 2026 losses started. The community-run YieldBlox pool priced USTRY from a Reflector feed sourced from a near-empty Stellar DEX market; one trade moved it about a hundredfold. XOXNO's "fail-closed" design rejects a price if either source fails or they disagree, which protects against that attack at the cost of blocking withdrawals and liquidations during outages. More: oracles on Stellar.

Liquidations and bad debt

Blend liquidates through Dutch auctions over 400 ledgers: the collateral offered rises for the first 200, then the debt the filler must take on falls. Bad debt passes to the pool's backstop, whose deposits are auctioned to cover it. XOXNO sizes repayment to restore a target health factor and pays a bonus that rises as health falls.

Blend's backstop is first-loss capital: depositors lock BLND-USDC 80:20 LP tokens, wait 17 days to withdraw, and if the backstop falls below 5% of its threshold while bad debt remains, the remaining debt is socialised across that asset's lenders. XOXNO has no separate insurance layer; its target health factor is 1.15 in most spokes and 1.02 in Stables & FX.

Incidents and current status

Blend's ecosystem had two incidents in 2026: about $10.2M drained from the YieldBlox pool in February through oracle manipulation, of which Halborn reports $7.2M was frozen by Stellar Tier-1 validators, and about $717K lost from the Comet BLND-USDC backstop pool in August through a same-asset swap bug.

The August loss still shows. On 8 October a read-only call to Blend's Fixed V2 pool returned status 3, which Blend's code defines as backstop "on ice": borrowing and cancelling liquidations are not permitted, while supplying, repaying and withdrawing continue. The V2 backstop reported its BLND-USDC token at a spot price of zero. On 4 October Blend's interface added support for a v2.1 deployment, the same audited contracts with a new backstop and a Comet v1.1 LP pool, and noted that a Fixed v2.1 pool appears once activated. For XOXNO, live on Stellar since September, we found no reported incident. Details: Blend's 2026 incidents.

Audits, governance and licence

Blend V2 lists audits by Certora and Code4rena, and Script3's chief executive personally offers a bug bounty of up to 9M BLND. Its pool contracts are immutable. XOXNO's repository has Certora specifications, a threat model and invariants, but its security policy says audit material will be linked "when available".

XOXNO's governance runs through a timelock whose configured minimum is 12 ledgers, about a minute, so a compromised owner key could change parameters quickly. Its code is under the PolyForm Noncommercial licence. Blend's risk sits with each pool's admin and with the oracle chosen at deployment. Reading audits critically: smart contract audits.

How they fit together

The two connect. XOXNO's migrate_from_blend function moves a position from an approved Blend pool (its mainnet config lists Fixed V2 and YieldBlox V2) into a new XOXNO account, and its repository includes a DeFindex strategy contract. DeFindex vaults, meanwhile, lend mostly on Blend; see what is DeFindex. Problems at Blend therefore reach users of products built on it.

The takeaway

Blend is where Stellar's lending liquidity is, and its contract design has held up; the risks are pool configuration, oracles and, right now, a backstop being rebuilt. XOXNO is the more flexible and more cautious design on paper, with better oracle defences and more collateral types, but it is tiny and unaudited by any published third party. Size positions accordingly.

Sources: DefiLlama protocol, fees and yield APIs, 8 October 2026; read-only simulations of Blend's Fixed V2 pool (get_config, get_reserve) and V2 backstop (pool_data), 8 October 2026; github.com/blend-capital/blend-contracts-v2; blend-ui pull request #238; Blend documentation (FAQ, choosing pools, liquidations, backstopping, auctions, risk parameters, oracle, audits, bad debt); github.com/XOXNO/rs-lending-xlm @ a83ec7c; Halborn's YieldBlox analysis; reporting on the August 2026 Comet exploit.