THE SHORT VERSION
Number Mazes: a grid of square cells. A numbered path visits every cell exactly once, starting at 1 and ending at the total cell count. Some numbers are given as clues; fill in the rest. Interior cells have four neighbours (N/S/E/W).
Hex Mazes: the same puzzle on a hexagonal grid. Interior cells have six neighbours instead of four. The path still visits every cell exactly once from 1 to total, with the same given clues.
The rules are identical. The geometry is not.
SIDE BY SIDE
Cell shape
Number Mazes: square
Hex Mazes: hexagon
Interior cell neighbours
Number Mazes: 4 (N / S / E / W)
Hex Mazes: 6 (NW / NE / E / SE / SW / W)
Edge cell neighbours
Number Mazes: 2–3
Hex Mazes: 3–4
Corner cell neighbours
Number Mazes: 2 (maximum forcing power)
Hex Mazes: 3 (less forcing power)
Valid routes per gap
Number Mazes: fewer (lower branching factor)
Hex Mazes: more (higher branching factor)
Relative difficulty at same gap count
Number Mazes: easier
Hex Mazes: harder
WHAT STAYS THE SAME
The core solving loop is identical. List all given clues in ascending order. Compute gaps between consecutive clues (gap = higher clue − lower clue − 1). Work from smallest gap to largest: a gap of 0 means adjacent clues must be in neighbouring cells; a gap of 1 means one cell sits between them.
Dead-end analysis works the same in both variants. Any cell with only one valid neighbour the path can enter from — or that must be the endpoint of a segment — is forced. Place it without counting.
Connectivity is equally critical. A candidate path segment that would cut the remaining unvisited cells into an isolated pocket is always illegal — the path can't reach those cells afterward. Eliminating routes that break connectivity works identically in square and hex grids.
WHAT CHANGES — THE BRANCHING FACTOR
Each extra neighbour creates more valid routes per gap. In Number Mazes, a gap-1 pair (one cell between two clues) means the missing cell must be a common neighbour of both clue cells — and with only 4 neighbours each, the intersection is usually small. In Hex Mazes, each clue cell has 6 neighbours, so the intersection is larger and more candidate cells survive the first pass.
Corner and edge cells are the weakest forcing points in both grids, but Hex Mazes weakens them further. A square-grid corner cell has exactly 2 neighbours — the path enters one and exits the other, fully determined. A hex-grid corner cell has 3 neighbours — entry and exit have multiple combinations to check.
This means Hex Mazes require more explicit route enumeration for the same gap sizes, and connectivity pruning has to work harder before a candidate set collapses to a single valid path.
WHICH IS HARDER?
Hex Mazes are consistently harder at the same gap count and grid size. More neighbours per cell means more valid routes to enumerate, fewer forced placements from corner/edge analysis alone, and more connectivity pruning required before a solution crystallises.
As a rough calibration: a Hex Maze at tier N is comparable in effort to a Number Maze one or two tiers higher. If you find a Number Mazes easy level trivial, try a Hex Mazes easy — the increased branching makes it feel like a harder Square Maze without changing the rules at all.
STRATEGY THAT TRANSFERS
Gap computation transfers exactly: gap = higher clue − lower clue − 1. Sort gaps smallest to largest and work in that order regardless of grid shape.
Connectivity pruning transfers exactly. A segment that isolates N unvisited cells with only one entry point forces the path through that entry. The logic is purely about reachability — the number of neighbours per cell doesn't change the argument, only the effort to verify it.
The "shared neighbours" technique for gap-1 transfers, but the intersection is usually larger in Hex Mazes so more candidates survive. Combine it immediately with connectivity pruning rather than treating them as sequential steps.
WHEN TO MAKE THE SWITCH
If you play Number Mazes: try Hex Mazes when you can solve medium Number Mazes quickly without needing to write anything down. The gap logic you've internalised transfers fully; you're just working with a higher branching factor.
If you play Hex Mazes: try Number Mazes when you want faster, more satisfying forced placements. The lower branching factor means dead-end analysis and corner forcing resolve more cells per scan, giving a different rhythmic feel.
Playing both builds spatial reasoning in two different neighbour geometries. Hex Mazes sharpens route-elimination discipline; Number Mazes sharpens the instinct for when a cell is truly forced.