The cheapest way to send money in 2026 (and why your bank hides it)

Every year, people move nearly a trillion dollars across borders to family back home. In 2024 the figure reached $905 billion, up 4.6% from the year before, and $685 billion of that went to low- and middle-income countries where a five-percent fee is not a rounding error but a week of groceries. Despite decades of promises about "disruption," the average cost to send $200 in the third quarter of 2025 was still 6.36%. Send that money through a bank and the average cost was 14.99%.

(Source: The World Bank, Remittance Prices Worldwide, available at http://remittanceprices.worldbank.org.)

Families on the receiving end in those countries can increasingly earn in dollars from home too, through AI training roles that accept applicants where they live: AITraining.jobs, a tracker we maintain, shows how many of those roles are open in each country.

That gap between what a transfer should cost and what most people actually pay is the reason we built the Remittance Rip-off Index. This is what we have learned pricing real corridors, honestly, both legs and both spreads included.

Your bank is the single most expensive option, almost everywhere

If you take one thing from this piece, take this: a bank wire is nearly always the worst way to send money abroad. The 14.99% average is not a worst case, it is the middle of the bank distribution. Banks do not usually charge a headline 15% fee. They charge a modest-looking flat fee, then bury the rest in the exchange rate. You send $1,000, the recipient gets what $850 to $920 of value would buy at the real mid-market rate, and the "$0 fee, low fee!" banner was technically true and completely misleading.

This is the trick the entire industry is built on: the cost lives in the FX margin, where you cannot see it, because you have nothing to compare the rate against. So the first honest number in any comparison is the mid-market rate, the neutral midpoint of the currency's buy and sell price with no margin added. Every real cost is the gap between that rate and what you actually get. We price every path against the same mid so a bank cannot hide behind a "no fee" claim.

The four paths, and how they actually rank

For a given corridor and amount there are really four ways to move the money, and the cheapest one is not fixed. It changes by destination, by amount, and through the day.

1. Bank wire. Slowest and, on average, most expensive (14.99%). Occasionally competitive on very large transfers where a flat fee gets diluted, but rarely the winner.
2. Transfer apps (Wise, Remitly, WorldRemit and peers). The mainstream honest option. Digital-only money transfer operators average 3.54%, and the best app on a given corridor is often better than that. On our India corridor, the cheapest app comes in near 1.1% all-in. This is the baseline everyone should already be beating their bank with.
3. The stablecoin route. Buy a dollar-pegged stablecoin, move it on-chain for cents to a couple of dollars, and cash out to local currency through an exchange in the destination country. On some corridors this is genuinely the cheapest path. On others it quietly loses to a good app once you count everything.
4. Cash and informal channels. Still enormous globally, but opaque, and outside what we price.

The interesting fight is between the best app and the stablecoin route. That is where the hidden money is, and where almost every "crypto is basically free" claim falls apart.

Why "stablecoins are free" is a lie by omission

Here is the pitch you have heard: a USDT transfer costs a fraction of a cent in network fees, so sending money abroad with crypto is "basically free." The on-chain hop really is cheap. On Solana, Polygon, BSC or Arbitrum a USDT transfer costs under one cent. Even the newly expensive Tron network, now the priciest common route because of an energy-fee change, is only around $1.16 to $2.25 per transfer.

But the network fee is the one honest cost in the whole chain, and it is the only one the "free" pitch ever mentions. The real cost is on the two ends:

• The on-ramp (buying the stablecoin). You pay a buy spread of roughly 1% to 4%, only near 1% for a US bank/ACH pro order, more for a card.
• The off-ramp (selling for local currency). This is where it hurts, and it is unpublished, which is exactly why we measure it. The stablecoin does not sell for the mid-market rate. It sells at a premium or discount set by local supply and demand. We measured, against the mid, a +2.96% premium in India, +1.15% in Nigeria, +2.85% in Kenya, roughly flat in the Philippines (+0.01%), and a small discount in Mexico (-0.28%). That premium is a real cost that "free" never counts.
• Cash-out and withdrawal fees. The Philippines charges roughly ₱10 flat plus 1.6% over ₱10,000. India runs ₹5 to ₹10 on IMPS. Small, but real.
• Local tax, which is not optional. India withholds 1% TDS at source on the transfer and taxes any crypto gain at 30%. That is a line item, not a footnote, and the recipient bears it.

Add the two spreads, the network fee, the cash-out and the tax, and the honest stablecoin cost for a typical $500 transfer lands in a 2.5% to 6% range, not a single flattering hero number. Anyone can quote you 0.9% by ignoring the buy spread, the off-ramp premium and the tax. We refuse to. We present the stablecoin path as a low/typical/high range, because that is the truth, and because on some corridors the range's midpoint is higher than the best app.

The risk nobody puts in the fee box

Cost is not the only thing that matters, and this is where we part company with every "just use crypto" thread. The off-ramp carries an enforcement risk that no fee table captures.

In Nigeria, the EFCC has frozen the bank accounts of crypto sellers: 1,146 traders in one sweep, and in September 2024 it froze accounts holding ₦548.6 million (about $330,000) of USDT sellers. A recipient's account can be frozen over funds they had no way of knowing were tainted, from a counterparty they never chose. In India, the Enforcement Directorate has frozen legitimate peer-to-peer sellers' accounts for weeks. These are not fines you can budget for. They are tail risks that can strand the money entirely.

So we show them as a plain-language risk block on the corridors where they apply, not as fine print. If you are sending to Nigeria, the cheapest number on the screen is not the whole decision, and we would rather you know that than convert on our word and get burned.

So what is actually cheapest?

It depends on the corridor, and that is the honest answer the industry avoids because it is harder to market than "always use us."

• USD to India: a top transfer app around 1.1% is hard to beat once you price India's 1% TDS into the stablecoin route. The app usually wins.
• USD to Philippines: the off-ramp premium is roughly zero, so the stablecoin route is genuinely competitive, and a good app is close behind. Either can win depending on the day.
• USD to Nigeria: the stablecoin route can look cheapest on paper, but the account-freeze risk is real and belongs in the decision, not just the fee.
• USD to Mexico: the stablecoin trades at a slight discount, which makes the crypto route attractive, though the best app is only a point or so behind and far simpler.

The winner moves. FX mid-rates update daily; the off-ramp premiums move in real time. A route that is cheapest at breakfast can lose by lunch. That is precisely why a one-time "here is the best way" article, including this one, is not enough, and why the live router exists: it re-prices your exact corridor and amount against all four paths continuously, and shows you the one that wins right now, with the receipts.

The honest bottom line

Stop using your bank for international transfers. That single change beats the 14.99% average for almost everyone reading this. Beyond that, compare every path against the real mid-market rate, count both spreads and the local tax, and treat the enforcement risk as part of the cost on the corridors where it is real. Sometimes the answer is a stablecoin. Often it is a good app. The point is never to sell you one route. The point is to price all of them the same honest way, so the cheapest path is the true cheapest path, and you can see exactly why.

Check your own corridor on the live router, and see the full Remittance Rip-off Index for what your bank overcharges, ranked worst to best.