First pioneered by the Finalmouse UltralightX, carbon fiber composite as a shell material has increasingly found adoption among enthusiast gaming mice, where saving every gram matters. The AKITSU—a collaboration between Arbiter Studio and Vancer—joins these ranks, and directly competes with the UltralightX Medium ("Lion"). At 39 g, the AKITSU only weighs a single gram more than the UltralightX Medium, albeit while being a decent bit smaller. In terms of shape, the AKITSU compares closely to the Endgame Gear OP1, though it ends up being even narrower, so those with larger hands may struggle to handle this one without cramps. Much like on the UltralightX, hexagonal holes are featured across the top, sides, and bottom, the latter of which is largely occupied by a depiction of a dragonfly. All the weight saving aside, the more pressing question of course is whether build quality holds up better than it did on the UltralightX, and at least based on my experience with a unit from the second batch I can say that it does. On my sample, there is some creaking when applying lateral pressure, but no flexing of the shell anywhere, no rattle when shaking the mouse, and actuating the side buttons by pressing below them cannot be done regardless of force used. In short, the AKITSU has the UltralightX squarely beaten in this regard. For those disliking the holes on the sides, a set of grip tape is also included, which can be used to cover them up.

For its main buttons, the AKITSU utilizes a set of switches from Huano, which are said to be customized. Be that as it may, I haven't seen this transparent pink shell, white plunger variety on any other mouse yet, and their actuation profile indeed seems largely unique. These are slightly duller, but far lighter than the Huano blue transparent shell, pink plunger variant, and lend themselves well to being spammed. The button design itself is sound as well, as lateral movement is rather low even when provoked, and even though post-travel is high, pre-travel remains pleasantly low. For the side buttons, low-profile 2-pin switches from Kailh are used, which shine with low pre and post-travel, although their feedback differs greatly between the forward and back buttons. Naturally, actuation isn't as satisfying as it would be on a full-sized mechanical switch, but that is entirely expected due to their size, and the actuation point at least is perfectly even. For the scroll wheel, TTC's blue/yellow encoder sees use, which is familiar from the Hitscan Hyperlight as well as recent G-Wolves releases, and while noise levels are somewhat elevated, tactility is very good, with distinct steps allowing for perfectly controlled scrolling. By default, the AKITSU ships with four non-dyed pure PTFE dot feet, which is mostly done to reduce weight, but a set of fitted feet is included with the mouse as well, and can be installed depending on preference. A sensor ring is not present either way, though given the small footprint of the AKITSU, its absence is likely negligible.

In terms of performance features, the AKITSU is in line with market trends. PixArt's PAW3395 sensor is paired with a Nordic nRF52840 MCU, and the included wireless dongle features a USB high-speed MCU, allowing polling rates of 8000 Hz in wireless operation. In addition, the dongle also features an LCD, showing various info such as the current CPI step, polling rate, MotionSync, or battery life. As far as numbers go, performance is largely in a great spot on the AKITSU. While a positive CPI deviation of 4.4% is present, this is easily accounted for and corrected. General tracking shows no issues, regardless of MotionSync being enabled or not, and polling is stable in wired operation. In wireless operation, however, elevated variance can be observed, though no actual missed polls. Another oddity exclusive to wireless mode is that when moved from a standstill, the first few counts are delayed, and the AKITSU only "catches up" fully after roughly 15 ms. This is somewhat comparable to a similar anomaly observed on the Glorious Model O 2/I 2 Wireless, and recently the Model O 2 Mini Wireless. At polling rates above 1000 Hz, the increased variance continues to persist, with the exception of 8000 Hz, which shows no issues in this regard. However, 8000 Hz suffers from a much graver issue, namely that motion delay randomly increases to significant and greatly varying degrees. From what I can tell, this typically happens more frequently the higher the CPI step is, though it'll ruin consistency no matter what. Accordingly, I'm unable to determine motion delay at 8000 Hz, but at 2000 and 4000 Hz, we're seeing excellent numbers: relative to the G403 (control subject), the AKITSU is ahead by 1.0 ms, which is on par with the very best. At 1000 Hz, the AKITSU is ahead by around 0.3 ms, and as always, MotionSync will add various amounts of motion delay depending on polling rate.

Click latency likewise is on the same, excellent level: regardless of polling rate, 0.4 ms are averaged in wireless operation, achieving equity with the Finalmouse UltralightX or Razer Viper V3 Pro at 8000 Hz. Keep in mind that this only applies with debounce time set to 0 ms, though since the AKITSU has a slam-click prevention algorithm implemented, there is no reason not to run 0 ms, provided no double-clicking is experienced. On firmware version prior to the latest 00.00.16.00, said slam-click prevention algorithm had a flaw where on lift-off, the around 17 ms worth of defer-type debounce were applied to the first click after lift-off, even when the mouse was no longer lifted, but thankfully, this has been fixed. However, there appears to be a minor idle delay present, in that with an interval of six seconds between samples, the average increased to 1.5 ms.

Another issue encountered on the 00.00.12.00 firmware in particular was that at higher polling rates, the AKITSU momentarily became unresponsive, for about 500 to 1000 ms, essentially ignoring user input during that period. On the latest 00.00.16.00 firmware, I'm no longer able to reproduce this behavior, and reports from other users seem promising as well, at least for the most part. Hence, while I cannot say with confidence that this issue is fully rectified, it appears to be improved significantly. On average, I would expect that the channel-hopping solution implemented on the AKITSU isn't as robust as others, so depending on environmental factors, issues may still crop up.

Much like the UltralightX and several other releases lately, the AKITSU is exclusively configured through a web driver, which is realized through WebHID and therefore restricted to Chromium-based browsers. In terms of features and functionality, I have no complaints, though given that there is no guarantee that the site will be hosted for years to come or is always available, I'd like to see a packaged desktop app as well. Some degree of on-device configuration is present at least, as both a CPI and a polling rate button are found on the bottom of the mouse, and this coupled with the LCD on the dongle gives one a decent amount of configurability even without the web driver. As far as battery life goes, up to 60 hours are cited at 1000 Hz, and up to 12 hours at 8000 Hz. Both the web driver and dongle include a (synchronized) percentage-based battery charge indicator, which is reasonably accurate and reliable. Unfortunately, I found myself switching through polling rates to nail down the connection issues, which is why I'm not in a position to give any estimates regarding the accuracy of the battery life claims. After six hours of mixed usage at every available polling rate, the indicator showed 89%, so there's that at least. Given the battery capacity of 200 mAh, I consider the values cited above plausible, and charging is a pleasant experience in general, as the charging cable is flexible and charging decently speedy.

With the latest firmware, several of the more significant issues the AKITSU was plagued with initially have been fixed, bringing it a lot closer to what it should have been to begin with. The only remaining issue is 8000 Hz, which continues to be wildly inconsistent. At the same time, not everyone is necessarily interested in using 8000 Hz, which is why this may be forgiven, and the AKITSU has plenty to offer otherwise. That said, one could make an argument that the Hitscan Hyperlight with the 8K dongle provides fierce competition, given how similar it is in terms of weight and performance while being more affordable. Still, when using the latest firmware, the AKITSU is a good mouse, albeit a pricey one for what is on offer, which is why it gets our Recommended award along with our Expensive badge.