Skip to content

October 4, 2026

SearchAbout
CRYPTO·COINBEAT

Journalism for the digital-asset economy

Ratings / Yield & Lending

Best Crypto Lending Platforms

Lending markets ranked on collateral design, liquidation behaviour in a crash, and whether you can see the risk.

8 services ratedLast verified August 16, 2026Methodology

What the score is made of

Weights total 100% · how each axis is recorded ↓
Collateral & liquidation
35%
Loan-to-value policy, oracle design and how liquidations behaved in past crashes.
Security record
25%
Audit depth, time in production and how incidents and bad debt were handled.
Rate transparency
20%
Whether the rate model is published and predictable, or set at the operator's discretion.
Counterparty clarity
20%
Whether you can identify who is borrowing your assets and against what.

The table at a glance

8 rated · top score 9.1 · tap a row for the full entry

  1. 01MorphoIsolated markets with curator-defined risk9.1
  2. 02CompoundConservative, single-borrow-asset markets9.0
  3. 03AaveThe default on-chain money market8.8
  4. 04SparkBorrowing against blue-chip collateral at predictable rates8.8
  5. 05EulerPermissionless markets with modular risk parameters8.4
  6. 06Maple FinanceInstitutional credit with named pool delegates7.8
  7. 07LednBitcoin-backed loans with open-book reporting7.6
  8. 08NexoCustodial borrowing with a consumer interface7.0
Editor’s pickRank 01

Morpho

Isolated markets with curator-defined risk

A minimal, immutable lending primitive with isolated markets on top, which contains a bad listing to the people who opted into it. The risk decision moves to the curator you choose, which is better design and a new thing for users to evaluate.

In its favour

  • Immutable core with isolated markets that contain contagion
  • Curators publish explicit risk mandates
  • Consistently better rates than pooled markets for both sides

Against it

  • Users must evaluate curators as well as the protocol
  • Shorter history than the incumbent it improves on
  • Launched2022
  • ModelIsolated markets, immutable core
  • ChainsEthereum, Base and others
Score9.1

The weighted mean of the 4 axes below — each read from the fact printed beside it.

Strongest
Collateral & liquidation9.2
Weakest
Security record8.8

Scorecard — and what it was read from

Collateral & liquidation
9.2
Each market publishes its own loan-to-value, oracle and liquidation parameters, fixed at creation and visible on-chain. When the April 2026 Kelp exploit hit the lending sector, exposure here was about $1m across two isolated markets and no other vault was affected — the isolation was tested live and held.
Security record
8.8
Immutable core contracts in production since 2022 with published audits and formal verification; no exploit of the core on record.
Rate transparency
9.0
Rates follow each market's published model; vault curators publish their mandates.
Counterparty clarity
9.2
Positions are on-chain and over-collateralised, and the curator responsible for each vault is named.

Read the full Morpho review →

The rest of the table

02

Compound

Conservative, single-borrow-asset markets

The protocol that defined algorithmic money markets, now in a v3 design that isolates each market to one borrowable asset. Conservative, well-tested and no longer where the innovation happens.

In its favour

  • Very long production history with a strong safety record
  • v3 isolation reduces cross-collateral contagion
  • Simple, predictable interest-rate model

Against it

  • Fewer assets and lower yields than competitors
  • Development and governance pace has slowed
  • Launched2018
  • ModelIsolated single-borrow markets
  • ChainsEthereum and L2s

Scorecard — and what it was read from

Collateral & liquidation
9.0
Collateral factors, liquidation penalties and oracle sources are published on-chain; a 2021 distribution bug over-paid COMP rewards without affecting deposits.
Security record
9.0
In production since 2018 with a long published audit history and no loss-of-deposit exploit on record.
Rate transparency
8.8
Rates follow a published algorithmic curve, with changes recorded as governance votes.
Counterparty clarity
9.0
Positions and collateral are visible on-chain, and all borrowing is over-collateralised.

Full Compound review →

Score9.0
03

Aave

The default on-chain money market

The most battle-tested lending market in DeFi, with a risk framework that has held through every major crash since 2020 and a governance process that actually adjusts parameters before things break. Everything about it is visible, including its bad debt.

In its favour

  • Longest crash-tested record of any on-chain lending market
  • Active risk governance with published parameter rationale
  • Full transparency of positions, rates and bad debt

Against it

  • Listed a bridged asset it could not verify: April 2026 left it carrying $177–236m of bad debt
  • Shared-pool design means one listing decision reaches every depositor
  • Launched2020
  • ModelPooled, over-collateralised
  • ChainsEthereum and major L2s

Scorecard — and what it was read from

Collateral & liquidation
8.2
Loan-to-value, liquidation thresholds and oracle sources are published per asset on-chain, and two bad-debt events are on the record: roughly $1.6m from the November 2022 CRV short squeeze, covered by the treasury, and between $177m and $236m in April 2026, when unbacked rsETH minted in the Kelp bridge exploit was posted as collateral on v3 and v4 and borrowed against. TVL fell by about $6.6bn in the days that followed.
Security record
9.0
In production since 2020 across many chains with a large published audit history and no exploit of its own contracts; the 2026 bad debt came from an asset it had listed, not from a flaw in its code.
Rate transparency
9.0
Rates follow a published algorithmic curve per asset, with every parameter change recorded as a governance vote.
Counterparty clarity
9.2
Every position, its collateral and its health factor are visible on-chain, and all borrowing is over-collateralised — though April 2026 showed that visibility does not help when the collateral itself was minted out of nothing.

Full Aave review →

Score8.8
04

Spark

Borrowing against blue-chip collateral at predictable rates

Backed by the Sky ecosystem's balance sheet, which lets it offer unusually stable and deep borrowing rates on its own stablecoin. The stability comes from a single ecosystem's decisions, which is the concentration to be aware of.

In its favour

  • Deep, predictable liquidity for stablecoin borrowing
  • Governance-set rates that do not spike with utilisation
  • Built on well-audited Aave-derived contracts

Against it

  • Tightly coupled to one ecosystem's treasury policy
  • Collateral list is deliberately narrow
  • Launched2023
  • ModelPooled, ecosystem-backed
  • ChainEthereum and L2s

Scorecard — and what it was read from

Collateral & liquidation
8.8
Collateral parameters and oracle sources are published on-chain; the collateral list is deliberately narrow.
Security record
8.6
Built on audited Aave-derived contracts and in production since 2023 with no exploit on record.
Rate transparency
9.2
Borrow rates are set by governance rather than by a utilisation curve, and each rate decision is published as a proposal.
Counterparty clarity
8.8
Positions are on-chain and over-collateralised; the backstop is the Sky ecosystem's balance sheet, which publishes its own accounting.

Full Spark review →

Score8.8
05

Euler

Permissionless markets with modular risk parameters

Rebuilt after a 2023 exploit in which the attacker returned nearly all the funds, and the v2 modular architecture is now one of the most flexible lending designs in production. The history is part of the assessment; so is the recovery.

In its favour

  • Highly modular design supporting bespoke market parameters
  • Full recovery and repayment after its 2023 exploit
  • Extensive audit and formal verification programme post-rebuild

Against it

  • Suffered one of the largest DeFi exploits on record
  • Flexibility places more risk-configuration burden on market creators
  • Launched2021 (v2 from 2024)
  • ModelModular permissionless markets
  • ChainEthereum and L2s

Scorecard — and what it was read from

Collateral & liquidation
8.4
Per-market parameters, oracles and liquidation rules are published on-chain, and v2 isolates each market in its own vault.
Security record
7.8
A March 2023 flash-loan exploit took roughly $197m and the attacker returned substantially all of it; the rebuild came with a published formal-verification and audit programme.
Rate transparency
8.8
Rates follow published per-market models set by the vault creator.
Counterparty clarity
8.8
Positions are on-chain and over-collateralised, and the vault creator responsible for each market is identifiable.

Full Euler review →

Score8.4
06

Maple Finance

Institutional credit with named pool delegates

On-chain institutional lending where pool delegates underwrite named borrowers, which is a real credit product rather than an over-collateralised money market. It suffered defaults in 2022, restructured, and now runs a more conservative, mostly collateralised book.

In its favour

  • Genuine institutional credit market with named delegates
  • Post-2022 model is materially more conservative
  • Detailed reporting on loan composition

Against it

  • Experienced significant borrower defaults in 2022
  • Depositors take credit risk, not just market risk
  • Launched2021
  • ModelDelegated institutional credit
  • ChainEthereum and Solana

Scorecard — and what it was read from

Collateral & liquidation
7.6
Loans are underwritten by named pool delegates and collateralisation varies by loan, disclosed per pool.
Security record
7.4
In production since 2021; borrower defaults in 2022 caused documented depositor losses in affected pools, after which the model was tightened towards collateralised lending.
Rate transparency
8.4
Rates are set per pool by the delegate and published with the loan terms.
Counterparty clarity
8.0
Borrowers are identified at institution level and loan composition is reported per pool — depositors carry credit risk, not only market risk.

Full Maple Finance review →

Score7.8
07

Ledn

Bitcoin-backed loans with open-book reporting

A focused Bitcoin lender that publishes proof-of-reserves and, unusually for the custodial category, discloses how its book is structured. The narrow product is the point: fewer assets, fewer ways to be surprised.

In its favour

  • Regular proof-of-reserves with third-party verification
  • Unusually open reporting for a custodial lender
  • Focused Bitcoin-backed product with clear terms

Against it

  • Narrow asset support compared with rivals
  • Still a custodial credit relationship with counterparty risk
  • Founded2018
  • ModelCustodial, Bitcoin-focused
  • DisclosureProof of reserves, book structure published

Scorecard — and what it was read from

Collateral & liquidation
7.8
Loan-to-value and margin-call terms are published, and collateral is held with named custodians.
Security record
7.6
Operating since 2018 with semi-annual proof-of-reserves attestations by a named firm — the most recent completed 31 March 2026 — and more than $10bn of loans funded without a client asset loss on record.
Rate transparency
7.8
Rates are published as a rate card rather than negotiated per customer.
Counterparty clarity
7.2
Publishes the structure of its book, including the share lent to institutional counterparties — unusual disclosure for a custodial lender, though individual borrowers are not named.

Full Ledn review →

Score7.6
08

Nexo

Custodial borrowing with a consumer interface

A long-running custodial lender that survived 2022 without halting withdrawals, which distinguishes it from most of its peers. You cannot see who is borrowing your assets, and that structural opacity caps how highly it can score here.

In its favour

  • Continued honouring withdrawals through the 2022 credit crisis
  • Real-time attestation programme on reserves
  • Straightforward consumer product with card integration

Against it

  • Borrower identity and loan book composition are not disclosed
  • Rates depend on holding the platform's own token
  • Founded2018
  • ModelCustodial lending
  • DisclosureReserve attestations, no loan-book detail

Scorecard — and what it was read from

Collateral & liquidation
7.4
Loan-to-value bands are published per asset, but liquidations are executed by the company rather than by a public contract.
Security record
7.0
Operating since 2018 and continued honouring withdrawals through the 2022 credit crisis; runs a real-time reserve attestation with a named accounting firm, though not a full audit of the loan book. In January 2026 California's financial regulator penalised it $500,000 for unlicensed lending to state residents between 2018 and 2022.
Rate transparency
7.0
Rates are set at the company's discretion and vary with how much of the platform token a customer holds; the tier tables are published.
Counterparty clarity
6.4
The borrower side of the loan book is not disclosed, and depositors are unsecured creditors of the company.

Full Nexo review →

Score7.0

↑ Back to the table at a glance

What the record supports

Morpho takes the table because April 2026 tested the thing it is built on: when unbacked rsETH reached the lending markets, its isolated design contained the damage to about $1m while Aave, which had listed the asset in a shared pool, was left carrying $177–236m of bad debt. Aave remains the deepest and most crash-tested market in DeFi and is still the default for most borrowers — but the gap between the two architectures stopped being theoretical this year. The custodial lenders at the bottom are convenience products, and their history should be read before their rate card.

A conclusion drawn from the facts above, and the only part of this page that is.

How a score is read

Each axis is read off the same five bands. They describe what is on the record, not how impressed we are.

9.0–10
Documented and independently verifiable
The claim is evidenced by a published record a third party can check — an attestation, an on-chain contract, a regulator's register — and nothing adverse is on file.
8.0–8.9
Documented, with gaps
Evidence exists but is partial, dated, or covers only part of what the axis measures.
7.0–7.9
Self-reported only
The operator publishes the information and no independent party has verified it.
6.0–6.9
Adverse event on record
A recorded incident, enforcement action or failure that has since been resolved, remediated or repaid.
Below 6
Undocumented or unresolved
No published evidence, or an incident with no resolution on the record. An absence of evidence is scored as an absence.

How we scored this table

Every large failure in crypto lending had the same shape: undisclosed counterparties, undercollateralised loans and a rate that looked good because the risk sat where the depositor could not see it. This table records against that pattern — whether positions are visible, whether every loan is over-collateralised, and who absorbs a shortfall.

On-chain markets score well on counterparty clarity by construction, because positions, thresholds and any bad debt are public. A custodial lender starts from a weaker evidentiary position and can only close the gap with disclosure — reserve attestations by a named firm, published loan-book structure, published rate cards.

Liquidation design is recorded against real events rather than parameters. Where a crash produced bad debt, the amount and who covered it is on the page, because that is the fact that distinguishes an oracle that held from one that did not.

  • Every score on this page carries the fact it was read from, printed beside the bar.
  • Bad debt is public on-chain, so it is checked rather than requested.
  • An undisclosed loan book is recorded as undisclosed and cannot score in the top band.

What each axis records, and where the facts come from

Collateral & liquidation35%
Published loan-to-value and liquidation thresholds per asset; the oracle used and how often it updates; bad debt recorded after past crashes and who absorbed it.
Source: On-chain risk parameters and oracle configuration, bad-debt dashboards, governance post-mortems.
Security record25%
Audits published, time in production, value held, and every recorded exploit with the amount and the repayment outcome.
Source: Published audits, incident post-mortems, on-chain recovery records.
Rate transparency20%
Whether the interest-rate model is published and algorithmic, or set at the operator's discretion, and whether rate changes are recorded.
Source: On-chain rate-model parameters, governance proposals and votes, terms of service.
Counterparty clarity20%
Whether borrowers and their collateral are visible, whether every loan is over-collateralised, and who bears a shortfall.
Source: On-chain position data, published loan-book disclosures, terms of service.

Frequently asked questions

What is the safest way to borrow against crypto?+

Over-collateralised, on a transparent on-chain market, at a loan-to-value well below the liquidation threshold — typically half of it. The failure mode is not the interest rate; it is a fast move that liquidates you before you can add collateral.

Why did so many crypto lenders fail in 2022?+

They made undercollateralised loans to a small number of undisclosed counterparties and funded them with retail deposits marketed as savings. When the borrowers failed, the depositors were unsecured creditors, which very few of them realised.

Is DeFi lending safer than a custodial lender?+

On counterparty risk, clearly yes — the collateral is visible and the rules are enforced by code. It substitutes smart-contract and oracle risk, which is real but has proved more manageable than opaque credit.

What loan-to-value is actually prudent?+

Well under half the liquidation threshold if the collateral is volatile. Anything closer assumes you will be awake, solvent and able to transact during exactly the market conditions that make all three unlikely.

More in Yield & Lending