Flash Loan Venues evaluated across single-transaction borrowable depth, loan cost in basis points, integration surface, and venue incident record.
MO
#1 of 8 · published ranking
Morpho (Morpho Blue)
82ChainChoice Score
4199
Why it leads
Best in the pool on loan cost bps (10/10; next 9/10)
Cost
Not priced· No comparable price is published
No provider can pay for a position in this table. The code that computes this order cannot read which links earn us a commission, and every build proves it. Every score below rebuilds from the published criteria.Ranking 2026.08 · 4 criteria · 8 products · same inputs, same order
Personal comparison
Flash Loan Venues ranked comparison
No comparable price is published; ranking still uses verified product evidence.
Ranked on 4 published criteria weighted 30/28/22/20, which set goal alignment — 30 of the 86 points. The rest: regional access 20, evidence depth 18, ease of use 10, institutional trust 8. Profile match is shown in breakdowns but carries no weight.
Audit
Infrastructure
Methodology
2026.09.15
published 2026-09-16
Providers tracked
980+
across 119 categories
Last verified
2026-09-27
newest dated pricing or sentiment read
Named on the roster
2 people
managing directors · 6 automated processes
Decision guide
What matters most before choosing in this category
Weighted on single-transaction borrowable depth (30), loan cost in basis points (28), integration surface (22), and venue incident record (20). This deliberately scores nothing a supply-APY comparison scores. The headline: SparkLend runs Aave v3 code - its own docs describe what it changed relative to "Upstream Aave v3" - and returns FLASHLOAN_PREMIUM_TOTAL() = 0 against Aave's 5, on a pool DefiLlama values at $4.51bn on Ethereum. On a $100m USDS or DAI flash loan, both sizes SparkLend can fund, the identical operation with the identical executeOperation callback costs $50,000 at Aave and nothing at SparkLend. What Aave's premium buys is breadth of asset and chain, not a better flash loan.
What matters most before choosing in this category
Weighted on single-transaction borrowable depth (30), loan cost in basis points (28), integration surface (22), and venue incident record (20). This deliberately scores nothing a supply-APY comparison scores. The headline: SparkLend runs Aave v3 code - its own docs describe what it changed relative to "Upstream Aave v3" - and returns FLASHLOAN_PREMIUM_TOTAL() = 0 against Aave's 5, on a pool DefiLlama values at $4.51bn on Ethereum. On a $100m USDS or DAI flash loan, both sizes SparkLend can fund, the identical operation with the identical executeOperation callback costs $50,000 at Aave and nothing at SparkLend. What Aave's premium buys is breadth of asset and chain, not a better flash loan.
Can the venue actually fund your size, today, in the asset you need?
What is the fee in basis points - read from the contract, not the docs?
Is an incident in its record an exploit OF the venue, or one that merely used it?
Current editor lead
Expert review and scoring weights
MO
Morpho (Morpho Blue)Data checked Sep 2026
Immutable lending primitive whose flashLoan draws on the entire token balance of the singleton contract, one asset per call.
Immutable lending primitive whose flashLoan draws on the entire token balance of the singleton contract, one asset per call. Strongest on loan cost in basis points (10/10): The Morpho Blue interface states that `flashFee` is zero, and the flashLoan code sends out and pulls back the same amount with no fee term and no fee parameter governance could raise. Weakest on integration surface (7/10): One dedicated call, morpho.flashLoan(token, amount, data), answered by one callback, onMorphoFlashLoan; it lends one asset per call, pulls repayment itself, and its interface warns that it is not ERC-3156 compliant. Published price: Zero, and structurally so. The interface source at raw.githubusercontent.com/morpho-org/morpho-blue/main/src/interfaces/IMorpho.sol, fetched 2026-09-27, states in the natspec directly above flashLoan, across four consecutive comment lines: "Warning: Not…
Best forLoan cost in basis points — 10/10
Main tradeoffThe callback is not told which asset it received. onMorphoFlashLoan(uint256 assets, bytes calldata data) passes the amount and your own calldata and nothing else, so a receiver written to handle more than one token must carry the token address through `data` itself; and because repayment is pull-based, a receiver that forgets to approve Morpho reverts the whole transaction after the gas has been spent — the docs put it as 'If your callback fails to approve the repayment, the entire transaction will revert'.
Verify before signupZero, and structurally so. The interface source at raw.githubusercontent.com/morpho-org/morpho-blue/main/src/interfaces/IMorpho.sol, fetched 2026-09-27, states in the natspec directly above flashLoan, across four consecutive comment lines: "Warning: Not ERC-3156 compliant but compatibility is easily reached:" / "`flashFee` is zero." / "`maxFlashLoan` is the token's balance of this contract." / "The receiver of `assets` is the caller." The implementation, raw.githubusercontent.com/morpho-org/morpho-blue/main/src/Morpho.sol fetched the same day, has no fee arithmetic anywhere in the path (lines 422-432): "IERC20(token).safeTransfer(msg.sender, assets);" then "IMorphoFlashLoanCallback(msg.sender).onMorphoFlashLoan(assets, data);" then "IERC20(token).safeTransferFrom(msg.sender, address(this), assets);". The amount pulled back is the amount sent out, and there is no fee parameter for governance to raise. Unchanged since the 2026-08-07 read.
Methodology
How this category is reviewed
Reviewed on single-transaction borrowable depth, loan cost in basis points, integration surface, and venue incident record.
The order on this page is the published ranking for this category. Every criterion, weight and source behind it is on the methodology page.
Frequently asked
Questions people ask before choosing flash loan venues
Which venue is cheapest?
Several charge zero, and the useful question is which of those can fund you. SparkLend returns FLASHLOAN_PREMIUM_TOTAL() = 0 with $4.51bn on Ethereum and 274,141,154.96 USDS plus 112,668,695.62 DAI available at the block read. Balancer V2 is the famous free venue and is now largely empty: its own live documentation still says "On mainnet the vault holds over a billion dollars worth of assets", while the Vault held 306,263.69 USDC, 1,298.08 WETH, 96,231.44 DAI and 11.79 WBTC at Ethereum block 25698583, against all-chain V2 TVL of roughly $26.4m. That is off by about 38x, on the single sentence a developer is most likely to act on. Aave v3 charges 5 bps and buys breadth rather than a better loan.
Does the flash-loan fee still go to suppliers?
At Aave, no, and the documentation has not caught up. The guide presents the split as "Fee to LP: FLASHLOAN_PREMIUM_TOTAL - FLASHLOAN_PREMIUM_TO_PROTOCOL" and reassures that "At initialization, FLASHLOAN_PREMIUM_TO_PROTOCOL is set to 0". Read on 2026-08-06, FLASHLOAN_PREMIUM_TO_PROTOCOL() returns 10000 on all five Pools checked, and the Pool reference documents that parameter as "expressed in bps" - so 10000 bps is 100%. The documented "Fee to LP" line evaluates to zero, while the same reference still describes a design where "A part is sent to aToken holders as extra, one time accumulated interest". Suppliers whose liquidity funds every flash loan now receive none of the 5 bps, and nothing on the guide says so.
Are flash loans responsible for all those exploits?
Usually they are the tool, not the hole, but the honest version is narrower than the usual defence. DefiLlama's hacks dataset carries 612 records, 93 of which name "Flashloan" as a technique - the largest being Euler V1, 2023-03-13, $197,000,000. In most of those the flash-loan venue was repaid within the transaction and lost nothing; the loss sat at whatever the borrowed capital was pointed at. But it is not true that none is a loss at a venue: CREAM, Rari, Sonne and UwU were themselves flash-loan lenders when they were exploited. When you read a venue's incident record, separate an exploit OF the venue from one that merely borrowed through it - the two say completely different things about the code you are about to integrate.
Not financial advice · For informational purposes only · Always do your own research
//Analytics consent·GDPR · ePrivacy · TTDSG
ChainChoice measures page views and conversions with two cookieless, EU-hosted services: Plausible and Cloudflare Web Analytics. Nothing loads until you accept, and rankings are identical either way.ChainChoice measures how the engine is used — page views, conversions, referrer — through two cookieless, EU-hosted services: Plausible and Cloudflare Web Analytics. No advertising cookies, no cross-site profile, no data resale. Neither script loads until you accept, and rankings are identical whether you accept or decline.
