Best DEX aggregator for low fees?
Don't take a ranking. Measure it.

What the field publishes, and what we publish

Not a ranking — a table of what each project documents about itself, read from its own pages on 2026-09-07, with the source beside each row. Where routing runs, what verification is published, and whether the design is a citable paper. Odos wound down its services in July 2026 and is not listed. The performance column stays empty on purpose: measure it yourself with the procedure below.

The six costs that decide what you receive

"Low fees" is the wrong frame. The fee is one of six terms, and on most trades it is far from the biggest:

The procedure

1. 1. Fix the variables. Same pair, same amount, same chain, quotes pulled within the same minute. A quote taken five minutes later measures the market, not the venue.
2. 2. Use your real size. Rankings invert with size — the venue that wins at $100 often loses at $50,000, because depth beats fee once the trade is large relative to the pool.
3. 3. Record tokens received. Raw output amounts, not percentages. Percentages are quoted against different baselines and are not comparable.
4. 4. Subtract every cost. Fee, price impact, the slippage you must genuinely set, and gas. Only the net number means anything.
5. 5. Check the quote against the fill. Compare what landed against what was promised. Almost nobody runs this test, and it is the one that separates a computed quote from a predicted one.
6. 6. Repeat. Three sizes across a few days. One sample is an anecdote.

Point it at us first

Our quote is computed on-chain by the Quoter contract — the same math and state that settles the trade — so you can reproduce it against any block before signing anything. You are not checking our marketing; you are checking the chain.

# What we would give you for 1 WETH → USDC on Base, right now:
cast call 0x4cEF0615614B212895F45Aa1D4833B16666E18d3 \
  "previewPlan(address,address,uint256)" \
  0x4200000000000000000000000000000000000006 \
  0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913 1000000000000000000 \
  --rpc-url https://mainnet.base.org

# Or let the site prepare that exact call for your own node (it performs no RPC itself):
curl 'https://blazephoenix.xyz/api/quote/prepare?chain=base&in=WETH&out=USDC&amountIn=1000000000000000000'

Now run the same pair and size on whichever venue you use today, within the same minute, and compare output amounts. The API is free, needs no key and has open CORS, so the whole comparison — including against us — scripts in a few lines: /agents.

Our fees, stated plainly

• • 0.28% (28 bps) on quoted output. That is the entire trading fee.
• • Execution surplus returns to you, fee-exempt. When the fill beats the quote, the difference is yours and we take nothing from it.
• • No deposit, withdrawal, account, subscription or API fee. Nothing is charged on a failed transaction.
• • The fee is a public constant in the contract, not a policy on a page — read it at /verify, including where every basis point goes.

What we will not claim: that we win every route. No aggregator does, and one that says so is telling you something you can disprove in five minutes. What we do claim, because it is checkable: our number is computed by the contract that settles it, our fee is a constant you can read, and the surplus goes back to you.

The checklist — choosing where you swap

Try it live: swap · reproduce any quote via the free API.

Comparing multi-chain staking rewards

Same discipline, one addition: before comparing advertised APRs, ask whether the pool can actually pay them. Most cannot be checked at all — ours answers with a single on-chain call.

Proof, not promises: live solvency report · the conservation identity · enter the staking engine.