Though competitively priced at $49.99, the M800 Pro (3395) simply performed too poorly to be worth a recommendation. With the M800 Pro Highspeed, DELUX improves on the former while keeping the pricing in check at $59.99, but performance unfortunately continues to be lacking.

Physically, there isn't much difference between the M800 Pro (3395) and M800 Pro Highspeed. At 72 g, the Highspeed weighs basically the same, the battery capacity is still 500 mAh, and charging speed is virtually identical, too. That said, the Highspeed is built more solidly by comparison. When shaking, there is still a rattle coming from an indeterminable location, but creaking or flexing of the shell is no longer present when applying lateral pressure. In addition, actuating the side buttons by pressing below them is no longer possible, as the side-button construction has been thoroughly revised. Whereas the M800 Pro (3395) had tactile switches mounted vertically atop the PCB, the Highspeed comes with proper mechanical switches from Huano soldered to their own PCB. As a result, the side buttons are improved as a whole, though the forward button in particular still suffers from noticeable pre and post-travel, whereas the back button fares much better. Nothing has changed in regards to the main button switches, which still are from Huano (blue shell, pink plunger), providing a snappy and firm button response, and the button design still affords low pre and moderate post-travel, coupled with minimal lateral play. The scroll wheel encoder likewise is the same model from Huano (black, pink core), which suffers from high noise levels and mediocre separation between steps, resulting in tactility being mostly lacking. Lastly, the feet are still black-dyed PTFE, whose glide is decent at best, though the inclusion of a replacement set is welcome at least.

Even though both the M800 Pro (3395) and M800 Pro Highspeed use PixArt's PAW3395 sensor, differences can be found in the MCU department. While the former utilizes a rather obscure model from XC-Tech, the latter comes with an MCU from Realtek, which has a USB high-speed PHY, allowing for polling rates of up to 8000 Hz in wired operation, along with up to 2000 Hz in wireless operation. On paper, this sounds great, and in testing, results aren't too bad actually. CPI deviation is rather low, though correcting the little deviation is made unnecessarily difficult due to the software UI limiting adjustment to increments of 100 CPI, despite the 3395 natively supporting increments of 50 CPI. Another rather curious UI limitation concerns polling rate, as only four options are present: 500, 1000, 2000, and 8000 Hz for wired, and 250, 500, 1000, and 2000 Hz for wireless. General tracking at least is fine for 1000 Hz and below, though the fact that MotionSync is enabled permanently isn't ideal, as it likely adds motion delay at least equal to half of the set polling interval. Smoothing likewise is enabled permanently and adds significant motion delay at higher CPI steps. Polling rate stability is an interesting topic on the M800 Pro Highspeed. First of all, except for the respectively highest polling rate for wired and wireless mode each, all polling rates suffer from intermittent bursts of elevated jitter, along with generally increased noise. Second, and most bizarrely, the M800 Pro Highspeed generates a significant number of null reports either before, during, or after a motion. While this has no direct bearing on performance, it does point towards things not behaving entirely correctly. Lastly, at 8000 Hz polling in particular, a large number of null reports is mixed in with regular reports, at a ratio of roughly 4:1. Though there is no packet duplication trickery of the kind Corsair likes to resort to going on, this no doubt seems odd and suggests that 8000 Hz may be faked in some regard unlike 2000 Hz, which behaves normally.

Such anomalies aside, the plain numbers are quite good actually. In wired mode at 2000 and 8000 Hz, the M800 Pro Highspeed is always ahead of the Logitech G403 (control subject) by up to 0.5 ms, and in wireless mode at 1000 and 2000 Hz, it trails the G403 by roughly the same amount. Click latency likewise is on a good level: In wired mode, 1.4 ms are averaged at 1000 and 2000 Hz and 1.5 ms at 8000 Hz, and in wireless mode, the average is 2.6 ms at 1000 and 2000 Hz. That said, while not bad when taken on their own, the values at 2000 and 8000 Hz in wired mode in particular fall massively short of wired-only 8000 Hz mice such as the Razer Viper 8K or Zaunkoenig M2K, which all muster no more than 0.1 ms. Likewise, the dual-connectivity ASUS ROG Chakram X too posts a click latency of just 0.1 ms in wired mode at 8000 Hz. A click latency of 1.5 ms is plenty low, but easily achieved by other mice limited to just 1000 Hz, so 8000 Hz is little more than a gimmick in this regard. Keep in mind that the values given above have been measured with debounce time set to lowest, which is 4 ms. The other possible settings are 8, 15, and 20 ms, though all of these are simply wrongly labeled, as 1, 2, 3 or 4 ms of defer-type debounce are added in actuality.

In any case, by the far the biggest problem with the M800 Pro Highspeed isn't how it fares in testing, but rather how it does in practice. In my in-game testing, I've been experiencing instances of the mouse locking up entirely; i.e., for several seconds, no inputs were accepted at all, the sensor didn't track and buttons did not work. This happened more frequently the higher the polling rate was, topping out at two to three times per hour at 8000 Hz, and one to two times per hour at 2000 Hz, both wired and wireless. At 1000 Hz, I could game for hours without issue, though of course, it is entirely possible to still happen at some point, given the finite amount of time spent with the mouse. Either way, even if we assume that 1000 Hz works fine, buying a mouse capable of polling rates above 1000 Hz and then being forced to never use them to avoid the mouse not working at all is certainly not a good thing. In short, this is very much a fatal flaw and sufficient to recommend avoiding this mouse altogether.

As mentioned earlier, the M800 Pro Highspeed comes with a 500 mAh battery, and depending on where one looks, either 70 or 80 hours of battery life are cited by DELUX. Of course, this still leaves the question which connectivity and polling rate one of these two figures applies to—2.4 GHz or Bluetooth, 250 or 2000 Hz? Nobody knows, and the battery life indicator within the software is no help either, as despite being percentage-based with seemingly single-digit accuracy, the value changes at random and does not align with expectations at all, as even after 10 hours of use, it still showed 97%, which is impossible. Thus, actual battery life is anyone's guess, though charging at least is a pleasant experience, as the cable is flexible if a little short at 1.55 m. The software itself is decent otherwise, as resource usage is low, though the aforementioned UI restrictions certainly aren't ideal.

Overall, the M800 Pro Highspeed could be a worthwhile budget recommendation, as the $59.99 it retails for on Mechkeys.com are fair, but a mouse that frequently stops working altogether is simply impossible to recommend regardless of price. As such, I'd once again steer clear of this one.