Sudoku Engine: Generate, Solve & Teach with Human Logic

The go-to Sudoku engine for apps, games, printables, and learning products. Generate unique puzzles with real difficulty bands, then solve them the way a strong human would, with named techniques, not blind computer search.

Keywords people search for: Sudoku generator · Sudoku solver · Sudoku API · puzzle difficulty · step-by-step Sudoku · Sudoku hints · Sudoku techniques (X-Wing, naked pairs, …)

---

What you can do

• Generate Sudoku puzzles at precise difficulties (L1–L10, easy → extreme)
• Batch-generate many unique grids with optional seeds (reproducible)
• Solve Sudoku using human tactics, with summary or detailed step traces
• Hint the next move, analyze difficulty, validate boards
• Teach step-by-step with candidates and technique payloads

---

Why this Sudoku engine

Most tools either spit out a random grid or dump a machine solution. This engine:

You need	You get
Fair printables / games	Puzzles graded to a real band (L1–L10)
Learning & tutoring UX	Named strategies + cause→effect steps
Stable integrations	81-char boards, seeds, analysis JSON

Unique solutions, human-logic paths, and technique metadata, built to be the default Sudoku backend you plug into anything.

---

Difficulty bands (L1–L10)

Band	Level	Typical techniques
L1	Easy	Open singles
L2–L3	Easy / Medium	Singles, simple locked candidates
L4–L6	Medium / Hard	Pairs, pointing, box-line
L7–L8	Hard / Expert	Wings, early chains
L9–L10	Extreme	Advanced & rare tactics

Start with L1–L3 while testing; raise timeoutMs (30–60s) for harder bands.

---

Operations

Set operation once per run. Results go to the dataset and key-value OUTPUT.

Sudoku generator

Operation	Input	Output
generate	difficulty, optional seed, timeoutMs	puzzle, solution, analysis
batchGenerate	same + count (1–50)	one dataset row per puzzle

{
  "operation": "generate",
  "difficulty": "L3",
  "seed": 42,
  "timeoutMs": 30000
}

Sudoku solver & helpers

Boards are 81 characters, row-major, digits 1–9, empty = 0.

Operation	Returns
solve	Solved grid + analysis
solveSummary	Compact human-logic steps
solveDetailed	Detailed cause → effect traces
analyze	Uniqueness, difficulty, hardest technique, score
hint	Next human step (no full solve)
validate	Rule / structure checks
teach	Teaching payload (board, candidates, step)
candidates	Pencil marks for every cell

{
  "operation": "solveSummary",
  "board": "530070000600195000098000060800060003400803001700020006060000280000419005000080079"
}

---

Input reference

Field	Required	Notes
operation	✅	See list above
difficulty	generate*	L1…L10 (default L3)
board	solve* / analyze / …	81-char string
seed	optional	Reproducible generation
count	optional	Batch (1–50)
timeoutMs	optional	Default 30000

---

Output

1. Dataset: one item per result (batch expands to N rows)
2. Key-value OUTPUT: full JSON payload

Console views: Overview · Analysis · Solve steps

{
  "puzzle": "054063009962010730…",
  "solution": "854763129962418735…",
  "analysis": {
    "hasSolution": true,
    "hasUniqueSolution": true,
    "difficulty": "L3",
    "hardestTechnique": "…",
    "score": 12.4
  },
  "seed": 42
}

Errors look like { "error": { "code": "bad_request", "message": "…" } }
(bad_request, unsolvable, timeout, internal, …).

---

How to run

Console: Start → choose operation → open Dataset / OUTPUT.

API:

curl "https://api.apify.com/v2/acts/komeil_mehranfar~sudoku-engine/runs?token=YOUR_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"operation":"generate","difficulty":"L1","seed":1}'

import { ApifyClient } from "apify-client";

const client = new ApifyClient({ token: process.env.APIFY_TOKEN });
const run = await client
  .actor("komeil_mehranfar/sudoku-engine")
  .call({ operation: "generate", difficulty: "L3", seed: 42 });

const { items } = await client.dataset(run.defaultDatasetId).listItems();
console.log(items[0]);

---

Pricing (pay-per-event)

Event	Free / Starter	Scale	Business
Actor start	$0.001 / GB	$0.001 / GB	$0.001 / GB
Result (primary)	$20 / 1,000	$12 / 1,000	$8 / 1,000

One puzzle on Starter ≈ $0.021. Scale and Business cut the per-result rate. Set a max charge per run if you want a ceiling (suggested min $0.05).

---

Performance

• L1–L3: usually a few seconds
• L7–L10: increase timeoutMs
• Batch size multiplies result charges 1:1 with puzzles written

---

Support

Questions or feature requests: use the Issues / contact options on this Actor’s Apify Store page.