Verdict
A mature multichain aggregator whose Fusion mode turns swaps into signed intents filled by competing resolvers — no gas, no failed transactions, meaningful MEV protection. Routing on the classic path remains strong and carries no protocol fee.
- Best for
- Multichain swaps with MEV protection
- Cost
- No protocol fee on classic swaps
What works
- Fusion mode executes gasless swaps filled by competing resolvers
- Routing splits across venues and chains with a mature pathfinding engine
- No protocol fee on classic swaps — you pay pool fees and gas only
- MEV protection is built into the default path rather than an opt-in toggle
What does not
- Fusion fills can be slower than a direct swap when resolver interest is low
- The resolver set is permissioned, which is a trust assumption
- Interface exposes a lot of settings most users will not understand
- Token approvals across many chains create a permission-management burden
The problem Fusion solves
A normal DEX swap has three costs, and users typically think about one of them. There is the pool fee. There is gas — which you pay whether or not the transaction succeeds. And there is MEV: the value extracted by whoever sees your pending transaction and trades around it.
Fusion mode restructures all three. You sign an intent — the assets, the minimum you will accept, a time window. Competing resolvers then fill it, paying the gas themselves and competing on the price they can deliver you.
You pay no gas. If nothing fills, nothing happens and it costs you nothing at all.
A failed transaction is not a category of event that occurs.
Sandwiching is much harder, because there is no public pending transaction with a known slippage tolerance to attack.
What you give up
Speed, sometimes. A resolver has to want your order. On a liquid pair, fills are near-instant. On an unusual pair or during a volatile minute, the order can sit in its window and expire unfilled — which costs nothing, but does mean you did not get your trade at the moment you wanted it.
The resolver set is also permissioned. Whitelisted parties compete to fill your orders. Competition among them is what protects your price, and that competition is a design property maintained by the protocol rather than a mathematical guarantee.
Classic routing
The original aggregator path is still there and still good. The pathfinding engine splits orders across venues and through intermediate assets, and on Ethereum mainnet it will routinely beat a direct pool quote by more than any fee you were trying to save by going direct.
There is no 1inch protocol fee on classic swaps. You pay the underlying pool fees and network gas. That is the honest comparison point against front ends that add a quarter of a percent of their own on top of the same pools.
Approvals are the real risk
Aggregators require token approvals, and using one across many chains means accumulating a long list of them. Each is a standing permission for a contract to move a token out of your wallet, and each persists indefinitely until revoked.
Audit them periodically and revoke what you are not using. This is the single most neglected piece of DeFi hygiene and the one most commonly exploited months or years after the transaction that created the permission.
The transaction you signed six months ago is still a live permission. An unused approval is an open door with your name on it.
How it compares
Against Uniswap's front end: same pools, no interface fee, better default execution. Against Jupiter: the same philosophy applied to EVM chains rather than Solana, with a more explicit intent mechanism. Against a wallet's built-in swap: dramatically cheaper, because wallet swaps are where the markups live.
Verdict
Limit orders and the rest of the suite
1inch supports on-chain limit orders through the same resolver infrastructure: you sign an order with a price and an expiry, and it fills when somebody wants it. No gas to place, no gas to cancel, and no centralised venue holding your funds while the order rests.
For anyone who has been tempted to leave money on an exchange purely to have a resting order, this is the feature that removes the reason. It is not as fast as a centralised order book and it does not need to be.
Chain coverage, and where the routing gets thin
Coverage spans Ethereum and most major EVM chains. Routing quality tracks liquidity: excellent on mainnet and the larger L2s, thinner on smaller chains where there are fewer venues to split across and fewer resolvers paying attention.
On a quiet chain, check the quote against a native aggregator before accepting it. Aggregation only helps when there is something to aggregate, and a multichain product will happily give you a mediocre route on a chain it barely covers.
Score: 8.0. A mature, honestly priced aggregator with the best default execution path in DeFi for ordinary users. Fusion's resolver model is both a real trust assumption and a real improvement, and the interface is more configurable than most people need it to be.