SanDisk MAX Endurance 128GB: the cheaper card is faster

Two cards, one brand, and the same numbers on both boxes

SanDisk sells two endurance-rated microSD lines. High Endurance is the one most people buy for a dash cam. MAX Endurance sits above it, and at 128 GB the two boxes make an unusual claim: exactly the same one. Up to 100 MB/s read, 40 MB/s write, Class 10, U3, V30, no app performance class on either.

Everything that differs between them is something a review cannot measure. MAX Endurance is rated for 60,000 hours of continuous recording against High Endurance's 10,000, and it carries a 10-year warranty against 2. Nobody can verify either figure in a review timeframe, and I am not going to pretend otherwise.

So the useful question is not whether the endurance rating is true. It is whether the two cards behave the same on everything a bench can measure, because if they do, the premium buys exactly what it says and nothing else changes. I ran the full suite on the MAX Endurance card, then ran a High Endurance card four hours later on the same reader, the same slot, the same room and the same software build. That answer turned out to be no.

Every printed claim, judged

This is the whole label, tested. Write speeds here are measured with every byte committed to the card, the way a camera actually writes, so they read lower than burst benchmarks like CrystalDiskMark. A burst benchmark measures the card's small buffer for a few seconds. This measures what the card sustains when the data has to land.

Five minutes of continuous writing, fan off

A dash cam does not care about burst speed. It cares about the slowest moment, because that is when a frame gets dropped. So the number this review turns on is the worst ten-second stretch out of a five-minute continuous write, with the cooling fan switched off.

The MAX Endurance held 51.6 MB/s through its worst ten seconds and settled at 52.8. That clears the 30 MB/s V30 floor by 1.7 times, and it clears the 26 Mbps its own endurance rating assumes by 15.9 times. There is no cache cliff anywhere: the card opened at 52.0 and finished at 53.0.

The whole-card fill is the stronger evidence. Writing all 127.83 GB, a job twenty-two times longer than the timed test, the rate started at 52.7 MB/s and finished at 52.9. Sampling the first, middle and last tenths of the card gives 52.8, 53.4 and 52.9, a spread of 1.2%. There is no fast zone, no slow zone and no cache to run out of. Many cheap cards print a big number they can only hold for a few gigabytes; a card with no cache has nothing to exhaust.

Both cards print V30. Only one of them is only a V30

Four hours after the MAX Endurance run I put a SanDisk High Endurance 128GB through the identical suite: same software build, same reader and slot, same drive letter, same 76°F room. Before comparing anything I checked the bench had not moved, and it had not. That card reproduced its own result from three weeks earlier to within a rounding error, 96.75 MB/s read against 96.77, and 88.47 MB/s sustained against 88.45.

Then it held 85.3 MB/s through its worst ten seconds, against this card's 51.6. It writes 39.5% faster, and where the MAX Endurance holds the V30 it prints, the High Endurance holds V60, a full video speed grade above the badge on its own package.

Nothing here is broken and no promise is missed. The MAX Endurance meets every claim on its own box, and the High Endurance simply exceeds a claim it shares. But a buyer moving up the endurance ladder is not getting a faster card with a longer rating, and there is no way to tell from the packaging, because the two spec sheets are word for word the same.

Both boxes print V30. Only one of them is only a V30.

The card registers back this up rather than sitting oddly against it. Every SD card carries an internal ID, and this one reports the product name SH128 at revision 8.0, made in June 2026. The High Endurance cards report SA128 at revision 8.7 and SN128 at 8.5. These are genuinely different parts, not one part sold at two ratings, and one of the High Endurance cards was manufactured within a month of this one, so it is not a case of newer silicon against older.

What that write speed actually supports

Every row below is judged against the worst ten-second figure of 51.6 MB/s, never an average, because the worst moment is the one that drops a frame.

For the job on the front of the box this is not close. A dash cam recording 4K30 needs about an eighth of what this card delivers at its worst. If you want the background on what those badges actually promise, I wrote up what the speed classes on an SD card really mean.

Capacity: all of it, verified byte for byte

Every card gets filled to the brim with verifiable data and read back, because a fraudulent card looks perfectly normal until you try to use the far end of it. This one wrote and verified 127,831,244,800 bytes across 120 files with zero mismatched bytes and zero read errors. That is 99.87% of the 128 GB label, and the small shortfall is ordinary formatting overhead rather than anything to complain about.

The package states the decimal-gigabyte convention directly, "1GB = 1,000,000,000 bytes. Actual user storage less", without publishing a usable figure. That is worth partial credit: it discloses the thing behind most of the confusion about missing space, then declines to say what you actually get.

Filling the card took 41 minutes 9 seconds. Reading it all back took 22 minutes 43 seconds. The High Endurance card fills in about 25 minutes, and if you routinely fill and offload a card, that gap is what stepping up the endurance ladder actually costs you in time. You pay it every single time.

The coolest of the family, for a reason worth knowing

With the fan switched off for the full five-minute write, the card peaked at 140°F in a 76°F room, a rise of 64°F. That is the coolest of the four SanDisk endurance cards on this bench: the High Endurance units ran 151, 155 and 156°F. It reads like an advantage, and mostly it is not one.

Divide the heat by the work and the ranking flips. This card produced 1.21°F of rise for every MB/s it delivered. The High Endurance card measured hours later produced 0.90. It runs cooler because it does about 40% less work, not because it wastes less energy doing it. Both numbers are honest and they answer different questions. The absolute figure is what matters in a sealed dash cam housing on a hot day. The per-MB/s figure tells you which controller is working harder.

Nothing thermal cost anything measurable. The write curve is flat from the first second to the last, the longest single blocked write was 105 milliseconds, and there is no throttling signature anywhere in the run.

How it scores on the CKBench rubric

My rating: 4.5/5, scored against the CKBench rubric.

That averages 4.4 and publishes as 4.5 out of 5. Nothing on the label is missed, and the half point comes off for a card that is outrun by its own cheaper sibling on the one thing both boxes describe identically.

Buy it for the rating, not for the speed

Judged on its own terms this is a very good card. It delivers its capacity, beats its printed write speed, holds every class mark on the package, has no cache to run out of, stays cool, and writes at one steady rate from the first byte to the last. If you are putting it in a dash cam or a security camera, which is exactly what the front of the box says it is for, you will get a card that does its job and a rating and warranty that no competitor at this capacity matches.

The thing to know before you buy is the part no spec sheet will tell you: the cheaper card in the same brand's own range is meaningfully faster, and both boxes describe themselves identically. Whether that matters depends entirely on what you need. If you need the rating, this is the card and the premium is buying something real. If you need throughput, you would be spending more for the slower option, and now at least you know.

One honest caveat to close on. I tested one MAX Endurance card and one High Endurance card. That is enough to compare these two units with every variable held steady, and it is not enough to describe two product lines. Cards from the same model number do vary, and I have measured a 15% spread within a single SKU on this bench before. If a second unit lands here, its numbers get published too.

Every card I have run through this bench, with the full method and the raw numbers behind each one, lives on the CKBench hub. The closest comparisons to this card are the SanDisk High Endurance 128GB it is measured against here and the VIOFO High Endurance cards, which are the budget end of the same idea.