Ether.fi Cash vs Tria Card: I hold both, so I measured both

The short version

One line decides it for most people in Europe, and it is the first one. Everything after that is preference.

What one euro costs

Both of these are now measured, by the same method, against the same benchmark. That was not true when this guide first went up: Ether.fi’s side rested on its help centre, which put the zero-FX promise in writing on 24 June 2026 and still calls the feature a beta.

Ether.fi charges nothing on euro purchases, and a live purchase says the same. A €138.00 charge on 4 August 2026 settled at $158.882436, an effective 1.151322 dollars per euro against a central bank reference of 1.1515 for that day. That lands 0.015% below the reference, which is not a discount and not a markup, just the noise you get from comparing a purchase at 09:14 with a rate published in the afternoon. Anything billed in a third currency is a different matter: roughly 1%, and on Core a further half point can ride on top, because the added margin is set by membership tier rather than by currency. The card account behind it also remains dollar-denominated, with native euro support promised rather than shipped.

Tria I measured the same way, and it does not land in the same place. A one euro item on 28 July 2026 came off the statement at $1.16, which works out to 2.64% above interbank once you divide by what the euro was actually worth that day. Two fees are named in the app and they total 1.50%. The other 1.12 points are in the rate itself: 1.1494 dollars per euro where the central bank reference said 1.1367.

Worth pausing on the smaller of those two named fees. Tria’s own front page advertises “0% Tria FX Fees” next to “1% VISA FX Fees”, and its card terms quote foreign exchange at up to 3%. The fee screen on my transaction charged 0.50%. Three numbers from the same company, and the one that came out of my account is the one that appears on none of its published pages.

That component is not a line item on any document. Reading the fee schedule harder would not have surfaced it, which is the whole reason we started buying things instead of only reading. There is a longer version of that argument in our guide to hidden FX fees.

Credit where it is due: very few cards break the charge down at all, and Tria does. It shows the rate it used and the fees it took. The shortfall only surfaces if you go and look up what the euro was worth independently, which almost nobody does. Honest presentation, incomplete arithmetic.

One more line that matters at small amounts. The app shows a $0.15 per-transaction convenience fee on the lower tiers, waived on Premium, and I can find it on no Tria page either. On a one euro purchase that single fee would have been 13%.

Cashback: confirmed against deferred

Ether.fi paid $0.15 on a $5.00 charge, which is 3% to the cent, settled and visible in the transaction detail. You choose whether it arrives as USDC or wETH. Captured 27 July 2026.

A second purchase a week later held the same rate at a size worth caring about: a €138.00 charge paid back €4.14, again 3.00% to the cent, and the on-chain record puts it at $4.766473 on a $158.882436 spend. Worth saying why that is not a surprise. The euro ladder below taper starts at the same top rate, and this purchase sat under the threshold where it bends.

Read the ladder before you plan around 3%, though. On Core the top rate covers your first $2,000 in a month; the next thousand earns 1% and everything past $3,000 earns half a percent. Luxe and Pinnacle stretch that top band to $10,000 and $50,000 respectively. Euros run on a tighter scale that starts tapering at €800, which is how the 0% euro FX gets paid for. And the tier is rented, never owned: the counter resets with the calendar, so Luxe has to be re-earned monthly at 10,000 points unless you stake or hold 15K ETHFI in the vault and bypass it. A balance sitting in an outside wallet does not count.

Tria promises more and settles later. Rates go 1.5%, 4.5% and 6% as you move from the virtual card up through Signature to Premium, and each rate has its own monthly ceiling: $100, $1,000 and $2,000 of spend. Past that ceiling the two paid tiers fall to 1% and the virtual card falls to half a percent. Payment lands in USDC or USDT in your own wallet, whichever of the two Tria picks, and its own Season 3 post says the rail in practice is USDT on Arbitrum. Two clauses in the terms matter more than the rate: payouts may arrive as much as twelve months after the period they were earned in, and closing your account before they arrive means losing them.

Tria’s help centre contradicts Tria’s terms on this, and the help centre is the more recently updated of the two. It says cashback earnings are currently uncapped. The terms and the company’s own product announcement both publish the ceilings above. When an issuer’s two documents disagree, the one you agreed to is the one that governs.

In practice my all-time counter read $0 for months. The first payout landed on the euro purchase above: six cents. At that size a cent of rounding cannot separate 5% from the advertised 6%, so I log it as proof the mechanism pays, not as confirmation of the rate. It is the one measurement in our log that stayed inconclusive.

Yield: on the balance or beside it

This is the structural difference, and it is bigger than the rates.

On Ether.fi the balance the card spends from is the balance that earns. In Direct Pay mode a purchase is deducted straight from the vault, out of the USDC or the yield-bearing LiquidUSD position, so the money works right up to the moment it leaves. Which of the two goes first the help centre does not say. Borrow mode does the opposite trick, drawing credit against collateral at 4% APY so the collateral never has to be sold. Either way there is nothing to move.

On Tria the card is a 0% APR credit product with collateral posted per dollar charged, and yield lives in separate self-custodial Earn strategies the card does not touch. To earn on money you also want to spend, you move it, and moving it costs.

The cost is measurable. A deposit of 1,000.25 AUSD made on 24 January was worth 1,015.54 when I closed it on 21 May: 1.53% across 117 days, or roughly 4.8% a year, against a tile in the app quoting up to 22.17% on the very same pool. Slippage on the way in and the way out ate most of the difference, and because that toll is close to fixed, a smaller deposit would have kept even less of it. Measured 28 July 2026.

What the cards refuse

Ether.fi will not pay for alcohol or tobacco. The first refusal I put down to a bad terminal. It happened again elsewhere, then in another country, and by the fourth time the pattern was obvious: the shops had nothing in common except selling something age-restricted. The email that follows a decline says so outright and points at a help article about high-risk categories.

Everything else about the card works, which is exactly why this belongs in a comparison. It is one narrow blind spot in a product that otherwise behaves, and no fee table anywhere would warn you about it. Carry something else if your spending runs through off-licences and tobacconists.

I have not reproduced anything equivalent on Tria. That is not the same as saying Tria has no restricted categories, only that I did not find one.

What happens if you get frozen

In May, Ether.fi suspended my card services on suspicion of cashback farming, which I was not doing. Card gone, bank transfers gone, Pix gone. Support reviewed it and restored the account in under 24 hours; the only lasting cost was reissuing the card.

The part worth carrying into any self-custody comparison is what happened to the money. Nothing. Withdrawals stayed open on-chain the entire time, so even if the appeal had failed, the funds would not have been trapped. That is the practical difference between a card sitting on your own vault and a card sitting on someone’s balance sheet, and it is very hard to learn without it happening to you.

Tria I have not had frozen, so I have nothing first-hand to compare against.

The app and the humans behind it

Tria takes this one, and not by a small margin. It is the best-built app I have used in this category: fast, coherent, and it does more than the card. Write to support and a human writes back, quickly, having read what you sent.

The plastic is where it falls down. Mine was paid for on 24 January and six months on it still has not arrived in Europe. Shipping opens one country at a time and that list is short. You can order the card; whether it gets to you depends entirely on where you live.

Ether.fi is plastic on Core and metal above it, though Core does not include a physical card at all: it costs $50, and free shipping starts at Luxe. The tier ladder is at least visible in the app instead of buried three menus deep. Its rough edge is the mirror image of Tria’s: the product works, the presentation asks more of you.

Which one to carry

Ether.fi if you spend in euros or dollars. Two measured euro purchases, 0.015% under the reference against 2.64% over it, is not a preference. It is a fee, and it recurs on every purchase. Add cashback that pays immediately and a balance that earns until you spend it, and the money side is settled. Accept that you will occasionally be declined somewhere that sells cigarettes, and check that your country is not one of the excluded ones.

Tria if Ether.fi will not have you, or if the app is what you actually live in. It covers 150+ countries, excludes the United States, and gives you the better daily experience by a distance. Read the cashback caps before believing the 6%, treat Earn as a separate decision from the card, and expect the euro rate to cost you about two and a half points more than the alternative.

Both are genuinely self-custodial and both are Visa. Between two cards that agree on the important architecture, the statement is what separates them, and the statement took six months and about two hundred dollars to read.