Crypto Payment Gateway Rates 2026: The Spread You Pay
We measured four crypto payment gateways' exchange rates against the live market on 18 September 2026. The hidden gap ran from 0% to 1.3% per payment.
Key Takeaways
- The crypto payment gateway exchange rate is set by the gateway, and four of them disagreed by more than a full percent on the same coin in the same minute. Measured on 18 September 2026, published rates sat 0.06% above the market mid at OxaPay, 0.31% below at Cryptomus, 0.61% below at CoinGate and 0.99% below at BitPay.
- The market barely disagrees with itself, and the gap never closed. Kraken, Coinbase and KuCoin sat within 0.073% of each other on average — but across 50 readings over 32 minutes, CoinGate and BitPay quoted below the market 50 times out of 50, on every coin, in every sample.
- Two gateways advertise conversion as free while quoting below the market. CoinGate's pricing page says the exchange fee excludes "automatic checkout conversions"; Cryptomus calls Auto-Convert "completely free of charge". Both published rates below mid on the day we checked.
- The spread reorders the fee table. On $10,000 a month, OxaPay's 1.5% headline came to about $144 all-in against CoinGate's 1% headline at about $161 — the cheaper sticker lost.
- Even stablecoins are not credited at par. CoinGate quoted 0.9943 USD per USDC, a 0.57% haircut on an asset pegged to the dollar.
Table of Contents
- Your gateway exchange rate is a fee nobody quoted you
- How we measured every gateway exchange rate
- What four gateways quoted in the same minute
- The control: the market barely disagrees with itself
- What the spread costs on real volume
- The stablecoin haircut nobody mentions
- Why "free conversion" and a 0.6% gap are both true
- What to check before you integrate
- FAQ
Your gateway exchange rate is a fee nobody quoted you
You signed up for a 1% gateway and you are paying closer to 1.6%. Nothing on your invoice is wrong and no line item is missing — the extra sixty basis points arrived in the crypto payment gateway exchange rate, the number that turns the customer's Bitcoin into the dollars credited to your balance. That number is set by your gateway, it is not audited by anyone, and on the day we measured it the four gateways we could check disagreed with each other by more than a full percentage point on the same coin at the same moment. We pulled every one of those rates from the gateways' own public endpoints on 18 September 2026 and priced them against three independent exchanges captured in the same request cycle, because this is the one gateway cost that never appears in a comparison table.
The reason it never appears is that it is genuinely hard to see. A processing fee is a percentage you agreed to and can find on a pricing page. A network fee is a blockchain cost you can verify on an explorer. The exchange rate is a single number handed to you at the moment of settlement, with nothing to compare it against unless you happened to record the market price in the same second. Most merchants never do, so the cost is real, recurring, and invisible.
- The processing fee is disclosed, agreed in advance, and identical for every transaction in your tier.
- The network fee is paid to a blockchain, verifiable by anyone, and usually the customer's problem rather than yours.
- The exchange rate is set by the gateway, disclosed nowhere, and applies to the entire value of every converted payment — which makes it the only one of the three that scales with your revenue without appearing in your accounts.
Every fee comparison we read before writing this one, including our own, hands the spread a range and moves on: 0.5% to 2%, source unstated, date unstated. So we stopped estimating it and went and measured it.
How we measured every gateway exchange rate
Four crypto payment gateways publish a rate through an open endpoint that needs no account and no API key, which means their pricing can be checked by anyone with a terminal. Three major exchanges publish a live order book the same way. Reading all seven inside a few seconds gives you a like-for-like comparison that no gateway publishes about itself.
The reference price for each coin is the average of whichever of those three exchange mids were available, and the gap is simply how far the gateway's number sits below it, as a percentage. One honest caveat, stated up front: only CoinGate labels a tier "merchant", so only CoinGate's figure is unambiguously the rate a merchant is credited at. For the other three this is the rate the company publishes about itself, and an individual invoice may settle differently. That ambiguity is not a flaw in the method — it is the finding. None of the four documents the relationship between its published rate and your settlement, and BitPay's developer reference for its rates endpoint still reads "TBD" where the calculation should be.
What four gateways quoted in the same minute
Every number below was captured at 14:41 UTC on 18 September 2026, inside a single pass that took a few seconds end to end. The percentage after each rate is how far below the three-exchange mid it sat; a negative number means the gateway quoted above the market.
The ordering is consistent enough to be a policy rather than an accident. OxaPay tracked the market on every single coin, landing within 0.22% and usually within 0.07%, on the generous side. BitPay sat a full percent below on six of the seven coins it quoted, and its one exception is worth naming honestly: Solana came in at 0.15%, which is nothing like the rest of its book and we have no explanation for it. Strip that outlier and BitPay's mean gap rises to 1.12%.
CoinGate is the interesting case, because it publishes the evidence itself. Its endpoint returns a merchant rate and a trader buy rate side by side, and in our first capture the merchant rate sat 0.34% to 0.50% below the trader rate across BTC, ETH, LTC, TRX, DOGE and BNB. You do not need an external reference to see that the merchant tier is the worse of CoinGate's two prices. You only need to read both fields in the same response.
The control: the market barely disagrees with itself
A gap of half a percent means nothing unless you know how much prices naturally vary between venues. If Kraken and Coinbase routinely disagreed by a percent, a gateway a percent off the average would be unremarkable. They do not, and that is what turns this from an observation into a finding.
Across all ten coins the three exchanges sat within 0.073% of each other on average, with a worst case of 0.104% on Ethereum. Kraken's own spread between its best bid and best ask — the price of actually trading, right now, in size — was 0.0001% on Bitcoin and never worse than 0.079%. So the cost of moving between these assets and dollars is a rounding error, and a gateway sitting a percent below the mid is not passing through a market cost. It is charging you.
That is the same shape of finding we reached on a different leg of the payment when we compared exchange withdrawal fees against real network costs: a number presented as a pass-through turns out, when you price the thing it supposedly passes through, to be a product with a margin on it.
What the spread costs on real volume
Adding the measured gap to each gateway's published headline fee changes the ranking, and it changes it in the direction nobody expects. Headline fees below are the public rates on 18 September 2026 for a merchant at the entry tier, which is where most readers of this site sit, taken from BitPay's and OxaPay's own pricing pages.
The gateway with the lowest sticker price is not the cheapest. CoinGate advertises 1% against OxaPay's 1.5% and looks like the obvious pick, but once the rate gap is counted OxaPay comes out ahead by 17 basis points. That is $17 a month on ten thousand dollars of volume — small, but it is the reverse of the decision the pricing pages point you towards, and it compounds. At fifty thousand a month the gap between OxaPay and BitPay is $775 a month, or over nine thousand dollars a year, on identical sales.
BitPay's entry tier deserves its own sentence. At 2% plus 25 cents plus a measured 0.99% rate gap, a small merchant is paying roughly 3% to accept crypto — which is what a card processor charges, without the chargeback protection a card network provides. That inverts the entire argument for accepting crypto in the first place, and we could find nothing on BitPay's pricing page that would let a merchant discover it.
The stablecoin haircut nobody mentions
Accept USDT and you have solved volatility, so the rate stops mattering — that is the received wisdom, including in our own guide to stablecoin payments. It is nearly right. There is still a rate, it is still set by the gateway, and it is still not one dollar.
CoinGate quoting 0.9943 for a USD Coin is the number to sit with. USDC is a fully reserved dollar token that traded within a hundredth of a percent of a dollar all day, and the merchant rate took 57 basis points out of it — more than half of CoinGate's entire 1% headline fee, applied to an asset with no volatility to hedge against. Whatever justifies a spread on Bitcoin does not obviously justify one here.
The practical reading: stablecoins reduce the spread, they do not remove it, and the ordering between gateways survives the switch. If you moved your book to USDC specifically to stop paying conversion costs, it is worth checking what you are actually credited per token.
Why "free conversion" and a 0.6% gap are both true
Here is the part that makes this hard to catch. Two of the four gateways tell you, in writing, that converting your crypto costs nothing — and they are not lying.
- CoinGate's pricing page states that its 1% exchange fee "applies to manual conversions on the platform" and that "automatic checkout conversions are excluded". Measured gap on checkout rates: 0.61%.
- Cryptomus describes its Auto-Convert feature as "completely free of charge". Measured gap: 0.31%.
Both statements are accurate about fees. Neither is a statement about rates, and the two are separate mechanisms. A fee is subtracted from an amount after the amount is calculated; a rate decides what the amount is in the first place. You can charge nothing for the conversion and still choose the number you convert at, and no pricing page in this industry treats the second one as pricing information.
There are legitimate costs buried in there. A gateway that quotes your customer a price and holds it while the transaction confirms is carrying real market risk — CoinGate's pricing page says it "automatically locks the exchange rate at the moment of payment", and as we found when we measured how long invoice windows actually stay open, a Bitcoin confirmation can take fifty minutes. Someone has to be paid for wearing that risk. But a hedging cost should scale with volatility, and it does not: the gap on a dollar-pegged stablecoin was 0.57% at CoinGate while the gap on Bitcoin was 0.69%. A risk premium that barely moves between the most volatile asset on the menu and the least volatile is functioning as a margin, not as a hedge.
The obvious objection to everything above is that one snapshot could be luck — a stale cache caught at a bad moment. So we sampled five coins every three and a half minutes for 32 minutes, from 14:38 to 15:10 UTC, and got fifty readings per gateway.
The gap moves, and it moves a lot — BitPay's Dogecoin gap swung from 0.89% to 1.45% inside half an hour, so where your invoice lands in the refresh cycle genuinely matters. But it never once crossed zero, for either gateway, on any coin, in any sample. That is the detail that rules out the innocent explanation. A price feed that simply lagged the market would sit above the mid roughly as often as below it, and this was not a one-way market: Bitcoin rose in four of the nine intervals and finished the window up 0.042%, while Dogecoin fell 0.382% over the same half hour. A stale quote on a falling asset reads high, not low. BitPay's Dogecoin rate stayed between 0.89% and 1.45% below the market throughout. A feed that never lands on the generous side of the market is not lagging it. It is offset from it.
What to check before you integrate
None of this makes a gateway a bad choice. It makes the published fee an incomplete way to compare them, and the fix is about fifteen minutes of work.
- Read the gateway's rate and a real exchange in the same minute. Every endpoint in our method table is public. Compare the number your candidate publishes for your busiest coin against a Kraken or Coinbase mid, three or four times across a day.
- Add the gap to the headline fee before you compare anything. A 1% gateway with a 0.6% gap is a 1.6% gateway, and it loses to a 1.5% gateway that quotes at the market.
- Check the stablecoin rate separately. It is not one dollar, and the ordering between gateways can differ from their ordering on volatile coins.
- Ask directly, in writing, how the settlement rate is derived and whether it matches the public feed. The answer, or the absence of one, is information either way.
- Consider not converting at all. A spread only exists where a conversion does.
That last point is the structural escape, and it is worth taking seriously. A gateway that forwards the customer's coin straight to your own wallet never converts anything, so there is no rate to mark up and nothing to measure — the trade-off is that you now hold the volatility yourself, which is a real cost and not a free lunch. NOWPayments runs non-custodially by default and lists a 1% service fee on its pricing page as of 18 September 2026, and BTCPay Server removes the intermediary entirely. If you need fiat in a bank account at the end of the week, you are converting, and then the only question left is who gives you the better rate for it.
Compare gateways on more than the headline number
We track fees, settlement models, supported networks and KYC requirements for every gateway in the directory, so you can see what a processor actually costs before you wire it into your checkout.
Compare Crypto Payment Gateways →FAQ
What is a crypto payment gateway exchange rate spread?
It is the difference between the market price of a cryptocurrency and the rate your gateway uses to convert it into the fiat balance it credits you. It is not a disclosed fee and it does not appear as a line item — it is built into the conversion itself. On 18 September 2026 we measured it at between 0.06% above the market and 1.30% below it, depending on the gateway and the coin.
How do I check my gateway's exchange rate myself?
Read your gateway's public rate endpoint and a real exchange's order book within the same minute, then compare. CoinGate, BitPay, Cryptomus and OxaPay all publish rates without requiring an API key, and Kraken, Coinbase Exchange and KuCoin all publish a live best bid and best ask. The endpoints are listed in the method table above. Take several readings across a day, because the gap moves as the gateway's feed refreshes.
Which crypto payment gateway has the best exchange rate?
Of the four we could measure on 18 September 2026, OxaPay was the only one whose published rates tracked the market, averaging 0.06% above the three-exchange mid across ten coins. Cryptomus averaged 0.31% below, CoinGate 0.61% below and BitPay 0.99% below. A follow-up run sampling five coins every three and a half minutes for 32 minutes found CoinGate below the market in all 50 readings and BitPay in all 50, so the ordering is stable rather than a snapshot artefact. It is still one day's reading on public feeds rather than a long-run audit — treat it as a method to repeat, not a permanent ranking.
My gateway says conversion is free. Am I still paying a spread?
Possibly, because a fee and a rate are different mechanisms. CoinGate's pricing page excludes automatic checkout conversions from its exchange fee and Cryptomus calls Auto-Convert "completely free of charge", yet both published rates below the market mid on the day we checked — 0.61% and 0.31% respectively. A zero conversion fee is a statement about what is subtracted after the conversion, not about the number the conversion is done at.
Do stablecoin payments avoid the spread?
They shrink it but do not remove it. On 18 September 2026 CoinGate credited 0.9993 USD per USDT and 0.9943 per USDC, Cryptomus 0.99733 and 0.99793, BitPay 0.99833 per USDT, and OxaPay 1.00000 and 1.00056. The 0.57% haircut on USDC is the one worth noticing, because USD Coin is a fully reserved dollar token with no volatility for a spread to hedge.
Is the spread just the gateway covering its own costs?
Partly. A gateway that locks a price while a transaction confirms is carrying genuine market risk, and confirmation can take far longer than the fifteen-minute window most invoices use. But the cost of actually trading these assets is tiny — Kraken's bid/ask spread on Bitcoin was 0.0001% and three major exchanges agreed within 0.073% on average. A gap ten to twenty times the market's own disagreement is a margin on top of a cost, not the cost itself.
How do I avoid paying an exchange rate spread at all?
Do not convert. A gateway that forwards the customer's coin directly to your wallet performs no conversion, so there is no rate to mark up — NOWPayments operates this way by default and BTCPay Server removes the processor altogether. The cost of that choice is that you hold the price risk yourself and you still have to off-ramp eventually, at which point you are paying someone a spread anyway. For most merchants the realistic goal is choosing a smaller spread, not escaping one.
Affiliate disclosure: payyd.co earns a commission on sign-ups made through our /go/ links, including the NOWPayments link above. We have no affiliate relationship with Kraken, Coinbase or KuCoin, and they appear here only as independent price references. Every gateway rate was read from that company's own public endpoint at 14:41 UTC on 18 September 2026 and priced against exchange order books captured in the same pass; rates change continuously, so re-run the check yourself rather than relying on a figure from a given minute. Only CoinGate labels a published tier "merchant", so for BitPay, Cryptomus and OxaPay these are the rates each company publishes about itself and an individual invoice may settle at a different number — that undocumented gap between a public feed and a settlement rate is the subject of this article, not an oversight in it.