FeeFloor Research

Which crypto exchange is actually the cheapest?

You are reading the live rolling series — these figures move every six hours as new sweeps land. The citable, frozen release is Version 1.0 — 50 sweeps, 73,001 observations, data cutoff 24 July 2026, SHA-256 verified. Quote Version 1.0 if you need a number that will still be true next month.

By the FeeFloor Data Team·Continuously measured since 15 July 2026·88 full measurements so far·Updated 3 Aug 2026, 12:02 UTC

The short answers

The cheapest venue for the trade most people actually make, by the dimensions that move it.

The scenario matrix (market buys)

This is the authoritative result: the cheapest venue for every asset × market × order-size combination we measure, by all-in cost of a market buy. Switch asset, market and time window below — each cell shows the winner, its all-in cost in dollars and basis points, and its margin over the runner-up. Windows that don’t yet have enough sweeps say so rather than pretending to a conclusion.

Asset

AllBTCETHSOL

Market

AllPerpetualSpot

Window

LatestSince start7 days30 days

Winner per cell over the latest window · 1 sweep

Asset · market	$1K	$10K	$100K	$1M
BTCPerpetual	GMTrade$-0.32-3.21 bps · vs #2 +$0.33	GMTrade$-3.18-3.18 bps · vs #2 +$3.26	GMTrade$-30.92-3.09 bps · vs #2 +$34.10	GMTrade$-109.00-1.09 bps · vs #2 +$254.80
BTCSpot	Toobit$0.000.00 bps · vs #2 +$0.51	Toobit$0.000.00 bps · vs #2 +$5.20	Toobit$0.010.00 bps · vs #2 +$56.41	Toobit$1.800.02 bps · vs #2 +$624.60
ETHPerpetual	GMTrade$0.040.38 bps · vs #2 +$0.02	GMTrade$0.380.38 bps · vs #2 +$0.28	GMTrade$4.070.41 bps · vs #2 +$4.80	GMTrade$126.601.27 bps · vs #2 +$83.10
ETHSpot	Toobit$0.000.03 bps · vs #2 +$0.52	Toobit$0.030.03 bps · vs #2 +$5.19	Toobit$0.270.03 bps · vs #2 +$59.25	Toobit$140.001.40 bps · vs #2 +$1,355
SOLPerpetual	Lighter$0.050.48 bps · vs #2 +$0.13	Lighter$0.480.48 bps · vs #2 +$1.34	Lighter$7.990.80 bps · vs #2 +$10.39	MEXC$438.704.39 bps · vs #2 +$80.30
SOLSpot	Toobit$0.070.69 bps · vs #2 +$0.50	Toobit$0.690.69 bps · vs #2 +$5.00	Toobit$17.151.72 bps · vs #2 +$50.13	MEXC$1,17811.78 bps · vs #2 +$210.80

How the cheapest venue changes as a BTC trade grows from $1K to $1M

Follow one asset and market — a BTC perpetual taker order — across four sizes. At the bottom of the range the trading fee dominates, so the venue with the lowest fee wins. As the order grows, the spread you cross and the slippage your size pushes through the book matter more than the fee, and the ranking shifts. By $1M, some venues cannot fill the order at all and drop out entirely — the line simply stops.

Fee vs. spread vs. slippage: what moves the ranking

Three costs make up every all-in number, and they scale differently. The fee is a flat percentage — identical at $1K and $1M, so it dominates only when the others are small. The spreadyou cross is roughly fixed in basis points but reads larger on lower-priced assets, which is why SOL’s spread looks higher than BTC’s even in a tight market. Slippage is the one that grows with size: a small order barely moves the book, a large one walks through several levels. The cross-over between fee-dominated and slippage-dominated is exactly where the cheapest venue changes hands.

Here is that composition for a $100K BTC perpetual, cheapest first. The bar length is total cost; the segments are fee, half-spread and slippage. Watch whether the venue tagged “lowest fee” is the shortest bar — a smaller fee can be undone by a wider spread or more slippage.

Spot and perpetuals are different races

We never blend the two, because they are won by different venues. Perpetual futures are dominated by low-fee perp DEXs and derivatives specialists; spot is a separate contest among the large centralized exchanges. A venue that is cheapest for spot BTC can be mid-pack for the perpetual, and vice versa — reading a single “cheapest exchange” across both is how most comparisons mislead.

Zero trading fees don’t mean zero trading cost

The headline fee is the most-advertised and least-complete number in crypto. In the latest completed sweep, the venue with the lowest advertised fee also delivered the lowest all-in cost in only 14 of 24 scenarios with enough venues to compare. In the other 10, a venue charging a higher fee was actually cheaper once spread and slippage were counted — the gap the fee page never shows you.

Across all 88 measurements so far (2112 comparable cells), the lowest-advertised-fee venue was cheapest all-in 67% of the time — the stable figure that supersedes any single sweep.

The $1M question: who has real liquidity

At size, depth decides. Below is how many venues per market could not fill a $1M order within tolerance in the most recent sweep — the ones that look competitive on a fee page but run out of book when the order is real.

This section is the seed of a companion study, The $1 Million Crypto Trade: Which Exchanges Have Real Liquidity?, which will go deeper on order-book depth alone.

Methodology, limitations and data

Every number is all-in cost — fee + spread + slippage — measured from live order books, exactly as described on our methodology page. The study uses standard list prices (taker, standard tier, no referral discount) so anyone can reproduce it; referral discounts are a separate lever, not baked into these results. We re-measure all 24 scenarios every few hours and report the winner per cell plus a scenario win rate over the whole window — we deliberately do notcollapse the scenarios into one average “overall winner”, because that number would depend entirely on how you weight assets, markets and sizes. The best measured order-book depth answer excludes oracle-priced venues (GMX, Jupiter, Gains and similar), which fill any size from an external price and so have no order book to measure.

These headline figures are buy-side. Every number in the matrix and short answers is a market buy, walked against the ask side of each order book. Sell-side execution (walking the bids) is now being collected as a separate side series with its own start date and coverage — shown alongside buys in the chapter below, never averaged into them, and only once each segment has enough history to be worth reporting.

Other limitations. These are estimates of immediate execution cost at measurement time; real fills move with the book. The window so far is 88 sweeps — the longer it runs, the more robust the win rates. Funding, withdrawal and regional pricing differences are out of scope.