Token trades on a DEX get matched by an automated market maker, an order book, or a solver auction, with a smart contract enforcing the final exchange. For an ordinary spot swap, choose the pool-based AMM first; order books and intent-based routes become more useful when trade size, price control, or cross-market execution matters.
The pool does the matching
An AMM matches your trade against pooled token reserves rather than against another trader’s standing order. Liquidity providers deposit a pair such as ETH and USDC, and the pool’s pricing rule calculates how much of one asset must leave when the other enters. In the classic Uniswap model, the reserve relationship is expressed as x × y = k.
That formula explains price impact. A small USDC purchase from a deep ETH/USDC pool barely changes the reserves, while a large purchase moves the ratio sharply and gives the buyer a worse average price. The pool is not finding a seller at one fixed quote; it is selling progressively more expensive units along its curve.
The router may improve the result by splitting the trade or using multiple pools. A USDC-to-MANTA swap might travel through USDC/WETH and WETH/MANTA if that route has deeper liquidity than a direct pair. The router proposes the path, but each pool contract computes its own output and the final transaction settles all legs atomically.
Uniswap v3 and v4 add concentrated liquidity. Instead of spreading capital across every possible price, an LP chooses a range bounded by ticks. During a swap, the contract consumes active liquidity in the current tick interval; when the price crosses a boundary, liquidity from the next interval becomes active. This can make stablecoin pools and heavily traded pairs much deeper, but a position outside its range stops serving trades.
AMM matching is not order matching
An order book matches discrete bids and asks, usually using price-time priority. A market maker posts a limit order to sell 10 ETH at a chosen price, and a buyer takes that quote. An AMM has no queue of individual offers: liquidity sits in a contract, and the curve determines the price continuously. That difference is why AMMs are permissionless and convenient for long-tail tokens, while order books offer finer control for limit prices and professional quoting.
Intent-based trading is a third approach, and it is often confused with an order book. An intent is a signed request describing an outcome, such as selling a specified amount of ETH for at least a specified amount of USDC. Solvers or fillers compete to satisfy it. CoW Protocol can batch orders and use “coincidence of wants,” where two opposite user trades settle directly against each other before touching an AMM. UniswapX uses competitive filler auctions, including RFQ and Dutch-auction mechanics, to search public and private liquidity.
That is the important change visible in 2026: the interface increasingly asks what result you want, rather than forcing you to select the exact pool route yourself. The solver may combine AMMs, private inventory, and direct matches. This can improve execution and remove gas costs for failed attempts, but the quality of the result depends on competition among solvers, the auction rules, and the settlement contract.
What the trade actually costs
The displayed exchange rate is only one part of the bill. Separate the costs before confirming:
- Pool or protocol fee paid from the swap, usually shared partly or wholly with liquidity providers.
- Price impact caused by your trade consuming available liquidity.
- Network gas, including the extra work required for approvals, multi-hop routes, or tick crossings.
- Slippage risk while the transaction waits, controlled by the minimum received, deadline, or signed price limit.
Price impact is caused by your order’s size; slippage is the change that happens between quotation and execution. A trade can have low expected price impact but still fail or fill worse if the market moves before the transaction lands. A strict minimum-received setting protects the boundary, although the transaction may revert instead of filling.
Where each model shines
AMMs are strongest for permissionless spot access: a new token can trade as soon as someone creates and funds a pool, without waiting for market makers to post quotes. They are also useful for wallet-native swaps and small trades where simplicity matters more than a custom limit price.
Concentrated-liquidity AMMs shine for stablecoin swaps and liquid blue-chip pairs, where LPs can place capital around a relatively narrow trading range. Check whether the liquidity is active at the current price; a pool’s headline liquidity is less useful if most of it sits outside the range your trade will cross.
Order books and solver auctions shine for larger, urgent, or more complicated trades. A limit order can state the exact worst price, while a solver can search several venues and potentially match your order directly with another user. Compare the guaranteed output, not just the quoted spot rate.
A bridge is a separate operation from matching. Celer Network’s cBridge, for example, moves assets between supported networks; it does not set the price of the DEX pool receiving the asset. On Manta Pacific, a bridged token still needs local liquidity before a DEX can swap it, even though both steps may appear in one interface. The Manta Bridge route starts at medium.com.
The practical verdict is simple: use an AMM for straightforward, liquid token swaps; use concentrated liquidity when the active range is deep; and consider an order book or intent route when price guarantees and execution quality matter more than one-click simplicity.