The cheapest Lexar on my bench is also the fastest

Why an entry-tier card got the full protocol

Lexar sells the Blue as its cheap line. It is the card you get when you sort by price and stop reading, and the packaging is written to match: one speed figure, a row of badges, and a footnote admitting that write speeds are lower without ever saying lower than what.

I bought it for a comparison rather than for itself. Lexar's Professional Silver Plus 128GB is the same form factor, the same bus and the same class marks one tier up, and it advertises 205 MB/s read and 150 MB/s write against the Blue's 160 and nothing. Two cards from one brand, one rung apart, is the cleanest way to find out what a tier name actually buys. The answer turned out to be almost nothing, and the cheaper card won the measurement that matters most.

It joins CKBench, the running index of every card on the bench, alongside the Lexar Silver Plus 128GB, the Kioxia Exceria G2 128GB that anchors the low end of my reference band, and the SanDisk High Endurance 128GB that anchors the high end.

What the box claims, and what the card did

Two rows there are unusual enough to be worth stopping on. The write row is blank on Lexar's side because the package states no write speed at all, and the video class row is the only time a card in this corpus has cleared its printed class by three full grades.

Does the Lexar Blue really read at 160 MB/s?

Yes, and it goes past it. The card averaged 168.4 MB/s across three read passes, or 105% of the printed figure.

That sentence is harder to earn than it looks. 160 MB/s is above what standard UHS-I signalling can carry, which tops out near 104 MB/s, so hitting it needs a faster non-standard mode that the card and the reader both have to support. Plug this card into an ordinary reader and you will measure about 96 MB/s, and the number you are looking at will be the reader's limit rather than the card's. It is the single most common way a card review gets a speed figure wrong.

My own tool flagged the risk automatically on this run:

"INTERFACE CEILING: sequential read (168.4 MB/s) is within 5% of UHS-I DDR200 (~170 MB/s), the reader may be the limiter, not the card."

So I ran the whole thing again on a second reader. The SanDisk SDDR-B751 has already been measured pushing a different card to 203.5 MB/s, so it has room to spare here. The Blue read 168.5 MB/s on it, within 0.06% of the first answer, with 17% of that reader's proven capability left unused.

Two readers, one of them with 35 MB/s of headroom it never needed, returning the same number. 168.4 MB/s is the card, not the bench. This is the first card here whose above-standard read claim I have been able to settle outright rather than mark unresolved, and the reason is slightly deflating: Lexar set the claim low enough to clear.

What 155 MB/s of sustained writing actually buys you

The number that decides whether a card can record video is not its headline speed. It is the worst ten-second window of a long continuous write, because a camera does not get to record over the average. One bad stretch is a corrupted clip.

Across five minutes of committed writing the Blue's worst ten seconds came to 155.1 MB/s. Its V30 badge promises 30. It cleared that floor by a factor of five, and it cleared V90, the top of the video class ladder, by 1.7 times.

The curve above is the least dramatic in this entire corpus and that is exactly what you want. The lowest single second of the run was 150.99 MB/s, the highest 159.38, and the first thirty seconds averaged within 0.1 MB/s of the following four and a half minutes. There is no knee, which is the moment a card exhausts a fast SLC write cache and drops to the slower speed of its underlying flash. Plenty of cheap cards look excellent for twenty seconds and then fall off a cliff.

Five minutes only proves five minutes, so the real evidence is the full-card fill further down: 14 minutes and 15 seconds of continuous writing across all 125.6 GB, starting at 148.3 MB/s and ending at 147.8. Its worst ten seconds over the whole card was 144.0 MB/s. There is no cache to run out of.

One caveat belongs with the chart above. Both anchors in my reference band stay on standard UHS-I signalling while this card uses the faster mode, so part of that gap is bus tier rather than card quality. The band is a comparison, never a score.

Is the more expensive Lexar worth it?

This is the comparison the card was bought for. Same brand, same form factor, same bus, same C10 / U3 / V30 / A2 marks, one tier apart, both measured on the same reader in the same room. On paper the Silver Plus promises 28% more read speed and publishes a write figure the Blue withholds.

The two cards read at the same speed. Not similar speeds, the same speed, 0.02% apart, which is closer than two runs of the same card usually land. The Blue promises 160 and delivers 168.4. The Silver Plus promises 205 and delivers 168.5. One card beats its claim by 5% and the other misses its by 18%, and they are doing identical work.

What the extra tier does buy is random 4K write, where the Silver Plus is 52.6% quicker. That is the small scattered-file workload behind the A2 badge, and it matters if you are running apps off the card in a phone or a Raspberry Pi. It buys a slightly better sequential write, half a percentage point more of your capacity, and a card that runs about 4 °F cooler. It does not buy read speed, and it does not buy sustained write, where the cheaper card is marginally ahead.

There is a wider implication I want to flag carefully. When I reviewed the Silver Plus I could not settle its 205 MB/s claim, because no reader on this bench had been proven past that figure and blaming the card for the instrument would have been dishonest. This card now shows the same platform stopping at 168.5 on a reader with 35 MB/s to spare. That is indirect evidence and it does not settle anything, since a claim can only be tested on the card that makes it. It does mean a re-run of the Silver Plus is worth doing, and I have a fair idea what it will say.

Why does a 128GB card show up as 117 GB?

Two different things are going on, and only one of them is the card's doing.

The first is arithmetic and it is not a fault. Card makers count a gigabyte as 1,000,000,000 bytes, which is what the SD specification and the speed class floors use. Windows counts in binary, where a gigabyte is 1,073,741,824 bytes. The same untouched card is therefore labelled 128 GB and displayed as about 117 GiB. Nothing is missing. Two counting systems are being shown with the same three letters, and every honest card on the market does this.

The second is real, and this card gives back more than most. I filled all 125.6 GB with seeded, unrepeatable data across 118 files and read every byte back: zero mismatched bytes and zero read errors. The space is genuine. But 125.6 GB of a 128.0 GB label is 98.2%, leaving 2.36 GB unaccounted for, and that is before Windows does its binary conversion.

For context, most name-brand cards I have tested land between 99.9% and 100.1%. The Silver Plus one tier up delivers 98.8%. At 98.2% the Blue hands back less of its label than any other name-brand card in the corpus, and only the bundled and bargain-bin cards do worse. Lexar does disclose the possibility on the package, in the standard line about actual usable capacity varying, but it publishes no figure to hold them to.

Speed was also even across the whole card, which is the other thing a full fill exposes. Writing to the first, middle and last tenth of the card gave 147.4, 147.3 and 147.7 MB/s, a spread of 0.26%. There are no slow zones near the end.

If a card ever fails this test badly, that is a different story entirely and a far more serious one. The counterfeit ImageMate I tested exposed 128 GB and delivered 26.8% of it, destroying everything written past the real capacity.

Does the Lexar Blue get too hot?

It gets hot. It does not get slow, and those are separate questions.

At the end of five minutes of flat-out committed writing, with no fan and a room at 76 °F / 24.4 °C, the card reached 162.3 °F / 72.4 °C. That is a rise of 86 °F / 48 °C over ambient, and it makes this the hottest 128GB card on my bench bar one.

Nothing slowed down. Not in the five-minute test, where throughput varied by 0.91% end to end and the longest single stall was 32 milliseconds, and not in the 14-minute whole-card fill either, which is a far longer thermal soak. Heat that throttles is a fault. Heat that costs no speed is just a fact about the card.

The fair way to judge it is heat produced per unit of work, and on that measure the Blue is third best of every card I have run, at about 0.56 °F of rise for each MB/s it sustains. It is hot because it is working hard, not because it is inefficient.

The caveat is headroom. Lexar prints no operating temperature range for this card at all, only heat and cold resistance icons with no numbers attached, so the reference point has to be the 85 °C these flash parts are typically rated for. At 72.4 °C on a desk in a 76 °F room, that leaves roughly 12.6 °C of margin. A closed car in summer eats margin like that quickly. I would still put this card in a dash cam, but if your car bakes, the cooler-running SanDisk High Endurance is the more conservative choice even though it writes at little more than half the speed.

What happens when you make it churn

Loop recording is not one long clean write. It is a delete and a rewrite in the same place, over and over, and that scattered pattern is what exposes a weak controller. The test is three back-to-back passes of random 4K writes, watching whether the card degrades as its flash translation layer fills with garbage.

The Blue recorded 1402, 1402 and 1403 IOPS across the three passes. That is not a small decline, it is no decline at all, a change of one tenth of one percent in the card's favour. Random reads behaved the same way, moving 0.3% across three passes. The cheap unbranded reference card in the same chart loses 71% by its third pass and needs about two minutes to recover.

Recovery was equally uneventful. After the earlier tests had thoroughly churned the card, it was back at its full media rate within 60 seconds, which is the earliest my analyzer is able to declare stability. It may well be faster than that.

Two honest limits on this. Three passes is a directional indicator, not an endurance test: no bench session can tell you whether a card will still be working in year three. And absolute random 4K figures move by as much as 24% on reader alone, which I confirmed on this very card by running it on both readers, so treat the shape here as the finding and the raw IOPS as approximate.

This card does not report itself as a Lexar

Every SD card carries a small read-only register called the CID, written at the factory, holding a manufacturer code, a product name, a revision and a date. It is the closest thing a card has to a birth certificate, and it is where a counterfeit usually gives itself away.

I knew what this one should say before I read it. The Blue's own packaging names Longsys, the company that owns the Lexar brand, in its legal small print. Longsys is manufacturer ID 0xAD. Both of the other Lexar cards on my bench report exactly that, along with the OEM code "LS" and the product name "LX128".

The Blue reports manufacturer ID 0xC9, an OEM code that decodes to nothing readable, and the product name MSSD0. Across all 30 register dumps I have collected, 0xC9 appears exactly once, and this is it.

I want to be careful about what that does and does not mean. It is not a sign of a fake. The capacity is real and fully verified, the speeds are excellent, the packaging is genuine retail with a proper Amazon barcode, and a counterfeit does not usually outperform the brand it is imitating. What it means is that whoever assembled this card did not program it with the identity its own box would lead you to expect.

There is a stranger detail on top. The Blue and the Silver Plus report different manufacturers, yet their SD Status registers, 64 bytes describing allocation sizes, erase timings and declared classes, are byte for byte identical. That is the only such match across my whole archive, and every other identical pair sits inside a single manufacturer. Combined with two cards reading at the same speed to two decimal places, it points at a shared controller and firmware wearing two different badges. I checked three other tempting similarities first and threw all three out, because they turned out to be shared with Samsung and SanDisk cards too and therefore prove nothing. This is an observation, not a conclusion. Confirming it would need a teardown I have not done.

The practical consequence for you is simple: a card with an unfamiliar controller identity is one worth re-checking if you buy another later. Parts get swapped inside a product that keeps the same name and the same box, which is exactly what my two SanDisk Ultra units bought eight years apart turned out to demonstrate. If you own one of these, the CID decoder will read your own card's registers back to you.

How the Lexar Blue scores

The bottom line on the Lexar Blue 128GB

Buy it. The Lexar Blue is the fastest UHS-I card I have put on this bench for the workload that actually matters, sustained committed writing, and it costs less than the Lexar sitting above it that it beats. It exceeds the one speed claim printed on its box, it holds a flat 155 MB/s from the first second of a full card to the last, it does not degrade under churn, and every byte it advertises is really there.

Two caveats travel with that recommendation, and neither is a dealbreaker. It gives back 2.4 GB against its label, more than any other name-brand card here, so if you are sizing a card to the edge of your needs, size up. And it runs hot: 72.4 °C with roughly 12.6 °C of margin left, which never cost it a single MB/s on my desk but is worth thinking about in a car that sits in the sun.

The wider lesson is the one I did not expect to write. A tier name is marketing, not a specification. Lexar's cheap card and Lexar's better card read at the same speed, share an identical firmware configuration, and the cheap one writes faster for longer. The only place the extra money bought measurable performance was random 4K, which is the workload most buyers of either card will never touch. Test the card, not the badge.

One unit, purchased retail from Amazon, tested 4 September 2026 on protocol v1.14. Three passes are an indicator, never an endurance test.

Disclosure: I bought this card myself at retail from Amazon and Lexar had no involvement in the testing or this review. This post contains affiliate links. As an Amazon Associate I earn from qualifying purchases, at no extra cost to you. Affiliate links like these and my YouTube channel are what keep the site running.