The free card that outwrote every card I paid for

Why test a card nobody chose to buy?

Because somebody did choose it. A bundled card is a decision a camera manufacturer made on your behalf, usually invisibly and usually to a price, and it is the card that will be in the mount on the day you need the footage. This one came in the box with a Mangoal dash cam, and it is the second Mangoal card on this bench: the first camera shipped a Kioxia Exceria G2 that turned out to be genuinely competent. Two bundled cards from one brand, both good, is not a pattern I have seen from anyone else. It joins CKBench, the running index of every card on the bench, alongside the SanDisk Extreme 512GB and the Samsung PRO Plus 512GB, both of which it beats at sustained writing.

What the card claims, and what the card did

Two rows need their reasoning attached. A2 is not testable on any USB card reader, mine included: its floors assume a command queuing mode that USB card readers do not perform, so the card is neither credited nor penalised for it. That is a limit of every bench like this one, not a mark against the card, and I explain it on the speed classes page.

The 205 MB/s read is the one number I could not settle, and it deserves more than a verdict word. See below.

Why 205 MB/s is not a number you should expect

Standard UHS-I signalling tops out around 104 MB/s. That is the wire, not the card, and it is why so many perfectly good cards in my corpus all read within a hair of 96 MB/s and stop. A claim of 205 is roughly double what that wire carries, and the only way any card gets there is DDR200, a manufacturer extension that is not part of the UHS-I specification and that the card and the reader both have to speak.

This card clearly speaks it. Reading 168.4 MB/s and writing 157.0 MB/s are both far above the standard ceiling, and that alone puts it in a small group. But I still could not reach 205, and I am not going to blame my reader for all of it: the same reader has read 180.3 MB/s from a SanDisk Extreme in the same slot, so it is not capped at 170 and the shortfall is more likely the card than the chain. What is true is that 205 is beyond anything this bench has ever produced, which is why the row above says NOT TESTABLE rather than FAILS. A reader rated well past 205 is on its way here, and when it arrives this card gets re-run and the claim gets a real answer.

The practical version is shorter. On an ordinary card reader without DDR200, this card and a cheap one will both stop near 96 MB/s. The speed is genuinely in the card. Whether it is in your hands depends on a specification almost nobody checks before buying, which is the same argument I made about whether you need UHS-II.

What happens when you write to it for five minutes straight

This is the test that decides a video speed class, and it is the one most cards fail interestingly. Five minutes of continuous committed writing, cooling fan off, sampled every second. The verdict number is the worst ten second window, never the average, because your camera does not get to record over the average.

The card opened at 154.6 MB/s and closed at 155.3. Across all 300 samples it ranged from 148.9 to 158.3, a band of about six percent, with no step, no sag and no periodic cycle. The worst ten seconds landed at 153.4 MB/s, which clears the V90 floor of 90 by a margin wider than most cards' entire result. On a card whose face is printed V30, that is three full grades of headroom.

More important than the speed is what is missing. There is no cache cliff. Many fast cards write quickly into a small pseudo-SLC buffer and then collapse when it fills, which is exactly the failure that ruins a long recording. The five minute test found no knee across 46.5 GB, and then the capacity test wrote the entire 126.4 GB card and stayed flat end to end, opening at 147.5 MB/s and closing at 146.2. A cliff shows itself long before 126 GB. This card either has no meaningful cache layer or one large enough that filling the whole card never exhausts it, and for your purposes those are the same thing: the speed you see in the first ten seconds is the speed you get for the whole card.

I ran the sustained test a second time, four minutes later, on a card whose every block had just been written and with no warm-up pass. It gave 147.8 MB/s at its worst, 3.6% lower, and still held V90. I do not have an explanation for the 3.6% that I can prove. The temperature was identical, so it is not heat. It is a small unexplained difference and I would rather record it than invent a cause for it.

How hot does it get, and does it slow down?

The frame below was taken the instant the five minute write stopped, because a card sheds heat in seconds. Ambient was held at 76 °F.

158 °F is a rise of 82 °F over ambient, and the card never slowed down reaching it. The write curve finishes higher than it starts, which is the opposite of throttling. Heat that costs no speed is a fact rather than a fault.

Context in both directions. It runs cooler than the cards nearest to it on speed: the SanDisk Extreme 512GB hits 172 °F and the Samsung PRO Plus units sit between 163 and 169 °F. Measured as heat produced per unit of work, it is the most efficient microSD card in my corpus. But it also runs 15 °F hotter than the Kioxia it replaced, which peaked at 143 °F. That is the price of writing 2.4 times as fast, and in a windshield mount in summer it stacks on top of a cabin that is already well above outside air. Lexar publish no operating temperature range for this line, so there is no manufacturer figure to check that against. I am not calling it a fault. I am saying a card sold into camera duty with no published thermal range is worth noticing.

Is the 128 GB real?

Yes, and this is the only test that can prove it. Every free byte was filled with unpredictable verifiable data and then read back and compared: 126.4 GB across 118 files, zero mismatched bytes, zero read errors. Nothing is missing, nothing repeats, nothing quietly wraps around. Filling it took 14 minutes 34 seconds and reading all of it back took 13 minutes 43 seconds.

What you actually receive is 98.75% of the printed 128 GB, a shortfall of 1.6 GB. That is not a defect: every card reserves flash for bad blocks, garbage collection and its translation layer. The point is that no standard says how much of that reserve comes out of the advertised number instead of out of raw flash, and the spread across otherwise comparable 128 GB cards on my bench is real. A SanDisk High Endurance delivers 100.0% of its label. The Kioxia that came with the previous Mangoal camera delivers 96.7%. This card sits between them, and it is the only card I have measured that lands in that gap. Against that Kioxia it hands you 2.66 GB more usable space on the same printed capacity.

Does it get worse the harder you hammer it?

Loop recording is not one long tidy write. It is scattered small writes landing on top of blocks the card has to erase first, over and over, for years. I cannot test lifespan in an afternoon and I will not pretend otherwise, but I can measure the behavior that predicts it: three back-to-back passes of random 4K writes, watching whether the card degrades as the churn continues.

This card gave 8.77, 8.77 and 8.76 MB/s across the three passes. That is a fade of 0.1%, which is another way of saying no measurable fade at all, and 2139 write IOPS is the highest of any card I have measured on this reader. For contrast, the unbranded reference card in the panel below loses 71% of its random write speed by its third pass and needs about two minutes to recover afterwards. That is the difference between a card built for this duty and a card that merely survives it.

How it scores, and why it is not a 5

Worth saying plainly, because it looks strange: this card scores exactly what the Kioxia it replaced scored. The Kioxia takes 4, 5, 5, 4, 5 as well, on a card that writes less than half as fast. That is not the rubric saying they are equal. It is the rubric running out of room: this card is already at the top of the three dimensions where it dominates, so its advantage has nowhere left to go, and both cards take the same deduction for handing back less than the label promises. The number cannot separate them. The rest of this review can.

The bottom line on the Lexar Silver Plus 128GB

This is the best bundled card I have measured and it is not close. It writes 157.0 MB/s committed, floors at 153.4, carries no cache cliff across its entire capacity, barely fades under scattered-write churn, and runs cooler for the work it does than any other microSD card on my bench. Among every UHS-I card I have tested, bought or otherwise, nothing writes faster.

And the honest reason to care is none of that. In the camera it came with, this card is enormously over-provisioned, and the Kioxia it replaced was already more card than that camera needed. Pulling it out to copy footage is where the speed lives, and even there an ordinary reader without DDR200 will close most of the gap between this card and a cheap one. What survives every one of those caveats is quieter and more useful: the space is real, there is no cliff waiting at the end of a long recording, and it does not degrade when you hammer it. Those are the things that matter at 2am on the side of a road, and they are the reason this card earns its score.

If you own the camera this came in, keep the card. If you are shopping for a card to fix a dash cam problem, this is not the fix, and the speed classes page explains why almost nothing you own needs more than V30.

Disclosure: This card was not supplied by Lexar. It arrived bundled with a dash cam I bought myself, and Lexar had no involvement in or knowledge of 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.