Guilty Gear Strive Open Beta Test 2: Postmortem

New characters, revamped lobbies, and balance changes

The second Guilty Gear Strive Open Beta Test began at 8:00 PM PST on May 13th, 2021 and concluded at 8:00 AM on May 16th. OBT2 was preceded by a Japanese arcade-only location test where balance changes had been implemented in the game, following feedback from OBT1 and the game’s delay from an April 9th launch to June 11th. The most significant changes to OBT2 were the inclusion of Anji and I-No, the fourteenth and fifteenth characters in the game’s base roster, an improved lobby system, and dramatic balance changes around air-to-ground offense.

For this postmortem, I won’t be retreading ground covered in my previous essays about the Closed Beta Test and the first Open Beta Test. I will also begin with a preface statement: I think OBT2 is generally an improvement over OBT1, although they still haven’t addressed the changes to Gatlings in a way that satisfies me. For the rest of this essay, I will mainly home in on the most noticeable changes between the two versions we’ve gotten to play. If readers want to hear my criticisms in a video format, I have a (largely unedited) Twitch highlight here. (25min long)

I will also not be focusing on aspects of the game that aren’t strictly functions of the game engine or the UI/UX. This is a beta, so things like server disconnects and crashes are to be expected- as are statements like ‘This is a beta.’ I should also mention that I was able to get into the beta right away at launch time (and I have the stream footage to show it) so I didn’t suffer from the extra two-hour long wait that most people seem to have run into. Overall, the frequency of errors I experienced when playing online or connecting to opponents was significantly less than the same errors during OBT1. There was also no major downtime globally during OBT2 like there was during OBT1 (where servers were down for over twelve consecutive hours) so it’s safe to argue that the servers and matchmaking queues are more stable. Of course, errors are frustrating no matter what, and this is not meant to diminish the experience of players who frequently had connection issues during OBT2.

The Lobbies: What’s changed?

The most significant change to the lobby system is the addition of Duel Stations. While not quite the cabinet system from Guilty Gear Xrd Rev2 and Granblue Fantasy Versus, they’re essentially the same function. Kiosks are now placed evenly around the ranked Tower Floors and unranked Park lobbies, and you can see when a player is on matchmaking standby if they’re queued up on either side of the kiosk. These cabinets also display the characters of the players on standby, so you can seek out- or more likely, avoid- certain matchups. This all replaces the function in the previous lobby where you would simply ready up at a random part of the room, get warped around for your effort, and then wait for someone in the crowd of opponents around you to successfully interact with you. The Duel Station setups greatly reduce the amount of clutter and haphazard organization of matches in the public lobbies.

The other major update to the lobby system was instant rematch. This feature is an absolute godsend, even compared to the kiosks. I probably only played against twenty people during the entire beta, but I played each of those players for an hour or more depending on if I was in ranked or the Park. Instant rematch completely skips menus and loading screens, and you can play against an opponent basically forever until you get tired. Even though the amount of people I played was less than that of OBT1, I got substantially more matches played during OBT2. I ended up spending more time in the unranked Park than in the ranked Tower, since you could rematch infinitely in the former but only up to three times in the latter. (This is three games total, by the way, not best-of-three. You can win the best of three and still play the third game out to 3–0 somebody. I don’t really know if I want this changed or not to be like other games, since I do like playing as many matches as possible.) Instant rematch is such a common-sense feature, but it’s also so crucial to have to facilitate long sets.

My speculation for why Arc System Works and Team Red are so dedicated to the 2D voxel platformer look is probably because it’s easier to navigate and see everything in the lobby, especially if you’re using a fightstick. In GBVS’s public airship-themed lobbies, I cannot see cabinets behind the camera without turning on a cumbersome right-analog toggle for my stick’s lever, and in Xrd I have to hold down a button and move the lever to do the same. I don’t think the lobby’s aesthetic is as important as how it functions, and it’s largely easier to see everyone in it now, especially with the member list (and the ability to finally hide the News tab with L3.) Also, as mentioned in previous postmortems, when GGST’s Quick Start queue is working properly, you get access to a wide variety of players without having to see the lobby at all. As of OBT2, I think the game’s online matchmaking has vastly improved between beta iterations, beyond what rollback alone allows. Gameplay is subjective, but if you can’t play the game with other people, it doesn’t matter if it’s good or not.

The New Daredevils: Anji and I-No

In OBT1, I was mainly playing Giovanna, but in OBT2 I stuck to I-No the entire time. I was mostly familiar with her Rev2 iteration- I had picked up I-No when I first got Rev2 and I revisited her when she was announced as character 15. In the shakeup from Rev2 to GGST, I-No’s most significant changes were the loss of Vertical Chemical Love and jump-cancellable j.D, Megalomania combining the function of Longing Desperation with a proximity unblockable, and her inputs being universalized as quarter circles or half-circle-back-forward (632146) super inputs. (There are also minor changes like the loss of 6S and the Sterilization Method command throw.) Overall, I-No is still the same queen of air movement and high-low okizeme. J.S and the ability to chain Sultry Performance hops for ridiculous overhead mixups fill in the gaps left by VCL and j.D, and her rushdown and pressure is still powerful. While overall one of the more difficult characters in the GGST roster, this is the easiest iteration of I-No to pick up, and Rev2 I-No already arguably had a low skill floor.

I never played Anji, but I played some long sets against friends who played as him. While Fuujin’s rekkas no longer have autoguard, it doesn’t really seem like his mixup potential suffers as a result. Fuujin itself seems safe enough on block to bait opponents trying to mash out of rekkas, and more importantly, every variant of the Suigetsu dodge (on its own or attached to Fuujin and Gou) lets Anji completely skip neutral. Shitsu butterfly oki is also incredibly underrated. Even without the autoguard and special overheads that aren’t Fuujin follow-ups, Anji is a threatening bruiser who can avoid your own pokes and pressure and punish you with big damage, either with a throw after Suigetsu into butterfly oki and Fuujin pressure, or just by spending 50 Tension on a big-ass counter super. Anji seems both simple to pick up and an especially oppressive gorilla who straight up does not care about half of what you attempt at mid-screen.

Rebalance: Air movement

The changes to air movement and ground-to-air interactions are the most controversial aspect of OBT2’s gameplay, so let’s go over them in detail.

  • Gravity seems to have been adjusted overall. In loops, repeating the same move applies more gravity and pushback, like how other Guilty Gear games already prevent infinites. This is obviously to prevent things like Sol’s 5K dash cancel Grand Viper infinite from OBT1.
  • Any air action except for empty single jump has recovery frames on landing. This has been found by SKD to be about three frames of completely inactionable landing recovery. For (brief and reductive) comparison, +R has one to four frames of partially actionable landing recovery, and Rev2 has no landing recovery on any air actions apart from j.D, which has three frames of inactionable landing recovery.
  • Jumps and air dashes had their height and distance reduced. This is global, but also affects each character differently. For instance, while Millia’s jump height and airdash distance are both lower, her airdashes are both still about the same length as each other, forward and back. Sol and Ramlethal, on the other hand, have extremely diminished backwards airdash lengths.
  • There are now about five frames of pre-jump animation. This prevents up-back chicken blocking, which was so prominent in OBT1.
  • 6P has been buffed across the board. Generally, the hitbox is larger, the upper body hurtbox is smaller, and the move is now more active. This makes it more effective as an anti-air but has some… unintended consequences when interacting with grounded moves.
  • Blockstun when using meterless blocks in the air has been increased for the defender, ostensibly to incentivize both anti-airs and aerial Faultless Defense.

Most of these changes are fine, and some of them probably will only feel ‘weird’ in awfully specific situations. As an example, someone who plays Rev2 might be able to jump over an opponent and then one-frame throw them immediately upon landing, but in that same situation in +R, the aerial opponent is more likely to get thrown upon landing instead since their offense has been delayed by a frame. Since you can still begin combos and continue pressure after successful air options, the landing recovery doesn’t seem to be a major detriment to offense. This works because your opponent is often in enough hit- or blockstun for the landing recovery to not matter. Punishes will only happen on successful anti-airs or on whiffed air buttons. Likewise, things like additional pre-jump animation and buffed 6Ps allow defenders to respond adequately to aerial offense.

The biggest change to game feel is, instead, the various tweaks to air movement itself: the height, distance, speed, and gravity of jumps and airdashes has been changed in a very disconcerting way. As mentioned previously, Sol and Ramlethal’s backwards airdash are very stunted- this is a poor metric, but they hardly move the width of a single average character when they IAD back. I-No, on the other hand, wasn’t in OBT1, and her airdash distance forward and back is about the same. Still, her jump forward IAD doesn’t seem to clear most characters’ heads without superjumping, a problem exacerbated by her downward angled airdash. Some actions are still fast, and it seems like the speed of the popgun IAD on most characters has been increased (at the expense of early actions and variable descents) but movement in other areas feels weirdly heavier. I believe when people refer to the air movement feeling ‘clunky,’ they are referring to the more obviously felt changes described here, rather than the less obviously felt changes to things like pre-jump and landing lag.

It’s mentioned repeatedly that, instead of making such drastic changes to air combat, the balance should be reverted to how it was in OBT1 or even past Gears, and the ability to block grounded anti-airs without Faultless Defense should just be removed. I partly agree with the argument to return air unblockables to GGST and I hope there is some middle ground or even a total reversion to the air movement. It took quite some time for me to make the adjustment to the revised system, and I would probably be more sour on it if I wasn’t playing I-No, who had a Morrigan hoverdash on top of an IAD that was at least fast and full-length relative to most of the cast.

On the other hand… this is a personal argument, but I’ve had situations happen to me in past Gears where I swore I was air FD braking, and then my meter ran out and I got hit anyway by my opponent’s anti-air, and that certainly feels bad. I don’t know how universal that experience is, but it sucks to have picked the ‘right’ defensive option and still be wrong. I imagine that removing the need for metered air blocking was intended to prevent those situations where you FD brake and get hit anyway at low Tension. (This is not to say that this is exactly why GGST made those design choices, nor am I suggesting that GGST should be balanced to my preferences. It’s just an educated guess based on having played a thousand-plus hours of Gear and seeing those situations.)


Hopefully, ASW will take feedback and criticism in good faith and make as many adjustments as they can before the June 11th launch. The game is in a precarious situation right now. If it ends up being delayed again, I imagine that would probably further erode a lot of the faith and expectations in the development team, even if such a delay and additional work were needed in order to make the game good enough to ship. There’s less than four weeks remaining until launch, and major changes to the game have certainly been made in a short span previously- see the gap between the February OBT and the April loketest. This game is also slated for a debut at Evo- if it doesn’t launch on time, it might get cut entirely, but if it launches and it’s bad, it makes a terrible first impression on the global FGC. There’s also some features that I’m still waiting for more information on, like hitbox and frame data viewer in Training Mode and a timeline on crossplay functionality between PSN and Steam.

I do want to make it clear that I still had fun with the game and I played it for like twenty-something hours while it was live. I think most changes to the game have been for the better. I can hop onto GGST and go into a Park and just play for ages and not worry about the skill or connection of my opponent. Damage seems to have been tweaked laterally such that earning wall breaks and the Positive Bonus meter buff feels useful (one of my biggest concerns with the past betas) even though combo damage is still high. Characters still exert massive pressure on knockdown- Sol, Ky, Millia, Zato, Ram, and now Anji and I-No. I can still squeeze air movement into my combos, even though I’m traveling horizontally rather than vertically. It’s still Guilty Gear, it’s just different. I understand that the game should be good at launch rather than waiting for a patch, next season, or version update down the line, but I also can hardly remember a time in the past decade where a fighting game didn’t have janky or straight-up busted shit on launch. In the end, I just want to have the final game in my hands so I can dive right in, play with a ton of people, and experiment and discover cool stuff. Of course, I can still want that final game to be better.

Nathan “Lite the Iron Man” Dhami can be found on Twitter (@LiteTheIronMan,) on Twitch (,) and at your local.