No. Atomic chess has not been solved. There is no published proof of the game-theoretic value of its starting position, no tablebase covering the variant, and no distributed effort that has come close to settling it. This is worth stating plainly, because atomic is frequently confused with antichess — and antichess was solved, in 2016. The two variants are often mentioned in the same breath as “the solved ones,” but only one of them earns the label. Atomic remains open: a draw is widely assumed at the top level of online play, but assumption is not proof, and nobody has produced one.

Atomic is one of the most popular online chess variants, built on a single brutal modification to the standard rules. The confusion with antichess is understandable — both are small, both are played heavily on the same servers — but their combinatorial structure is completely different, and that difference is exactly why one fell to computation and the other has not. For the general question of what “solved” even means, and the larger story of standard chess, see Is Chess Solved? and the deep dive in Antichess Was Solved. Can Chess Be?.

What is atomic chess?

Atomic plays on a normal board with normal piece movement, but every capture triggers an explosion. The captured piece, the capturing piece, and every non-pawn piece on the eight surrounding squares are removed from the board at once. Pawns are immune — only non-pawn pieces detonate — so pawns survive an adjacent blast while knights, bishops, rooks, and queens do not. The king cannot capture, because doing so would blow itself up, which is suicide. The full ruleset lives on the Caissly Atomic page.

You win in three ways: by ordinary checkmate, by stalemate counting as a draw, or — the signature condition — by exploding the enemy king. Any capture made on a square adjacent to the opposing king removes that king in the blast and ends the game immediately. This single rule reshapes everything. Kings huddle behind pawn cover, because a pawn shield blocks the explosion that pieces cannot survive, and the entire game becomes a race of tactical threats radiating outward from every capture.

Has atomic chess been solved?

No, and there is no near-term prospect of it. To solve atomic in the formal sense would require proving the value of the opening position with perfect play, ideally backed by a tablebase mapping every reachable position to its outcome. Neither exists. There is no atomic-chess endgame tablebase the way there is for standard chess; the explosion rule changes the geometry of every capture, so the standard Syzygy and Lomonosov tablebases — built for ordinary capture rules — do not apply at all. A solution would have to be constructed from scratch for atomic’s own mechanics, and no such project has been completed or, to public knowledge, seriously attempted at scale.

What does exist is strong engine and human practice. Specialised atomic engines and top online players have mapped sharp opening lines and know which early sequences are dangerous — the variant is famously full of forced mates in the first dozen moves if a player wanders into a known trap. But knowing a sharp line is not solving the game, any more than knowing the Sicilian is solving chess. The value of the atomic starting position remains, formally, unknown.

Why atomic is harder to solve than antichess

This is the crux, and it is where the two variants split. Antichess was solved in 2016 by the mathematician Mark Watkins, who proved that 1.e3 is a forced win for White. The reason antichess yielded is that its rules shrink the game tree: captures are compulsory, so in most positions a player has only one or two legal moves. That compulsion collapses the branching factor and produces short games — twenty to thirty moves is typical — which is precisely what makes exhaustive search tractable. Antichess has roughly 1040 positions, and the compulsory-capture rule prunes the live tree far below that.

Atomic does the opposite. Its captures are optional, not forced, so the branching factor stays high. The explosion rule does not constrain choice — it expands consequence. A single capture can remove up to nine pieces at once, which means positions transform violently and the tree of meaningful continuations stays wide rather than narrowing. Atomic games are short in practice because someone usually blunders into an explosion, but short typical games are not the same as a small or shallow game tree. There is no compulsion to thin the branches, no equivalent of antichess’s forced-capture funnel. The structural feature that made antichess solvable is simply absent in atomic.

So the comparison that lumps the two together is misleading. Antichess was solvable because its rules are self-pruning. Atomic’s rules are self-amplifying. Solving atomic would be closer to solving a small but full-branching chess-like game than to repeating the antichess proof, and nobody has the method or the resources to do it.

What about a draw claim?

Among strong atomic players the working assumption is that the game is a draw with best play, and engine results at the top level are consistent with that. But this is the same situation as standard chess: a belief grounded in evaluation, not a theorem. No proof has been published. The starting position’s value is unverified, the game is not weakly solved, and it is certainly not strongly solved. Anyone stating flatly that atomic “is a draw” or “is solved” is reporting a hunch, not a result.

Where this leaves atomic

Atomic sits in the same unsolved territory as standard chess and the other modern variants such as King of the Hill — played at a high level, mapped in its sharp lines, but formally open. That is a perfectly healthy state for a game to be in. Antichess is the outlier, the variant whose rules happened to make a full proof reachable. Atomic, with its optional, board-clearing captures, is built the opposite way, and it will stay unsolved for the foreseeable future. The honest summary: atomic chess is not solved, it has never been solved, and it is materially harder to solve than the variant it is so often confused with.

References

Cross-links inside Caissly: the rules in full are on Atomic; the solved variant it is confused with is Antichess; the broader questions are in Is Chess Solved? and Antichess Was Solved. Can Chess Be?.

Issue Nº 006 · The Magazine · The Caissly Editorial