Suguru — also called Tectonic, Number Cages, or Suguro — is a number-placement logic puzzle invented by Japanese puzzle author Naoki Inaba. Each board is divided into oddly-shaped cages (also called regions or zones). Your job is to fill every cell with a number, following two simple rules.
Choose any difficulty from Beginner to Cyborg ☠. Beginner/Easy run on a 5×5 grid with many given cells. Medium uses a 6×6 grid. Hard, Very Hard, Scientist and Cyborg all use the full 7×7 grid with progressively fewer given clues — Cyborg leaves only ~28% of the cells filled at the start.
Every day, the 🗓 Daily button serves the same puzzle to every player worldwide. Solve it once a day to build your streak.
Suguru was popularised by Dutch newspaper NRC Handelsblad under the name Tectonic. It belongs to the same family of "cage" puzzles as Killer Sudoku and KenKen, but with a much simpler rule set — no arithmetic, just the two rules above. That makes it approachable for beginners while still rewarding the kind of careful constraint-propagation thinking that Sudoku players love.
If you enjoy Suguru Max, you'll probably also like our Calcudoku (cages with arithmetic), Sudoku, and Flag Zone (Queens-style logic).
What separates it from Sudoku in practice is the neighbour rule: identical numbers may never touch, not even diagonally, so every placement radiates constraints in all eight directions. Each cage sets its own number range by its size. Nurikabe and Calcudoku continue the same Japanese logic-puzzle lineage.
I'll say this upfront: Suguru is the puzzle that humbled me as a builder. Most grid games I've made follow a clean generative recipe — construct a solution, carve away clues, verify uniqueness, ship. Suguru refused the recipe. The same grid can carry multiple valid results, which I found genuinely fascinating and slightly maddening in equal measure. It limits how many puzzles we can generate, and it took a lot of research before I understood why the puzzle behaves this way and how to work with it instead of against it.
Here's the design insight that eventually made large boards possible. Suguru has two rules — every region fills with 1 up to its size, and no digit may ever touch the same digit, diagonals included. Of those two, the neighbour rule is the workhorse. It prunes far more candidates than the region rule does, because every placed digit immediately poisons up to eight surrounding cells. So the generator leans on neighbour-elimination first. That single ordering decision is what keeps the big grids honestly logic-solvable rather than guess-and-check, especially through the diagonals, where most of the hidden pressure lives.
Now, the mistake I watch Sudoku players make here, over and over. Sudoku trains you to scan rows, columns, and boxes. Suguru has none of those. A digit can repeat across a row as freely as it likes — put a 3 at both ends of the top row and the puzzle doesn't care, as long as the two aren't touching. The only laws are the region fill and the no-touching rule. And Sudoku training whispers something actively harmful in your ear: ignore the diagonals. That is exactly the reflex Suguru punishes. The cell that kills your solve is almost always diagonal to the one you just placed.
My advice is to unlearn on purpose. When you place a digit, trace the full ring around it — all eight neighbours, corners included — and mark what that placement forbids. Then look at your smallest regions. A one-cell region is a gift (it must be 1), and a two-cell region is nearly one. Small regions plus the neighbour ring will cascade you into the medium regions without a single guess.
One more thing, since the honest version of this page should say it: if you ever find a second valid arrangement on one of our boards, you haven't broken anything. That's the nature of this puzzle family, and it's the exact property that sent me down the research hole in the first place. I built the Cyborg sizes for the players who want the diagonals to really bite. Start smaller than your ego suggests. The neighbour rule scales; pride doesn't.