1inch is a DEX aggregator with Pathfinder split-route swaps

1inch is a token-swap service that uses Pathfinder, a route finder, to search many on-chain markets called decentralized exchanges. It compares pool prices, gas estimates, and available depth, then builds one atomic swap that divides an order among several liquidity sources when the combined route returns more tokens.

A stablecoin trade across fragmented pools

A USDC-to-DAI trade shows 1inch at its clearest: Pathfinder compares direct pools, connector-token routes, and split allocations before the wallet signs a final quoted transaction.

Curve supplies one source of deep stablecoin liquidity, while Uniswap and Balancer expose different prices and fee structures. Pathfinder evaluates those pools and connector assets such as WETH, rather than assuming that the apparent direct pair is best. USDC uses 6 decimal places on Ethereum, whereas DAI and WETH use 18. The router normalizes those units, checks how much each pool returns for each allocated slice, and combines compatible legs inside one transaction. Raw token amounts become comparable only after decimal precision and pool mathematics align.

Split routing has the clearest value on larger orders and pairs whose liquidity sits across several venues. A shallow direct pool moves farther along its pricing curve as the order grows. Dividing the amount between Curve, Uniswap, and Balancer reduces that pressure when their combined depth justifies the added calls. The same mechanism serves stablecoin exchanges, treasury rebalancing, and ordinary token swaps without moving custody to a centralized order book.

The useful output is the amount received after route costs, not the prettiest spot price.

What does a split route cost?

A split route costs the blockchain gas for every executed contract call, plus the trading fees embedded in the selected pools; 1inch displays a combined estimate before submission on the chosen network at that moment.

On Ethereum, 1 Gwei equals 0.000000001 ETH, and a plain ETH transfer has a fixed intrinsic cost of 21 000 gas before smart-contract execution. EIP-1559 separates the base fee from the priority fee, producing 2 visible fee components before the gas limit enters the calculation. A token swap consumes more gas than a transfer because the router calls contracts, moves ERC-20 balances, and records pool state. The wallet multiplies expected gas use by the offered fee per gas to display an estimate in the native token.

On Ethereum, established Uniswap v3 fee tiers include 0.01%, 0.05%, 0.30%, and 1%. A routed swap pays the fee that each chosen pool charges only on the portion sent there. Curve pool parameters and Balancer designs follow their own rules. Pathfinder contrasts those deductions and added gas with the tokens returned. A direct single-pool route wins when its slightly weaker gross quote costs less to execute.

Pathfinder balances rate, gas, and market depth

Pathfinder selects an executable route by comparing net token output across direct swaps, multi-hop paths, split allocations, and the gas burden created by every additional step.

The algorithm models liquidity at different depths, so it does not treat a pool as one flat price. It also consolidates compatible steps and uses concentrated liquidity where the selected range remains active. The 2025 Pathfinder upgrade delivered rates up to 6.5% better than its predecessor in a comparison covering more than 30 000 real-time trades. That figure describes the measured comparison, not a fixed improvement for every order. Token pair, amount, pool state, and gas price still determine the route that reaches the wallet.

Price impact explains why the best path changes with order size. A route that sends everything through one Uniswap position suits a small swap, while a larger order reaches better aggregate depth through several pools. Connector tokens such as WETH also open paths where no efficient direct pair exists. For tiny trades, a simpler path beats a marginally better gross rate after gas.

Classic, Fusion, and Fusion+ settle different intents

The 3 main 1inch execution modes solve separate jobs: Classic submits a quoted on-chain route, Fusion invites resolver competition, and Fusion+ coordinates an intent across chains.

Classic uses the Aggregation Protocol and Pathfinder output directly. The connected wallet pays network gas, and the whole route settles atomically. A separate Limit Order Protocol handles signed orders with a target price; a taker later executes an acceptable order and pays that execution gas. These paths expose more direct control over route settings or price conditions than an intent auction.

Fusion turns the swap into a signed intent. Resolvers compete through a Dutch auction whose rate descends until one accepts the order within its conditions. The resolver submits the blockchain transaction and covers its gas, while the economic cost enters the offered execution terms. Partial fills divide an enabled order into smaller fills. An existing token allowance still matters, so a first interaction with an ERC-20 token requires an approval transaction unless the token supports a permit.

Resolver access follows defined staking rules. An approved resolver needs at least 5% of total Unicorn Power, and the listed resolver set has a maximum of 10. Staking lock choices run from 1 month to 2 years, with 2 years as the default. Those parameters connect the 1INCH token to resolver qualification and 1inch DAO participation, rather than to Pathfinder’s pool-by-pool quote calculation.

Fusion+ extends the intent model across networks, replacing a separate bridge-and-swap sequence with coordinated cross-chain settlement.

Starting with a wallet, network, and approval

A first 1inch swap requires a compatible wallet, the correct chain, the tokens being exchanged, and either native gas for Classic or a usable Fusion allowance before that quoted action can settle on-chain.

MetaMask, Coinbase Wallet, Ledger, and WalletConnect cover common Ethereum Virtual Machine connections, while EIP-6963 lets compatible browser wallets announce themselves to the interface. Phantom serves Solana connections. Network identity deserves an exact check: Ethereum uses chain ID 1, BNB Chain uses 56, Polygon uses 137, Arbitrum One uses 42161, Base uses 8453, and Optimism uses 10. The same token ticker may exist under different contracts on different chains, so the selected network and token contract must agree before any quote has meaning.

Use these 5 conditions to choose the execution path:

• Select Classic when the wallet holds enough native gas for both approval and swap transactions.
• Select Fusion when the token allowance exists and resolver-paid execution matters more than manual route settings.
• Select Fusion+ only when the source and destination assets sit on different supported networks.
• Compare the final receive amount after gas, pool fees, and price impact, not the headline exchange rate.
• Reject a quote when its token contract, destination network, or minimum received amount differs from the intended trade.

Once those conditions line up, the quote becomes a transaction choice rather than a network-setup problem, as covered in Swap Preview.

Limits that matter before signing

The main 1inch trade-offs come from route complexity, persistent token allowances, changing pool depth, and the network fee already committed when a Classic transaction reaches validators. An atomic route has 2 outcomes: every leg settles, or the transaction reverts. A Classic revert still consumes gas. ERC-20 allowances persist until their amount is spent or changed, and shallow liquidity leaves Pathfinder with only 1 meaningful pool. A smaller order or deeper market then beats additional routing complexity.

When does a direct DEX or another aggregator fit better?

A direct DEX fits when one known pool dominates the trade, while another aggregator fits when its auction, solver set, or venue coverage produces a stronger executable quote for the same order on the same chain.

Uniswap gives direct access to its pools and concentrated-liquidity positions. Curve focuses strongly on like-asset and stablecoin liquidity. CoW Swap uses solver competition and batch auctions, ParaSwap aggregates routes across decentralized liquidity, and Matcha presents 0x routing. Each system builds or sources execution differently. Compare the same token contracts, chain, amount, minimum received value, and network cost. One venue should win because its final executable terms fit the order, not because an aggregator or direct DEX carries a familiar label.

The routing idea began on May 19, 2019, when Sergej Kunz and Anton Bukov built the prototype at an ETHGlobal event in New York. The 1INCH token launched on December 25, 2020 with 1.5 billion tokens minted at genesis, and its mint function was permanently disabled on April 6, 2022. The 1inch DAO uses the token for governance, while Pathfinder remains the mechanism that discovers routes. The project grew from one routing prototype into a broader execution network, but the choice still starts with the quoted trade.

Useful questions about 1inch