Rabbit and Steel

Rabbit and Steel has the best recent record I've seen: 100 of the last 100 English reviews are positive. Not 98, not 99. All of them.

So the useful work here is finding what the other 2.1% object to, because on this evidence you're otherwise very likely to enjoy it.

The clarity problem

Visual clarity, colours and visibility appear in 19 of the hundred most-helpful negatives, and the complaint is precise rather than vague.

A 29-vote reviewer: "This game has really REALLY bad visual clarity. The colors are not distinct enough, I was having trouble discer[ning]." A 21-vote reviewer with 27 hours, who otherwise likes it: "Better colors for attacks and markers. Better/Customizable colors for other players, more visibility for markers. I will update this review if they add stuff that lessens the chance for you to die an unavoidable death."

In a game built entirely on reading telegraphed attacks — where the whole skill is seeing the pattern and moving — colours that blur together aren't a cosmetic issue. They're a fairness issue, and for anyone with colour-vision difficulty they're likely disqualifying. Accessibility scores a 5 on that basis.

The one encouraging note: dating the complaints gives 11 in 2024, 4 in 2025, 4 in 2026, and just one in the hundred most recent reviews. So it has improved. It hasn't gone away.

The scheduling problem

The second real objection, from a 28-hour player at 37 votes, opens by saying they don't want to be writing it: the game has "one absolutely crippling feature of being full committal game with no mid run saves in SP and MP. That means you need to dedicate a full 40m+ to a run. And 4 peoples schedules can very easily get interrupted."

That's a small mention count — three reviews — but it's a genuine practical constraint that prospective buyers should weigh, and it compounds the general co-op difficulty this library keeps running into with Machine Party and Big Walk: you need people, at the same time, for a fixed block.

Solo play exists but is penalised. A 57-hour reviewer: "I don't recommend this game for solo players… you have to beat hard mode to be able to unlock most of the stuff, or farm bosses over and over in the lower difficulties."

What it gets right

The premise, completely. If you've done end-game content in an MMO, the vocabulary is instantly familiar — telegraphed floor markers, stack points, spread patterns, mechanics that punish the whole group when one person misreads them. Compressing that into a 40-minute roguelike with four rabbits is a genuinely smart piece of design, and it's why FFXIV, MMO and raid terms appear in 20 of a hundred negatives, usually admiringly.

Classes matter properly — they change your job in the group rather than your damage number — and bullet-hell density gives the moment-to-moment play real texture.

The monetisation is clean. Two years of patches through version 2.0.2.8 in May 2026, with content added free; the only things you can pay extra for are two soundtracks. One negative reviewer notes approvingly that the DLC content was free.

The difficulty is serious: 54 achievements, the most common held by 79% of players but the median by 5.1%, and the rarest by 0.1%.

The verdict

Rabbit and Steel earns an 8.5. It's an inventive, well-supported, fairly priced co-op game that translates something genuinely difficult — raid design — into a form you can finish in an evening, and its current players are unanimously happy with it.

It falls short of a 9 because the thing it asks you to do, read attacks precisely, is undermined by colours that reviewers repeatedly say they can't distinguish, and because a 40-minute uninterruptible run is a real commitment for four people with lives.

If you have a group and good colour vision, buy it — this is one of the more original co-op designs of recent years. If either of those is in question, read the clarity reviews before you spend.