Chess Evolved Online Patch Notes — May 8, 2026
Aggregated from Steam, cross-tracked with Battle.net coverage on GamePatchNote.
behold the true power of a king
The amount of XP a player has will determine how many army points they can assign towards a custom king. There is no way to buy additional levels or unlock them with any currency.
/
XP requirements are high, and it is currently gated behind an entry barrier so that the narrow scope of "cost limit 1" is not a situation anyone will find themselves in.
/
after 50k, the +1cost levels are 10k each with a shallow curve. additionally, the base set moves of a king total to 16 cost, so the ability to have 9cost king is actually around 25pt of movetype selections.
/
XP gain itself will be adjusted over time if this is too low, but the learning curve of the game is high enough already so this is being treated as an advanced long term progression system for the actual players of the game rather than a playground for tourists. Everyone should remember that there are people spending their entire lives trying to master the use of a pawn wall in a vanilla mirror match, so you can wait a while to unlock king+++ and beyond.
/
costs are mapped on a per-slot basis for individual movetype:slot pairs, as well as an elegance and complexity system that rewards keeping the number of individual movetypes low. If you think about it, complexity of a custom king is its own kind of hidden balance problem because the opponent is expected to parse what they are even looking at in the middle of the game, and beyond that it makes it more frustrating to plan around, so having an incoherent pile of random movetypes piled into every slot is heavily penalized here even if they are the cheapest ones - if you make something clean and refined the elegance system gives you a discount. This is a simplified version of how new units are made.
/
by the way, the tech for the king lab includes the ability to adjust costs serverside and add new movetypes to king slots, as well as some future proofing towards handling 500+ movetypes per slot. much like gluttony, any new unit added into the game is an axis of multiplication for the total number of kings, which means the number of possible armies just became even more exponential.
-----
new units: 
most of these units introduce some kind of new mechanics
pegasus 'mount' toggles on to allow the pegasus to carry a unit behind,
transposer can mass-swap all adjacent units on kill,
witch can summon frog or transform enemy into toad, with a single-use magic,
spriggan summons a barrier of saplings into empty locations around itself on kill,
bunny/rabbit gain value on the opponent's side and can multiply if not contested,
priestess can enchant the king at game start or when revived or spawned from promotion,
flying swords can teleport to the king from anywhere, but only attack while adjacent to the king,
decoy can swap places with the king within augment range,
also sloth... a warning, an ominous delay, and then a rampage - fancy indicators included.
I am too lazy to explain why sloth is so legendary,
but maybe after the 8th sin is added I will explain.
---
massive balance shift:
the historical balance of this game often involves cases where a unit is overpowered at 1cost and then suddenly becomes 'useless' at 2cost, because in some abstract sense it is worth ~1.5cost. This kind of rounding error creates an additional nightmare because the rounding errors can compound across the tiers of the units, as strict upgrades are prohibited by design, despite the lowminded trolls claiming 'p2w'. This means a minion would often follow the minimal increase pattern like 1/2/3/4, and inevitably make many of the minions weak as a result of granularity itself, leading to many tier shifts or moveset alterations trying to avoid the rounding errors and snap onto exact 1.0 precision, which then may conflict with a different unit and shove them forward by +1cost. This means with most minions hovering in the 0~5cost range it was literally impossible to reach a perfect balance, and by the same logic 200pt is impossible as well, but because the rounding errors accumulate across tiers it turns out to be more than 2x precision overall after accounting for the game's strict balance rule. The clear downside is that it makes the game harder to understand because of the slightly larger numbers, but the clean elegance of '100pt' does not realistically fit the game's complexity, so it was always going to force balance problems. The remaining issue here is that minions at 0cost do not gain any precision from the 2x granularity shift, so more work has to be done in order to make some of those stronger. Additionally, greed doubles to 6/4/2/0 and rebalances to 4/3/2/1, allowing the weakest champions to drop below the previous floor.
--- 
about rebalance and new unit mechanics:
the sheer scale of rebalance being done makes most individual cost adjustments summarize into "everything's cost has been adjusted", but there are also mechanical changes along the way. Almost everything that was weak has been buffed and everything that was overpowered has been very slightly nerfed, which probably sounds like 'power creep' except this game has such extreme emphasis on balance and so little power creep that people were starting to ask that the base tier queen be nerfed by +1 just for good measure. This time around the idea is that 50 overpowered units instead of 5 is a more balanced game - the scale of theatrical pandemonium that is possible in CEO is going to continue scaling up, starting with some of these changes. Weak units will get overhauls more often than strong units get heavy nerfs. Most things were already within the range of a +/-1 rounding error with very few exceptions, and the classic units will remain strong as well.
--- 
a common problem across CEO history has been the infinite repeating perpetual threat type of patterns that can form against minions from a ranger/poisonmage/etc or against king with threats like angel/ninja or anything else that forces the match into move decay and projections of inevitability where any dispute in assessments result in playing through all 50+ moves, or one player simply surrendering. The exhaust mechanic solves this in almost all cases, by inflicting a single turn stun on the repeating unit, allowing one side to pivot and forcing the exhausted side to move something else for at least 1 turn. More complex repeat structures are still possible involving multiple units or custom kings, but for now all of the most common and most annoying mindless repeats are prone to triggering exhaust and are no longer considered valid gameplay.
--- 
some people prefer to think with visual anchors and some people try to convey things in streams but had no convenient options available. these arrows are a janky prototype but technically can draw onto the board, coloring based on the last unit team they were dragged from.
--- 
effects like the new null magic or any growth/decay effects, temperance/elementals/etc were all invisible from a screenshot or intuitive visual parsing point of view. To solve that missing axis of visual information, this mechanic displays a small '(X)' value label onto any unit that has its value changed away from the base value, and will also display custom king levels in army edit or army select screens to better differentiate them.
--- 
some keywords allow edge-case details that clutter movetype descriptions to collapse down into secondary info inside the keyword rollover, as well as including some more details that were never added to movetypes in the first place. the initial pass across keywords was a combination of sparse selection of ambiguity and complete overkill in a way that the highlights themselves are not an indicator of warranted attention. These also have a string layer stacking such that something like '\'can also have a nested keyword '\', but the current keyword selection, definitions ,and tech are all essentially prototypes (ex: you cannot chain into the description to open keyword sub-panels, yet)
+
unit previews example
about unit previews:
promotion has always been a layer of pseudo-hidden information, as a new player may see something like "promotes into priestess" and have no idea wtf that even means. Additionally, the ambiguity of promotes added a level of cognitive load even for veteran players, since each minion type promotes to a different champion/tier and required a mental map of their exact movetypes/value in order to properly plan around promotions with any level of accuracy. For this reason, the rollover of unit name keywords provides a fancy reference frame of their moveset/value, and clicking on it will change over to that unit's display card for easy reference during gameplay.
---
about scroll bar:
in the past I have mentioned a 'complexity cap' rule that by design prohibits me from expanding unit card info in any way, often needing to rephrase movetypes or minify down their variety in order to fit all of it onto the unit panel at once. Any circumventions like smaller text were also not allowed. However, the custom kings can have multiple extra movetypes added onto the king in addition to their other info, and if an opponent has a custom king there must also be some way to reference what the tiles are. This means the complexity cap is completely antithetical to custom kings and no longer valid. The king will list all of their extra movetypes in their info panel and has a scrollable list viewport to see them all (mouse wheel only atm) Since the king is "a unit" this also means other units could in theory have higher complexity, but I enforced the complexity cap for a reason so those would probably be barely over the line or on some kind of special PvE units or w/e. An opponent should not have to read 16 scrolling lists of info to know what your army does, so these are intended mostly for the king.
--- 
the augment range effect that was added to unit moveset grids works very well for an intuitive visual of its effect without parsing the text for the listed range number, but that was missing from the board during gameplay which requires a subtle friction of mapping the moveset's augment range onto the unit position. Having augment bounds visible directly on the board allows a more intuitive visual of what the unit does.
--- 
in the past, mastery XP gain could only be assessed by taking note of mastery and comparing after a match, making it into a kind of invisible background system detached from any direct action:reward correlations. The mastery system is one of the primary sources of rubies, so in many cases this mastery XP value is effectively the highest reward from victory, but people were made to only look at 'coins' and approximated their progress based on that alone. The progression in CEO scales across multiple things, so having some added coherence here to make that more obvious is likely to have a larger impact than the trivial change might imply.
--- 
past chaos portal challenges have broken replays because the RNG of chaos effects quickly desyncs from whatever the person posting the replay into the discord had witnessed before. This means any other kind of PvE-only chaos effects would break as well, so the replays are going to use a deterministic seed in cases where a chaos unit is part of the replay, and in theory this will allow replays to work with ANY chaos effects, which are of course the exclusive domain of non-ranked CEO
--- 
these are mostly a novelty, but since half the people on kongregate had CEO avatars it tells me that the ability to export them in some way is considered useful. The avatar exported is a simple and direct attempt to export a .png directly from the current avatar, while the prototype art lab is a bit more involved and allows editing like a canvas (the tools are all janky, so don't bother treating it like it's supposed to be full-on photoshop... yet)
--- 
across the years requests for metrics have been a recurring theme, as it is usually somewhat interesting to see winrates/usage/stats of various units. the metrics themselves come with 40 asterisks of criteria for if the winrate is useful or not, but overall usage is a solid signal, and the metrics lab currently doubles as a master list unit reference for the entire unit library (except special units like sapling/pillar/twins/etc.) please refrain from spamming the metrics lab or trying to break it or I'll just keep it to myself. The lab is mine, after all.
--- 
most of the bugs in this game amount to edge case nuance and specific interaction quirks across the extreme scale of possible interactions. for example: aqua-vs-fireball included a major bug. the complexity of the movetypes/passives/effects interacting in a type-agnostic way is then amplified tremendously with the inclusion of things like gluttony or custom kings, so presumably there are lots of strange interactions laying around. The priority for fixing gameplay bugs found in this version will start with the most common/prevalent ones rather than severity (from the example above, aqua-vs-fireball was severe, but it's such a meme scenario that almost nobody encountered that problem in normal play) The current version has more edge case problems but delaying the update to test for and resolve them would be silly since the vast majority are rare or small quirks.
--- 
these patchnotes do not even cover everything, nor do they explain everything that is covered in a complete way. the design theory involved goes beyond what could be contained into a coherent format, and 'full patch notes' would in theory provide a complete explanation of balance changes, but you are already looking at like ~3000 words, and it takes time to write all of this.
...
--- 
the primary use case for larger board sizes will be in custom lobby system for duels and various PvE systems such as challenges. the tech is being setup to handle smaller boards as well as very large boards, but is currently not optimized so a small warning will show if you're playing in the lab and try to scale it up to like 100x100 or w/e. Part of the lag is that it's using a mostly-old challenge gen logic and trying to make a valid challenge generation on the larger board, which means there is a ton of extra work going on in the background in addition to 100x100 being 10000 tiles. That being said, something like a 50x50 board would take forever to play across and there is no camera pan/zoom yet. A more sane range of board sizes is in the 6x~20x range, but a lot of things have to be made in order for any of that to be useful (ex: the CEO AI is made of magic pixie dust, but currently it thinks it's on an 8x8 board even on the larger one) I'd like to take a second to mention that despite the CEO AI having many errors, it actually is capable of scaling up to extreme board sizes, because it assesses abstract value heuristics rather than trying to tree-search a path to victory. Some people already know this, but the CEO AI literally does not look ahead at all, meaning everything it manages to do it does before thinking, if you can imagine that. The AI will be improved more later - for now it still has the ghostline error and will often suicide even at higher difficulty.
this means for now you can only tinker with the concept in the lab.
larger board sizes will be added as PvE after some of the larger issues are sorted out.
---
for now, you'll need to figure out how to deal with having 999999 different kings on the board.