Borderless Benchmark Report

01 · Key Learnings

The quarter in five findings, stated plainly. Each one is unpacked with full numbers in the sections that follow.

• Stablecoin delivery is now cheaper than the banks' own rate. Matched day for day against the interbank rate, the rate banks trade at with each other and customers never receive, stablecoin delivery priced below it for the whole quarter. We track that difference as the Parity Gap; it crossed below zero in February and has drifted further under nearly every month since. FX markup on major corridors is no longer a durable margin line, and any contract priced off last year's spreads deserves a rebenchmark.

• The price of delivering a payment has stopped moving. Delivering $10,000 cost about $27 at the typical corridor, and it's been within 30 cents of that level for five straight months. That's not providers agreeing; it's competition doing its job. Delivery at the going rate has commoditized, and the next finding is where the remaining money went.

• No provider stays the cheapest for long, and that gap is where the money is. The best price changed hands every few days in the busiest markets, so a business wired to one provider pays the going rate over time, about $2,330 more per $1 million than the best available price. That's the Routing Tax, the cost of running on one provider, and it's the biggest cost lever left in cross-border payments.

• The quarter's price shocks hit where there was no backup. Every major repricing, led by Malawi, landed on a corridor where a single provider was the only option, so the new price simply became the

  price. The one event on a corridor with backup providers kept a much cheaper path open the whole way through. A corridor with a fallback is a different product from the same corridor without one.

• None of these numbers is your number. Every figure in this report is a network-level number, and nobody actually pays the median: your cost is set by your own corridors, payment sizes, and providers. The Cost by Country section has a row or every currency we price, and your own number only exists once someone measures it.

02 · Key Numbers

The Benchmark Metrics scoreboard for Q2 2026, with change versus Q1:

Consensus rates cited in this report are the Borderless Mid. Summary of the dataset:

A quick word on "spread," because it means one specific thing here: the gap between what a provider charges to buy stablecoins and what it pays to sell them, in basis points (bps, hundredths of a percent) of the mid. It's a provider execution cost, not the premium of stablecoin pricing versus traditional FX. Those two numbers answer different questions and live in different sections. Counts behind corridor statistics are provider-days (one provider quoting on one day), not raw quotes.

And an honest note on that -37: median spreads hit 98.8 bps in March and haven't moved since, so most of that compression happened inside Q1, not during Q2. The Year So Far section shows it month by month.

03 · The Network

The numbers in this report describe a delivery network, not a rate feed. Here's its shape in Q2:

• 260 corridors across 59 currencies and 108 countries, up from 258 in Q1 (two added, none dropped); the roster of active providers changed on 21 corridors.

• 377 payout routes, where a route is one provider quoting one asset on one chain into one corridor. Routes ride seven blockchains, with USDC quoted on six chains and USDT on five, so chain choice is a third routing axis after provider and asset.

• 131 local rails on file across 70 countries, spanning 60 distinct schemes. That's a rail per country: PIX in Brazil, UPI in India, and mobile money in each market it serves all count once. Completed deliveries this quarter ran over 22 of those schemes (PIX, SPEI, BECS, NPSS, EFT, ACH, SEPA, COELSA, NIP, IBFT, FPS, MADA, SINPE, ZAHAV, mobile money). The first number is what the network can reach; the second is what actually carried payments this quarter.

• 82 corridors with failover, spanning 18 currencies: two or more providers steadily active all quarter, with 18 corridors carrying three or more. On those corridors, when a leader's pricing slips, the next quote is already there.

The quarter tested that last line. Its largest verified repricings, Malawi's move to spreads around 1,975 bps and Sri Lanka's 4.3% jump in May, along with Zambia's premium swing, all happened on corridors outside the redundancy set, with no failover path: the new price simply became the price. Ghana, the one liquidity event on a redundant corridor, kept multiple providers quoting on all 88 of the quarter's observation days, and on a typical day the best quote was 258 bps cheaper than the median quote. Redundancy didn't prevent Ghana's repricing; it kept a much cheaper path open while it happened. Expect this to become a pricing tier in the market: a corridor with a failover path is a different product than the same corridor without one, and the quarter just priced the difference.

04 · The Cost of Delivering a Payment

Total Delivery Cost is the all-in number: what providers pay out versus the Borderless Mid, plus the rail fees and local taxes on file for each corridor. It moves with ticket size, so we price every corridor at three sizes.

A flat rail fee is a rounding error on $1,000,000 and real money on $500, which is why a single bps figure with no payment size attached hides more than it says. At the median corridor, delivering $500 cost 46.2 bps ($2.31), delivering $10,000 cost 27.2 bps ($27.20), and delivering $1,000,000 cost 26.2 bps ($2,620). Routing to the best-priced provider changed those numbers to 20.0, 3.0, and 3.0 bps respectively. The best route delivered a $10,000 payment for $3.

All values in bps of the ticket, median provider unless marked; medians describe the network, not the price any one flow pays. The corridors shown are drawn from the highest-flow payout corridors in the network and chosen to span its distinct cost structures; the same computation covers all 205 priced corridors. Rail fees and taxes come from Borderless fee records as of 2026-06-24; 201 of 205 priced corridors carry the fee layer. Best-route figures are the best quote in the market at each day's pricing: you earn them by routing dynamically, not by picking today's leader once, and a single quote at or better than the mid can be a provider pricing to attract flow rather than a standing offer.

Four things worth reading off that table:

• The ticket-size crossover. The Philippines costs 22.9 bps at $500 and 2.9 bps at $1M, an 8x difference that's entirely flat rail fees fading to nothing at larger sizes. Nigeria is expensive at every size because its cost lives in the FX layer, not the fees.

• Best routes on deep corridors are near zero. Mexico's best route priced 0.1 bps off the consensus mid, and Kenya's delivered at the mid with no rail fee at all.

• India flips the structure. Its FX priced at the mid (0.0 bps), and nearly all of the 101 bps delivery cost at $10,000 is the tax and rail layer on file. Provider choice can't lower that price; only structure can. The small flat fee fades if payments are batched, but the 1% tax (a statutory withholding on these transfers, not a provider fee) scales with the payment and never fades, so India is a corridor you price into the product rather than route around.

• The Mexico rows preview the asset decision. The same corridor costs 18.5 bps on USDC and 23.0 on USDT at the $10,000 ticket; more on that under Routing & Best Execution.

The same table in dollars: delivering $10,000 to Mexico cost $18.50 at the median and ten cents at the best route. The identical payment to Nigeria cost $159 at the median and $118 at the best route. To Colombia, $81 at the median and $6.60 at the best route: a 12x difference on one routing decision. Four of the 205 priced corridors lack fee records and are reported as FX-only costs; every corridor in the table above carries the full fee layer.

The fixed-fee layer is a routing decision in its own right, because it differs across routes into the same currency:

• On the 33 currencies with per-payment fees on file from more than one provider, the gap between the cheapest and priciest route's fee ran $1.00 per payment at the median currency and $24 at the widest. Part of that range is the rail itself: an instant-rail quote and an RTGS quote are different products.

• A one-dollar difference in fees is invisible at $1,000,000 and worth 20 bps at $500.

• Measured across the 201 fee-covered corridors, routing to the best provider was worth a median 15 bps more at the $500 ticket than at the $1M ticket, and that extra value comes from differences in fees, not rates. Hong Kong and China are the clean cases: provider choice there was worth 0.1 bps at $1M and 80 bps at $500, four dollars on a single payment.

Small tickets and large tickets are optimizing different layers of the same price, often across different routes. For remittance-scale businesses the implication is that unit economics are won in fee routing, not in the FX negotiation.

05 · Corridor Grades

226 corridors earned a Corridor Grade in Q2: 3 rated A, 165 B, 36 C, 22 D.

The A list is USDC→BRL, USDC→ARS, and USDT→MXN, which joined it this quarter, upgraded from B on a score of 80.9. A grade squeezes five measured properties (depth, spread, continuity, stability, and quote coherence) into one letter a treasury can act on. A B-rated corridor delivers reliably at a defensible price; the step from B to A is usually depth and pricing discipline, not reliability. That's why the middle of the distribution matters more than the podium: 165 of 226 graded corridors now sit at B, up from February, when C was the most common grade.

Migrations against Q1 ran heavily positive: roughly 150 upgrades against 12 downgrades. The single biggest driver was European corridors. EUR corridors across dozens of receiving countries stepped from C to B (and several from D to B) on better continuity and stability, alongside upgrades in TRY, SAR, AED, QAR, HKD, VND, CNY, and IDR. The downgrades are the more useful operational signal:

Kenya and Mexico slipped below the A threshold rather than falling apart; both are still among the deepest corridors in the network. Malawi's move is the real event: the grade catching up with the April repricing and the wider spreads that followed. Zambia's downgrade landed the same quarter its currency premium collapsed into a discount, a useful reminder that cheap and reliable are different things.

The full D list is short enough to print: Botswana, Congo (CDF), Switzerland, Malawi, Serbia, and Zambia, plus the XOF zone across Burkina Faso, Benin, Côte d'Ivoire, Mali, and Togo, each on both assets. Switzerland is the surprise there, and its D is depth and continuity, not turbulence: thin, intermittent quoting on an otherwise stable corridor. We print the list because that's what the grade is for. A corridor that needs redundancy planned around it should be knowable before an event, not after one.

Grades and their components are defined in the Borderless Benchmark methodology; a data-quality limitation never moves a grade.

06 · Quarter over Quarter

Against Q1, the headline metrics barely moved. The turbulence was all regional.

The scoreboard deltas: the Routing Tax at 23.3 bps (-0.1), the Parity Gap at -3.2 bps (-0.1), Total Delivery Cost at the $10K ticket flat at 27.2 bps. Underneath:

• Spread moves concentrated in Africa. Malawi widened 1,680 bps to a 1,975 bps median after an April repricing; Uganda widened 177 bps and Tanzania 141. The compression side was also African: Kenya tightened 87 bps, Nigeria tightened 84 on USDC and 67 on USDT, and Zambia 47. Of 124 corridors comparable across both quarters, those moves bracket a market that otherwise held its pricing.

• Currency premiums repriced hard in both directions. Ghana's premium widened 219 bps to 375.7 and the Dominican Republic's 218 bps to 236.8; Argentina drifted 67 bps wider to 538.5. On the other side, Congo's parallel-market premium narrowed 179 bps (still 3,220), Nigeria's halved to 108.5, and Zambia swung 172 bps to close the quarter at a 61 bps discount to interbank.

• Regional spreads: LatAm compressed 7.5 bps to 89.0, Europe widened 3.6 to 15.8, Asia was flat at 6.1, and Africa widened 165.7 to 512.8.

Spread deltas only compare corridors whose data was reliable in both quarters; a corridor whose data quality changed is excluded rather than reported as a market move.

07 · The Year So Far

The quarterly numbers only make sense against the year's trajectory. One column per month, January through June:

January reflects a 68-corridor network before the February coverage expansion; its cost and spread levels describe a smaller, thinner market, so read that column with that in mind.

Two trajectories tell the year's story. Delivery cost found its level almost immediately after the network scaled: 27.1 bps in February, and never more than 0.3 bps away from that since. Whatever individual providers did with their pricing, and they did plenty, the cost of delivering a payment across the network has been a flat line. That's not providers agreeing; it's competition doing its job. When the lead changes hands every few days, nobody can hold pricing above the pack for long, and the aggregate settles where the market clears. For payments companies, that flat line means delivery at the median has commoditized: picking a good provider is no longer a differentiator, and the 23.3 bps between median and best is the main cost lever left, and it accrues to whoever industrializes routing. The Parity Gap is the opposite: a slow, one-directional drift from +0.6 bps in January to -5.9 in June. Stablecoin delivery pricing didn't converge to traditional FX this year. It crossed below it in February and has priced further under it nearly every month since. Q2 continued both trends rather than breaking them.

The churn row rewards a second look. March's 31 reshuffled corridors was the year's high, the tail of the February expansion as newly added providers found their pricing levels against incumbents. Since then churn has stepped down through May and held at 15 in June, without the routing gap narrowing. Fewer corridors are reshuffling, but where the lead does move, it still moves by the same margin. Competition is consolidating into the corridors that matter rather than fading out.

08 · Stablecoin vs Traditional FX

Day-matched against the same day's interbank rate across 57 currencies, the median premium for the quarter was -3.2 bps. Stablecoin delivery didn't just reach parity with traditional FX this year; it crossed below it.

Regionally:

If pricing below interbank sounds impossible, remember what interbank is: the rate banks trade with each other, which customers never receive. A traditional cross-border payment passes through a chain of correspondent banks that each add a margin on top of that rate. Stablecoin delivery replaces the chain with one conversion, made by providers competing openly for the flow, and Q2 is what that competition looks like: the middle layer's margin compressed to zero and, at the median, slightly past it.

The impact is blunt: FX markup on major corridors is no longer a durable margin line, and any contract priced off last year's spreads deserves a rebenchmark. Pricing power in this market is moving from whoever holds the spread to whoever controls the routing. What's left to compete on is settlement speed, coverage, and redundancy.

One reading note the Parity Gap always carries. Rates from a subset of providers include embedded provider fees that can't currently be separated from the exchange rate, so the Parity Gap reflects all-in client-received pricing rather than isolated FX execution. A negative gap therefore means something stronger than cheap FX: the full delivered price, fees included where embedded, landed below the interbank mid. That's rare for any cross-border delivery mechanism, and it's the main reason this number is worth tracking monthly rather than quarterly.

The two ends of the distribution:

• Deepest discounts: Zambia (-61.2 bps), Colombia (-33.8), Rwanda (-19.5), Uganda (-13.3). Zambia's number deserves the same skepticism the premiums get: a delivered price 61 bps below interbank on a managed currency, quoted on a D-rated corridor outside the redundancy set, reads less like a bargain and more like an official reference rate lagging the real market.

• Biggest premiums: the parallel-market roster: CDF at 3,220 bps, ARS at 538, XAF at 380, GHS at 376, BWP at 367. These aren't stablecoin markup. When a country's official exchange rate sits far from what its currency actually trades for, any route that genuinely delivers local money prices at the street rate; the premium is the honest price of access, and it should be read separately from the market-structure numbers.

Within the quarter the overall premium drifted just 1.4 bps tighter, so the June deepening visible in the yearly table came from the mix of currencies in the June sample rather than a mid-quarter shock.

09 · Routing & Best Execution

The Routing Tax, what a payer gives up routing at the median provider instead of the best one, held at 23.3 bps for the quarter, or $2,330 per $1M, across 81 corridors with meaningful provider choice.

The same headline figure applied on-ramp, but the two sides diverge sharply at the corridor level: Ghana's collection-side gap ran 593 bps against 258 on payout. If you run both pay-ins and payouts, benchmark each side separately, because the corridor that prices well in one direction can be the expensive one in the other.

Ranked by where the money actually flows (World Bank remittance inflows) times the routing gap:

Dollar figures are computed from unrounded rates; bps are shown to one decimal, so the two may differ by a few dollars per $1M. Leakage figures are illustrative: they assume the corridor's full remittance flow priced at the observed gap between the median and best quote. They exist to rank corridors by consequence, and the ranking is the insight. Mexico's modest 21.5 bps gap on $67.6B of flow costs as much as Colombia's 122.8 bps gap on a sixth of the volume. A treasury that only fixes its widest corridor is optimizing the wrong list. The same logic applies inside a single business: which of your corridors are leaking, and how much, is a property of your flow, not of this table.

Lead-Provider Churn explains why the gap persists:

• USDT→BRL: the best-priced provider changed 34 times in 88 days, roughly every 2.6 days, each change worth at least 5 bps. No single provider held the top spot even half the quarter.

• USDC→BRL: 28 changes, and on both Brazilian corridors every active provider held the lead at some point.

• Elsewhere: Mexico reshuffled 18-19 times, Nigeria 16-18, Chile 17, Uganda 16.

The churn isn't noise. Providers price around their own liquidity and inventory, so whoever most wants the flow on a given day quotes tightest, wins it, and steps back. That's healthy market structure, and it's also why no provider choice stays right for long: a business wired to a single provider, including whichever one is best today, pays the median over time, and the median costs 23.3 bps more than the best quote. (Three multi-provider counts appear in this report: 82 corridors with two or more steady providers, 81 with enough daily choice to price the Routing Tax, and 62 churn-scored with at least ten multi-provider days.)

Asset choice is a second routing decision on top of provider choice. Across 315,086 hour-matched USDC/USDT pairs this quarter, the median pricing gap between the two assets was 0.4 bps: at the network level they're interchangeable. At the corridor level they're not. Peru priced USDC at a persistent 99 bps discount to USDT, Chile and Switzerland ran roughly 10 bps the other way, and USDC buy-sell spreads ran consistently tighter than USDT's at the median (187 versus 259 bps). A router that treats the two assets as one instrument leaves the Peru gap on the table every day.

The quarter also recorded 8,217 hours of crossed quotes across 14 corridors, where one provider's sell price exceeded another's buy price. Brazil alone contributed 1,731 crossed hours on USDC at a median crossing of 26 bps. These are pricing inconsistencies across providers, not prices anyone could have traded on, and they're what a market that hasn't finished converging looks like.

10 · The Quarter in Motion

Month by month, the quarter went shock, turbulence, consolidation.

April's shock was Malawi: a 5.8% repricing on April 9, after which its typical spread moved from roughly 296 bps to 1,975 and stayed there. April was otherwise the quarter's calmest month, with the Routing Tax at its yearly low of 22.9 bps and the Parity Gap briefly narrowing to -2.0.

May was the turbulent month. Outlier detection flagged its middle weekend, May 16-17, as the quarter's densest anomaly cluster, Sri Lanka jumped 4.3% on May 20, and churn stayed elevated before settling. The Parity Gap resumed its slide to -3.8.

Ghana ran its own quarter. USDC→GHS spreads widened 992 bps between the quarter's first and last weeks, a 596% increase, while the currency's premium to interbank widened 219 bps and its routing gap reached 258 bps. Every layer of the corridor repriced at once, which is what separates a real liquidity event from a data artifact; the filters kept it as genuine.

June consolidated. Spreads held the 98.8 bps line for a fourth straight month, churn matched May's 15 reshuffled corridors, and the Parity Gap hit its deepest level of the year at -5.9 bps. More corridors slipped to a C in June (79, versus 30 in May) even as the quarter's aggregate grades improved, a reminder that monthly grade snapshots are noisier than quarterly ones.

11 · Cost by Country

This is the lookup section. Find the countries you deliver to and read across.

How to read the tables:

• Cost, $10K is the all-in cost of a $10,000 payment at the median route in Q2, in bps: fees and taxes included, median across a currency's routes, both assets.

• Single-provider tax is what routing at the median instead of the best route cost. A "–" means there wasn't enough provider choice on file to price a gap, or no measurable gap existed.

• Premium columns are monthly medians versus the same day's interbank rate: positive is above interbank, negative is below. A "–" means we couldn't measure that month.

• Big positive premiums (Congo, Argentina, Ghana) are parallel-market access pricing, not stablecoin markup.

• The standing caveat applies here most of all: these are network medians, and your flow prices at its own number.

Latin America

LatAm's median spread compressed to 89.0 bps this quarter, and the region owns all three A-rated corridors (Brazil and Argentina on USDC, joined this quarter by USDT→MXN). The table's standout is Colombia: a 97 bps median single-provider tax against $11.9B of annual inflows makes it the most consequential routing decision in the network. Argentina pairs a 538 bps parallel-market premium with an A-grade execution profile, an odd but real combination. And Peru's tax column is bigger than its cost column because it's a median across both assets: the USDT quotes there run far apart, which is exactly the asset-choice story from the routing section.

Africa

Regional spreads widened 166 bps to 512.8, but the average hides a fork. Kenya tightened 87 bps, Nigeria tightened 84 bps on USDC and 67 on USDT, and Zambia's premium collapsed into a 61 bps discount, while Malawi, Uganda, and Tanzania went the other way, led by Malawi repricing to spreads around 1,975 bps. Ghana pairs a 376 bps premium with a 121 bps single-provider tax: expensive to deliver, and expensive to deliver badly. Watch the monthly premium columns here; this is the region where they actually move.

Asia-Pacific

The quietest delivery surface in the network at a 6.1 bps median spread, and home to the tightest corridors we price: the Philippines delivers at $1M for 2.9 bps all-in. India is the exception that proves the payments-cost frame: FX at the mid, 101 bps of delivery cost from the tax and rail layer, and a B-to-C downgrade on continuity. China and Hong Kong look cheap here at the $10K ticket, and they are, but their small-ticket story is different: the fee gap between routes is worth 80 bps at $500.

Europe

Spreads widened slightly to 15.8 bps, while EUR corridors produced the quarter's biggest coordinated grade improvement: dozens of C-to-B moves on continuity and stability. Europe is turning into the network's most uniform delivery surface. The Eurozone row's 27.2 bps cost is a median across 79 receiving-country corridors; the sub-2 bps rows below it show what the region's tightest routes look like.

Middle East & Türkiye

The quietest table in the report: single-digit delivery costs and premiums hugging interbank. These corridors produced several of the quarter's grade upgrades (TRY, SAR, AED, QAR) and almost none of its drama.

North America

One row: Canada. It behaves like the European majors, cheap and quiet.

12 · When to Transact

There's no timing penalty. If your payout operations still schedule around market hours, you're managing a constraint that no longer exists, except on the short list below.

The evidence first, from 19,261 provider-days of quarter data:

• Weekends price like weekdays. Weekend spreads (98.5 bps) matched weekday spreads (98.7); of the 154 corridors with enough weekend activity to compare, 130 priced weekends within 1 bps.

• Coverage holds around the clock. The network posted as many provider-day quotes per weekend day as per weekday (219 against 219), and median hourly spreads across all 24 UTC hours stayed inside a 0.2 bps band. Delivering at 03:00 UTC on a Sunday priced the same as noon on a Tuesday.

• Month-end isn't an event. Traditional FX treats month-end as a turbulence window; here, the last two days of each month priced 0.7 bps tighter than the rest of the quarter, which is to say flat.

The exceptions are few enough to print, and they aren't random:

With 154 corridors tested, a couple of large gaps would show up by chance, so the rows carry different confidence levels. Mexico and Colombia are solid: statistically strong, on both assets for Colombia, and weekend quoting density doesn't drop, the same lineup simply quotes wider. Nigeria's gaps are large but noisier. Zambia tops the raw penalty list at 88 bps wider but doesn't separate from its own noise on a D-rated corridor, so it stays off the table.

The operational read: for most of the network, send when the payment is ready. Weekend flow into Mexico, Colombia, and Nigeria paid a real premium all quarter, consistent with providers pricing the cost of holding local currency over the weekend, so batching non-urgent weekend payments to a weekday pays there.

13 · What This Means for Payments Providers

The quarter's findings compress into five operating rules.

Optimize the fixed fee at small tickets and the FX rate at large ones. At the median corridor, $500 costs 46.2 bps and $1M costs 26.2, and the difference is the fee layer. Provider choice was worth a median 15 bps more at the $500 ticket than at $1M across fee-covered corridors, which means the small-ticket routing decision is won on whose per-payment fee is lowest, not whose rate is best. A single blended bps number hides which layer to optimize.

Route dynamically or accept the median. The best-priced provider changed hands in 22 corridors this quarter, every 2.6 days on the busiest one, and a single-provider setup pays the median over time. The median cost 23.3 bps more than the best available price. That gap is the measured return on multi-provider routing.

Treat USDC and USDT as separate instruments, corridor by corridor. At the network level the two price 0.4 bps apart. At the corridor level, Peru ran a persistent 99 bps gap and Mexico a 4.5 bps one. A router that treats them as one instrument leaves those gaps on the table daily.

Plan redundancy from grades, before an event. The quarter's largest repricings all landed on corridors outside the redundancy set, where no failover path existed; on the one redundant corridor that repriced, the best route stayed 258 bps inside the median throughout. The D list above is the place to start.

Stop scheduling around market hours. Weekend and weekday pricing matched to within 0.2 bps and provider coverage was flat around the clock. There's no timing penalty left to manage.

All five rules are stated at network medians. The numbers that matter are the ones your flow pays, and those only exist once someone measures them.

14 · Anomalies & Data Quality

Outlier filtering excluded 11,131 records this quarter, 0.4% of the dataset. Honesty here is what makes the rest credible.

Of the excluded records, 8,049 came from a single corridor, USDC→XAF (Cameroon), where one provider's quotes swung beyond the outlier threshold; treat XAF corridor statistics with matching caution. The May 16-17 weekend produced the densest anomaly cluster. The quarter's largest verified daily moves were Malawi's -5.8% on April 9 (a genuine repricing: the filter recognized a new price level and kept it) and Sri Lanka's +4.3% on May 20.

The dataset also contained 32 provider-days where a provider's buy rate sat below its own sell rate, concentrated in the UAE and Brazil. A negative spread isn't a tradable gift; it's a provider updating its buy and sell prices at different speeds, and those records are excluded from spread statistics.

Static-rate corridors, where quotes never moved all quarter, were excluded from spread analysis; Congo (CDF) is the notable case. Nineteen corridors remain precision-limited, meaning their published rates carry too few decimals to measure their spread, and they're excluded from spread rankings rather than reported at false precision. USD-quoted corridors are stablecoin redemption pegs, not FX, and are excluded from spread rankings, the Routing Tax, and Corridor Grades throughout.

15 · Methodology

Every number in this report comes out of one measured pipeline; here's how it works.

All corridor statistics aggregate per-provider period medians (two-stage aggregation), so a provider quoting every minute carries no more weight than one quoting hourly; counts are provider-days. Weekend effects are tested per corridor with a two-sided Mann-Whitney test on provider-day medians, with a qualification floor of 15 weekday and 6 weekend provider-days. Outliers are removed by a Hampel filter (z > 6) with regime-change recovery so genuine repricings survive. Providers whose median mid diverges more than 10% from multi-provider consensus are excluded as parallel-market quotes. Corridors whose source rate precision cannot resolve their spread are excluded from spread-based metrics and disclosed.

Traditional FX comparisons are day-matched: each day's stablecoin consensus against the same day's historical interbank rate (ECB dailies for majors, historical daily sources for frontier currencies), never period-pooled and never latest-snapshot.

Total Delivery Cost is quoted, not executed. Provider quotes are all-in deliverable prices; per-payment rail fees and local taxes come from Borderless fee records (as of 2026-06-24, covering 61 currencies), and corridors without fee coverage are reported as FX-only costs and labeled as such. Real executed transactions validate the quote layer separately through execution alignment. Rails on file count country-rail combinations in those fee records; rails carried are the schemes with completed deliveries in the quarter. Quarter-over-quarter deltas are computed against the prior quarter's persisted period summary, using only corridors reliable in both periods, and both the Year So Far series and the Cost by Country monthly premium columns are built from persisted monthly summaries on the same basis. Full metric definitions live in the Borderless Benchmark methodology.

The full benchmark is at Borderless.xyz/benchmark, where the same measurement can be applied to a single flow: what it pays today, what the best route would have paid, and what the providers it has not integrated were quoting.

About Borderless

Build your own smarter payments network. Three layers, one platform. Connectivity to every provider, the operations to run them, and the intelligence to make every payment smarter.

• 01 — Connectivity, the basics: One integration. Every provider. A single API and standard schema gives you access to the full network of on- and off-ramp providers. Add a new provider in days, not quarters.

• 02 — Platform, control center: The tools to actually run it. A working operations layer on top of the network: reconciliation, exception handling, audit trail, and compliance. The internal tools your team would otherwise build for 12+ months.

• 03 — Intelligence, the brain: A system that gets smarter every payment. Reliability scoring, execution benchmarking, and real-time routing get smarter with every payment across the network. The more the network is used, the smarter every decision gets.

15+ providers · 100+ countries · 63+ currencies · 3-4 weeks to go live

Direct relationships, full routing control, and the freedom to switch providers without re-integrating.