How apys picks where your stablecoins go to earn
apys is an app where stablecoins earn interest. The obvious first question is: where does that interest come from?
It comes from borrowers who want capital against liquid, non-stablecoin collateral, and from traders paying for leverage. That demand is met through lending markets built and run by Kamino, Jupiter Lend, and Loopscale. Between them, these three venues hold roughly 92% of all tracked stablecoin lending value on Solana, as per DefiLlama.
Each of the three solves the lending problem differently, and that variety is what lets apys match a deposit to a user's actual constraints, namely size and risk tolerance.
Kamino: one big pool, risk walled off
Kamino runs one shared pool of liquidity for its main assets: your USDC and other depositors' USDC all sit together in the same place, rather than being split up into lots of small, separate pools. A bigger shared pool means steadier rates and deeper liquidity—there's always capital available, instead of your deposit sitting in a thin, isolated pocket.
Your deposit represents a growing claim on that pool: the exchange rate climbs continuously as interest accrues, so your balance works for you automatically. Within that shared pool, similar assets can be grouped together: for example, USDC and USDT, so borrowing against one to access the other is more efficient, without that pairing putting the rest of the pool at extra risk.
Separately, Kamino keeps its riskier or more exotic collateral types (volatile or less-liquid assets) walled off in their own isolated markets, so if something goes wrong there, it can't spread and affect the main pool where stablecoin deposits sit.
Kamino also has the deepest published security record: 20 external security reports, 15 traditional audits, 4 formal verifications, and dedicated fuzzing campaigns. These are all publicly accessible. Auditors span across some of the most reputable security and audit firms, such as OtterSec, Sec3, Offside Labs, Certora, and Ackee Blockchain (fuzzing). This shows that Kamino deliberately uses multiple firms and methodologies so no single auditor's blind spots carry through.
Kamino launched in 2023 and, as of publication, has not experienced any public exploits.
Jupiter Lend: one pool, two jobs
Jupiter Lend is built on a shared Liquidity Layer which is a single core program that holds every deposited and borrowed asset. Rather than each product attracting and fragmenting its own pool, Earn (where USDC/USDT depositors sit) and Borrow (leveraged vaults borrowing against collateral) both plug into that same underlying pool.
For a stablecoin depositor, this matters because your USDC isn't sitting idle in an isolated Earn-only pool waiting for lenders to show up. The same USDC pool simultaneously backs Borrow-side leverage positions, so utilization (and the interest rate) respond to demand across the whole protocol, not just one product silo. Jupiter frames this as the core capital-efficiency case: shared liquidity means users borrow more and earn better yields without requiring separate deposits for each product.
The architecture also has a built-in safety valve for stablecoin lenders: dynamic borrow and withdrawal ceilings. Borrowing and withdrawal capacity expand gradually rather than jumping instantly, and the ceiling contracts the same way if withdrawals accumulate. That's a deliberate throttle against a liquidity shock draining a USDC pool all at once.
Jupiter Lend has 10 published security reviews: seven firm-conducted audits (Zenith, three from Offside Labs, Mixbytes, two from OtterSec) plus two Certora formal-verification rounds and one Code4rena contest.
Jupiter's docs flag the standard DeFi risk set for stablecoin lenders: smart contract risk, oracle risk, borrower default, and depeg risk on either side of the pool. No publicly confirmed exploit to date.
Loopscale an order book, not a pool
Loopscale's core primitive is the Credit Order Book (COB) which is technically a matching engine, not a pool. Lenders place limit orders (rate + duration + collateral type they'll accept), borrowers place market orders, and the COB pairs them directly.
For a USDC depositor who doesn't want to place individual orders, Loopscale Vaults abstract this: you deposit USDC into a vault, a Vault Curator (an individual or team responsible for managing it) sets which markets and durations that USDC is allowed to lend into, and the vault allocates across multiple borrower-facing markets on your behalf.
Curators are required to hold back a liquidity buffer so USDC withdrawals can be met without waiting on loan maturities, and if that buffer runs low, maturing loans are refinanced to other lenders to protect withdrawal capacity. Lenders know their rate going in, rather than watching it drift after the fact.
Loopscale is the only one of the three with an incident on record—and the response is why it's still in the set.
On April 26, 2025, an attacker exploited a price feed which allowed them to take out undercollateralized loans and drained around $5.8M USDC. Loopscale negotiated a 10% bounty with the attacker, and all funds were returned in full within 48 hours. Since then, Loopscale has published five additional third-party audits, and publicly adopted a hard rule: no program update ships without external review first.
You don’t have to pick
No single one of these venues is "the best" place to park a stablecoin: they're each optimized for a different tradeoff between rate predictability, depth, and capital efficiency.
Kamino offers the deepest, most liquid variable-rate market with pooled efficiency for correlated pairs. Jupiter Lend runs a shared liquidity layer where your USDC simultaneously backs Earn and Borrow activity, with dynamic ceilings that throttle liquidity shocks. Loopscale trades pool depth for rate certainty (a fixed-rate order book where you know your yield before you deposit).
So which one should you use? Depends on what you're optimizing for—and that's exactly the part apys handles. Enter your deposit size and pick growth style now, and we’ll propose an allocation you can run in one go.