Borderless Benchmark Report

Executive Summary

2026 has ushered in the beginning of enterprise stablecoin adoption. In the year since the GENIUS Act became law, we have seen billions of dollars of stablecoin M&A, the two largest remittance companies each launch a dollar stablecoin, and more than 140 companies, Visa and Mastercard among them, come together behind Open USD. As this report goes to print, Samsung has announced stablecoin transfers in Samsung Wallet for 82 million eligible Galaxy devices, with payouts to bank accounts in more than 60 countries, Bastion as the regulated institution and Borderless as the network of local-rail providers behind it.

The year in three quarters

Q1 First quarter

• Tether launches USAT, a US-regulated dollar stablecoin.

• Stripe's Tempo chain goes live with Visa and MoneyGram among its validators.

• Mastercard agrees to buy BVNK for up to $1.8 billion.

Q2 Second quarter

• Western Union launches USDPT and MoneyGram launches MGUSD.

• Thirty-seven European banks join a consortium to issue a euro stablecoin, and Japan's three largest banks announce a joint yen stablecoin.

• Visa, Mastercard, Stripe, Coinbase and Shopify announce Open USD, a shared stablecoin.

Q3 Third quarter

• Mastercard closes the BVNK purchase.

• Twenty-one of the world's largest banks announce a joint dollar stablecoin.

• Circle agrees to buy Tazapay for $400 million and launches its Arc chain.

• Visa reports stablecoin settlement above $20 billion a year, fifteen times the year before, and Open USD goes live with more than 140 partners.

For payments companies, the question has moved from whether to use stablecoin rails to where, with which providers, and how to maximize margin, speed and reliability. The data to drive those decisions is fragmented across providers, each of which sees only its own rails and flows, creating a partial picture. The Borderless network has grown significantly on both sides this year, generating more and more data with each transaction processed. This drives the intelligence layer, helping companies improve the margin, deliverability and speed of their stablecoin-funded payments.

This report is a high-level look at that data, which lives for customers directly in the platform and drives detailed insights for their stablecoin payment strategies. This quarter, the benchmark adds two new dimensions, both in early beta: reliability and settlement speed.

Cost

Delivering a stablecoin into a local bank account is priced like delivering a dollar, with a spread in the exchange rate plus a flat fee for the local rail. At the median provider, delivering $10,000 cost $52, or 52 basis points (bps), across every corridor with a price on file. A basis point is a hundredth of a percent.

Total delivery cost by region · $10,000, median provider, bps

The vertical rule marks the all-corridors median of 52 bps.

Stablecoins versus interbank. Across the 57 currencies with a same-day interbank rate, the Stablecoin Mid sat a median 4 bps below interbank this quarter. It was within a few basis points of interbank in every region but Africa. It sits well above interbank in Congo, Argentina, the CFA franc zone, Botswana and Ghana. That gap is the parallel market. A recipient there gets 4 to 33% more local currency than the official rate would give.

The gap between the median and the best price. Where more than one provider priced a currency, routing to the best-priced provider improved margin by 43 bps over the median provider this quarter, or $4,340 per $1 million. That is up from 32 bps in the second quarter. The gap widens to 56 bps in Africa, where the choice of provider decides the margin.

Lead-Provider Churn. Sixteen of the 18 currencies priced by more than one provider changed their best-priced provider at least once this quarter, up from 13 of 17 in the second. Colombia's best-priced provider changed 21 times in 92 days, Brazil's and Chile's 14 times and Mexico's 13. In Rwanda and the CFA franc zone it barely moved, and in the eurozone it changed once.

Reliability Beta

On the most active corridors, providers deliver 97.6 to 98.6% of payouts and fewer than 1.5% fail. Across all providers' rails, deliverability ranges from 99% on the best to 73% on the least reliable.

The best and worst rails. Mexico's SPEI delivered 98.6% of payouts and Brazil's PIX 97.8%. At the other end, Costa Rica's SINPE delivered 73% and Colombia's PSE 79%.

Downtime. The second quarter was the worst for reliability, with about 170 service-affecting incidents at providers against about 30 in the third. A typical incident lasted an hour, but the long ones added up. One provider's rails had an incident open for a sixth of the quarter, and two currencies lost their only rail at a provider for three days each.

Failures versus returns. In Colombia, 11.8% of payouts were rejected by the bank rails and another 11% went out and came back. In Costa Rica a quarter of payouts came back after being sent. Rejections are the rail's problem. Returns usually trace to the beneficiary details the payment carried.

Speed Beta

A stablecoin payment is only as fast as the local rail that delivers it. Payouts settle in minutes on instant rails and the next business day on batch rails. Over SWIFT they take one to three business days.

By rail. PIX, SPEI and PSE settled a typical payout in 2 to 3 minutes. US ACH payouts land the next business day, T+1. A typical SWIFT payout settled on T+1 and the slow tenth on T+3.

In the tail. A typical SEPA payout lands in 10 minutes, but one in ten takes most of a working day. In Israel one in ten takes 19 hours.

Where stablecoins are being used

Flow across the network nearly doubled from the second quarter to the third. It falls into three uses.

01 B2B cross-border payments

Transfers of $10,000 or more are a minority of transactions but carry most of the flow, and demand is growing fastest in Latin America and Asia.

02 Global payouts

Transfers under $2,000 make up most transfers but a minority of flow, and demand is growing fastest in Latin America, Africa and Türkiye.

03 Treasury

Businesses move dollars between bank accounts and stablecoin as working capital. Funds can be deployed and repatriated instantly, so there is no need to prefund.

Does the stablecoin matter? Does the chain?

The coin barely changes the price, and the chain decides where you can send.

Providers that offer both coins priced a USDT payout below a USDC payout on the same corridor and day, by a median 6.6 bps in Europe, 5.4 bps in Asia and 3.4 bps in Latin America. In every African currency the difference was 0.0 bps, because providers there hold both coins at par. USDT has traded a median 3.5 bps below USDC this year and within 10 bps of par on 86% of days. It dipped to 33 bps below in January and spiked briefly above par in April, May and September.

The chain decides reach · of the network's 72 currencies and 132 local rails

Reach through the providers that support each chain, according to the providers' own documentation. Providers that publish no chain list are not counted. Plasma, Avalanche, Monad, TON, Algorand, HyperEVM, Linea, XDC, Aptos and Arc are also supported, with reach not yet measured.

Benchmark: Cost

Stablecoin delivery costs less than the traditional alternative at every ticket size, and the choice of provider is worth most in Africa.

The named metrics

The Stablecoin Mid — A credibly neutral benchmark rate for stablecoins into local currency, administered by Borderless from the network's data: the midpoints of multiple providers' prices for each corridor, aggregated the way a composite mid-market rate is built from many venues. The other metrics use it as their reference.

Total Delivery Cost — The all-in cost of delivering a payment, which is a provider's rate against the Stablecoin Mid plus that provider's rail fee and the corridor's local tax. It is always stated for a given ticket size, because flat fees weigh more on small payments.

The Interbank Gap — The Stablecoin Mid against the same day's interbank mid, per currency. Zero means the stablecoin dollar is priced at par. A positive gap is the parallel-market premium. The recipient gets more local currency than the official rate gives.

The Routing Gain — The median provider's cost minus the best provider's, for each corridor on each day that more than one provider prices it. It is the margin a business gains by routing instead of using one provider.

Lead-Provider Churn — How often the best-priced provider changed hands, counting only changes of 5 bps or more.

Headline metrics · third quarter against second

Stablecoins versus the traditional rails

The comparison uses the interbank rate for the exchange rate itself, the World Bank's published cost of a $200 remittance for retail tickets, and a business wire for large ones.

Stablecoins price within a few basis points of interbank in most markets. Across the 57 currencies with a same-day interbank rate, the Stablecoin Mid sat a median 4 bps below the interbank mid this quarter. By region it sat 5 bps below in Latin America, 2 in Europe, 6 in Asia and 5 in Oceania. The Stablecoin Mid and the interbank mid are both benchmark rates. An actual payout, once the provider's spread and the rail fee are included, lands about 0.5% below the interbank rate, and a bank's business rate about 1% below it.

Parallel markets. In some countries the government sets an official exchange rate and the market trades at a different one. The difference shows up here as a large positive Interbank Gap. These premiums are why Africa's median sits 92 bps above interbank. In these markets a stablecoin pays the recipient the dollar's actual local value instead of the government peg.

The Interbank Gap in parallel markets · Stablecoin Mid above the official rate, end of quarter

Retail remittances cost a fraction of the traditional price. Delivering $200, the World Bank's standard remittance ticket, cost 1.0%, or $2.00, at the median provider. The World Bank put the traditional cost at 6.36%, or $12.72, in the third quarter of 2025. The improvement, especially into emerging markets, comes primarily from removing intermediaries that each add their own margin on top of the rate, and from the working-capital efficiency of reducing the costs associated with prefunding.

Retail remittance · $200 delivered

Business payments cost less, settle faster, and need no prefunding. The median provider priced a $10,000 payout at $52. In the eurozone the median provider priced it at $38 and the best route at $26. At the large US banks, an international wire carries a published fee of $35 to $50 before any exchange-rate margin. A 1% margin, the low end of what studies of small-business wires find, adds $100 on the same $10,000. At $100,000, the median provider priced delivery at 50 bps, or $505, against about $1,050 for the wire.

Cost to deliver a $10,000 business payment

Business payment cost by region and ticket size

Business payments are also moving onto these rails for speed and to end prefunding. A wire crosses one or more correspondent banks, and each takes a margin. It also requires prefunded balances in every market a business pays into. This quarter a typical wire settled the next business day and the slow tenth took three. A stablecoin-funded payout settles into a local account in minutes on instant rails, with one rate and one fee known before it is sent. A single dollar balance can fund any corridor and come back the same day, instead of sitting prefunded in each one.

Where cost is compressing and where it is expanding

Prices moved all quarter. Nigeria, the CFA franc and Botswana repriced by 70 to 130 bps against the official rate. Ghana's providers moved hundreds of basis points apart, and Bolivia's corridor reset to a new level in a day. In September the best-priced provider changed hands on 15 of the 16 currencies that more than one provider priced all year, up from 8 in July.

Expanding: official rates moved faster than providers. Nigeria's premium rose from 108 bps in Q2 to 242 in Q3 and 317 in September. The interbank rate moved about 2% in late August, and payout prices followed. The Central African CFA franc's premium rose by 128 bps, to 508, and Botswana's by 73, to 439. None of these moves came from the stablecoin layer.

Compressing: Southern and East Africa. Quarter on quarter, Zambia's cost to deliver $10,000 fell by 138 bps, to 265. Botswana's fell by 76, to 303, and Kenya's by 22, to 56. Tanzania's premium narrowed by 22 bps. Ghana went the other way. The median provider's cost rose from 284 bps to 350 while the best route held near 163. The gap between its providers widened to 187 bps.

Bolivia moved to the parallel rate. Through June the boliviano was priced on the network at the official rate, 6.91 to the dollar. On July 22 pricing moved to the parallel market in a single day, to 10.9, and reached 12.0 by September 30. That is 74% above the official rate, the largest parallel-market premium in Latin America. The parallel rate itself swung 10% in a week in late September, and the network's pricing swung with it.

Lead changes, the Interbank Gap and the Routing Gain, month by month

* From February the lead-change and Routing Gain columns are measured on the 16 currencies priced by more than one provider in every month. January predates the corridors added in February, so its row covers all currencies.

Where providers disagree and where they agree

In some markets providers price within a few basis points of each other and the best price changes hands constantly. In others the gap between providers is wide and the leader rarely changes, so the choice of provider decides the margin.

Provider agreement by region

Africa: structural gaps, stable leaders. Ghana's median-to-best gap was 187 bps, Rwanda's 117, Tanzania's 105 and the two CFA francs' 42 and 56. In Rwanda and the CFA franc zone the same provider was best nearly all quarter.

Latin America: small gaps, constant churn. In Brazil the gap between providers was 28 bps and the best price changed hands 14 times in 92 days. Mexico's gap was 28 bps with 13 changes, Chile's 29 bps with 14, and Peru's 36 bps with 11. Colombia, the busiest corridor on the network, had a 90 bps gap and a new leader every four days. Argentina looks more like Africa, with a 160 bps gap and two lead changes.

Europe: consensus. Across the eurozone the gap between the median and the best provider was 12 bps, and the leader changed once in the quarter.

September: a second provider on 18 more currencies. In September a second provider began pricing the Australian, Canadian, Hong Kong and Singapore dollars and six more Asian currencies, the rupee, rupiah, won, dong, renminbi and Philippine peso. In Europe, Türkiye and the UAE it added sterling, the Swiss franc, the Danish and Norwegian kroner, the zloty, the leu, the lira and the dirham. On every one of them the incumbent was best every day. The gap between the two ran from 8 bps in Singapore to 57 in Türkiye, with a median of 32. These figures leave out the won as an outlier, because the second provider's rate there looks stale rather than priced.

Cost by Country

One row per currency, grouped by region. The Stablecoin Mid and the interbank rate are the September 30 values in local currency per dollar. Median cost is the median provider's all-in cost to deliver $10,000 over the quarter, across the currency's corridors. The Routing Gain is the gap between the median provider and the best route on the days when more than one provider priced the currency. A † marks currencies whose second provider arrived during the quarter, measured on the days it was there. The gap columns are the Interbank Gap. Positive means the stablecoin fetches more local currency than the official rate. The euro row covers 20 countries, the West African CFA franc 8 and the Central African CFA franc 6. These are medians, so no single payment is priced at exactly these figures.

Cost by Country · September 30 values

Benchmark: Speed Beta

A stablecoin payment is only as fast as its local rail, and a payment that lands in two minutes is a different product from one that lands in two days. This section measures how fast each local rail settled completed transfers this quarter. This benchmark is new this quarter and in early beta. Figures are printed only where there is enough data behind them.

The named metric

Settlement Speed — The time from a completed transfer's creation to its final status, per rail and direction. It is reported as the median (p50), the 90th percentile (p90), and the share settled within one hour and within 24 hours. ACH payouts and SWIFT are reported in business days, where T+1 means the business day after the payment is created. It is reported only where a rail carried statistically significant volume in the period. Timing is measured on this platform. It includes the provider's processing and the rail's own clearing, but not the customer's funding leg.

How fast each rail settles

Instant rails settle a payout in minutes, almost always inside the hour. Batch rails settle the next business day, and correspondent banking takes one to three business days.

Settlement speed by rail, third quarter · time from creation to final status, log scale

“· in” marks collection rails; every other rail is a payout.

Settlement speed by rail · full table

On ACH the provider's final status is the same-day hand-off to the batch, and the funds land on T+1.

Instant rails settle in minutes. PIX, SPEI and PSE delivered a typical payout in 2 to 3 minutes and 96 to 98% of payouts inside the hour. On these corridors the speed depends on the rail rather than the provider.

Most bank rails settle inside the hour, and the difference between them is in the slow tenth. South Africa's EFT, the UAE's NPSS, East African mobile money, Australia's BECS and the generic local bank transfer all settled a typical payout in 5 to 10 minutes, with 89 to 93% inside the hour. The slow tenth waited 20 minutes on BECS and 65 on NPSS.

SEPA and Israel have long tails. A typical SEPA payout lands in 10 minutes, but one in five takes more than an hour and one in ten takes most of a working day. Israel's rail behaves the same way, with one in ten waiting 19 hours. On Peru's CCE one in ten waits 5.5 hours.

Batch rails settle next day and SWIFT in days. A typical SWIFT payout settled on T+1, a quarter took more than two business days, and the slow tenth settled on T+3. Where a corridor has both an instant local rail and SWIFT, the choice of rail can change delivery by two days.

Rails got faster this year. The slow tenth on Australia's BECS fell from 11.9 hours in the second quarter to 20 minutes, and on Israel's rail from 43 hours to 19. The UAE's fell from 90 minutes to 65, and PIX's halved to 5 minutes. The data shows the improvements but not what caused them.

Benchmark: Reliability Beta

Price is the first thing a payments team optimizes for, and while providers' rate sheets are based on different rates, the Stablecoin Mid compares them against the same reference. Whether the payment arrives is harder to measure and harder to compare. Deliverability depends on the provider and the rail together, and the biggest lever is a second provider to fail over to. This benchmark is new this quarter and in early beta. Figures are printed only where there is enough data behind them.

The named metrics

Completion Rate — The share of created transfers that reached completion, by rail and by currency, with failures reported separately from cancellations and refunds.

Incident Hours — Service-affecting incidents at providers, measured by how often they happen, how long a typical one lasts, and the hours a provider's rails had an incident open. Overlapping incidents are merged, and maintenance windows and stale monitors are excluded.

Failover Coverage — Corridors and currencies with two or more providers covering the rail.

How often payments complete

Across providers' rails, deliverability ranged from 99% on the best to 73% on the least reliable this quarter. On the most active corridors it ran 97.6 to 98.6%, with failures below 1.5%.

Completion by rail, third quarter

Where two providers serve a rail the figure is a range; the bar shows the lower bound and the range is printed beside it.

Completion by rail · full table

The most active corridors deliver. Mexico, Brazil, South Africa, Australia, the UK and the UAE each completed 97.6 to 98.4% of payouts. Brazil's PIX improved from 93.5% to 97.8%, and the eurozone from 65% to 87%.

Two providers on one rail can be nine points apart. South Africa's EFT completed 89.8% of payouts with one provider and 97.7% with the other. SEPA completed 80.3% with one and 89.7% with the other. A business locked into the weaker provider loses eight to nine points of deliverability. A business that can route to either gets the better number on every payment.

Colombia is the least reliable large market. 77% of Colombian payouts completed and 11.8% failed, the highest failure rate of any large currency. Colombia's bank rails logged 35 major incidents in the second quarter alone, almost all at individual banks and on the PSE and Bre-B schemes. The failures sit in the banking rail. The stablecoin leg plays no part in them.

Returns are a different problem from failures. A failed payment was rejected by the rail. A returned payment went out and came back, usually because of the beneficiary details it carried. They need different fixes, so they are reported separately. Costa Rica's rail completed 73% with only 2.5% failed. A quarter of its payouts went out and came back. Peru's CCE, Israel's ZAHAV and Chile's high-value rail behave the same way, with 12 to 16% of payouts returned and under 1% failed. Colombia's shortfall splits evenly between rejections and returns.

Failures are concentrated in a few currencies. The highest failure rates were Colombia at 11.8%, USD payouts at 10.4%, Singapore at 8.1%, Thailand at 6.5%, the eurozone at 4.3%, Korea at 3.3% and Ghana at 3.1%. Everything else was under 3%, and the most active currencies, Colombia aside, were at or under 1.5%.

Incidents and downtime

The second quarter was the worst for reliability, with about 170 service-affecting incidents at providers against about 30 in the third. The typical incident lasted an hour. Five providers had more than 24 hours of cumulative downtime on their rails in the second quarter.

Provider downtime, second quarter

No provider is named next to any benchmark figure. Providers A to E are anonymized.

Downtime is concentrated in a few providers, and where it occurs it is long. Two providers accounted for most of the hours. Provider A's rails had an incident open for 354 hours, a sixth of the quarter, including one stretch of nine and a half days. Provider B's Colombian bank rails had one open for 297 hours across 71 incidents. Two more providers each had a single currency switched off for three days. At least one provider on the network had an incident open for 37% of the quarter.

Frequency and duration are different risks. The Colombian rails had an incident most working days and cleared each in about two hours. Provider A had one every four days and cleared most in half an hour, until the one that took nine and a half days. The short, frequent kind costs retries and support tickets. The long kind takes the corridor down.

This is the case for a second provider on the rail. A business with one provider on East African mobile money lost the equivalent of two weeks of the quarter. A business with two providers on the rail could route around it. The three-day currency outages worked the same way. A second provider was the only way to deliver those currencies on those days.

Failover: a second provider on the rail

Corridors with two or more providers on the rail rose from 78 at the end of the second quarter to 148 in September. Currencies with a second provider rose from 17 to 36. A second provider came onto the rail for sterling, the Swiss franc, the Canadian and Australian dollars, the Nordic and Central European currencies, Türkiye, Bolivia and the Dominican Republic, and in Asia for India, China, Korea, Singapore, Indonesia, the Philippines and Vietnam.

Deliverability comes from the second provider. Failures cluster by provider and rail rather than by country. Colombia's 11.8% failure rate sits on specific bank rails, and the second quarter's Incident Hours fell on two providers.

On the majors the second provider is for the bad days. On rails where completion runs above 97% and a typical payout lands in minutes, a second provider still pays for itself on the bad days. Those are the week a rail's slow tail stretches from minutes to hours, and the three days a currency's only provider has it switched off.

Completion and reach by currency

Payouts completed and failed, by currency

Currencies and rails without statistically significant volume are excluded.

Building an enterprise-grade stablecoin program

By Kevin Lehtiniitty, Borderless.xyz

Real-time local rails now exist in most of the world's large markets, and a stablecoin is an asset that can move 24 hours a day, every day of the year. Bridging the two means a payment can be funded and delivered at any hour, which removes most of the capital that cross-border payments have always tied up in prefunding. That capital has a cost, and holding it in a local market carries FX risk, or the cost of hedging that risk away. For SME and mid-market payments companies, this greatly simplifies the business: with a stablecoin balance you can pay out to bank accounts across all of these markets without complex treasury management. For an enterprise, stablecoins open access to more liquidity. And with wallets as the new bank accounts, a business holds one centralized balance that it can deploy and repatriate, pay out from and collect into, in real time from anywhere in the world.

Analyzing all the payments and pricing data we have seen this year, we find the result is faster and cheaper than the traditional route. A $200 remittance cost 1.0% at the median provider against the World Bank's 6.36%. A $10,000 business payment cost $52 against about $135 to $150 for a wire. A payout on an instant local rail landed in two to three minutes. A wire took one business day for a typical payment and three for the slowest tenth.

Stablecoin route vs traditional route · 2026 year to date

Almost everyone starts with a single provider, often an aggregator, because it is easy, with one contract and one integration covering every corridor on day one. That simplicity has three costs. The aggregator marks up the underlying provider's price and keeps the margin the rails would have given you. The counterparty underneath is unknown to you, so you carry risk you cannot see or diligence. The aggregator mixes your flows with every other client's, so one high-risk client can put the aggregator's bank relationships at risk, and your payments with them.

Once you have outgrown a single provider, enterprise-grade means a network of providers. Four design decisions matter most.

99% → 73% - Deliverability, best rail to least reliable

Work with local providers in each region you serve. Deliverability ran from 99% on the best rails to 73% on the least reliable this quarter, and the provider and the rail together set that difference. Colombia's failures sit in specific bank rails. Costa Rica's returns trace to the beneficiary details the payment carried. A provider who knows the market fixes both. A generalist does neither.

43 bps - Routing gain over the median provider, or $4,340 per $1M

Improve margin with more than one provider per corridor. The best-priced provider on a corridor keeps changing. It changed on 16 of the 18 currencies priced by more than one provider this quarter, 21 times in 92 days in Colombia and 14 times in Brazil. Routing each payment to the best-priced provider was worth 43 bps over the median provider this quarter, or $4,340 on every million. In every month since February it was worth between 28 and 55 bps. A business on one provider gets the best price for a week and the median price after that. In Africa the gap between the median and the best provider was 56 bps, so the same decision is worth more there.

5 providers - More than a day of cumulative downtime in Q2

Increase reliability with a second provider to fail over to. Five providers had more than a day of cumulative downtime on their rails in the second quarter. One had an incident open for a sixth of the quarter, including a stretch of nine and a half days. Two currencies lost their only rail at a provider for three days each. A business with one provider on those rails lost those days. A business with two could route around them.

Stablecoin Mid - The neutral reference rate

Benchmark against the market, not just your own transactions. None of these decisions is made once. The best-priced provider kept changing, Bolivia repriced to the parallel market in a day, and the downtime moved from one provider to another. Your own payments cannot tell you whether your rate is good, whether your failure rate is normal, or whether a provider is down for everyone or only for you. That takes a data set that spans providers and corridors, and a neutral reference rate that puts every provider's price on the same scale. The Stablecoin Mid is that reference. Each month, measure your Total Delivery Cost against it, and track deliverability by provider and rail, with failures kept apart from returns.

Methodology and data quality

Sources. Provider pricing was observed across the Borderless network from January 1 to September 30, 2026. Quarterly figures cover July 1 to September 30 unless stated. Transfer data covers completed and attempted transfers by originators on the platform over the same period. Incidents are those providers reported in the platform's incident record. Rail fees and local taxes are the published figures on file as of October 8, 2026. Traditional remittance costs are the World Bank's Remittance Prices Worldwide, Issue 54. The wire comparison uses published outgoing wire fees of $35 to $50 at large US banks and a 1% exchange-rate margin. That margin is the low end of what published studies find banks charge small and mid-sized businesses.

How prices are measured. A corridor is a currency into a receiving country, with USDC and USDT counted once. Before any corridor statistic is computed, each provider's observations are reduced to that provider's median, so a provider that updates every minute carries no more weight than one that updates hourly. Each provider's prices are sorted by what they represent, and the cost metrics use each provider's published delivery price. A few providers charge part of their margin at execution, so the best route can read lower than the price a payment settles at.

The Stablecoin Mid. Borderless administers and publishes the Stablecoin Mid from the network's observations. The Q2 2026 edition published it, with the same definition, as the Borderless Mid. For each corridor and day, it is the median of the providers' own mid-rates. The period figure is the median of the daily values. Every contributing provider carries equal weight, outliers are filtered before aggregation, and the window is the full UTC day.

Total Delivery Cost. A provider's rate against the Stablecoin Mid plus that provider's published rail fee and the corridor's statutory local tax, in basis points of a stated ticket. A fee applies only to that provider's route. A tax applies to every route in the currency. The best route is the provider with the lowest all-in cost each day. Where no provider publishes a deliverable price, cost is measured on completed transfers instead. For each day, it is the median of the rate recipients received against that day's interbank rate, using only tickets large enough that fixed fees do not dominate. Headline and regional figures are medians at $10,000 across every corridor with a price on file.

The Routing Gain and the Interbank Gap. The Routing Gain was published as the Routing Tax in the Q2 edition. On each day that more than one provider priced a corridor, it is the median provider's cost minus the best provider's. The corridor figure is the median of those days, and the currency figure is the median of its corridors, with the euro counted once. The headline is the median across currencies that had a choice of provider on at least half the quarter's days. Lead-Provider Churn uses the same currencies. The Interbank Gap was published as the Parity Gap in the Q2 edition. It compares each day's Stablecoin Mid with that day's interbank mid, takes the median of the daily values for each currency, then takes the median across currencies. It compares mid with mid. No payout or collection price is measured against interbank. Interbank rates are ECB reference rates for the majors and historical daily series for frontier currencies, matched by day. Currencies without coverage for the same period are excluded. Parallel-market currencies are flagged and read separately.

Cleaning. Outliers are removed with a Hampel filter that recognizes sustained regime changes, so Bolivia's July reset was kept as a new level. Four kinds of data are excluded from the affected metrics. They are providers that diverge more than 10% from the Stablecoin Mid, corridors whose rate precision cannot resolve their spread, static-rate corridors and USD-denominated corridors. Provider feeds that stop updating are excluded by rule. Outlier filtering removed 5.8% of records this quarter, almost all from one provider's euro pricing.

Speed and reliability. Settlement Speed and Completion Rate are computed on transfers from originators, excluding internal and test accounts. Currencies and rails without statistically significant volume are excluded.

Comparisons and anonymization. Quarter-on-quarter changes compare against the prior quarter recalculated with the current fee records and method, using only corridors with reliable data in both periods. Payout prices for USDC and USDT are compared only where the same provider priced both on the same corridor and day. Open-market prices for the two coins are Binance.US daily closes. Chain coverage is what providers' own documentation says is possible, counted against the network's 72 currencies and 132 local rails. No provider is named next to any benchmark figure, and provider counts and transfer volumes are not printed. Where two providers serve a rail, the figure is shown as a range or the pair is anonymized. Usage figures are shares of this network's completed flow, and growth indices set the prior quarter at 100. They describe this network only.