Price per level depends on puzzle dimensions. More categories means exponentially more cross-references to audit for mutual exclusivity—a 4×4 board requires 48, a 12×5 requires 660. That’s the hard part.
| Words per category | ||||||
|---|---|---|---|---|---|---|
| 4 words | 5–6 words | 7–9 words | 10–13 words | 14–17 words | ||
| Categories per level | ≤6 | $40 | $50 | $60 | $80 | $90 |
| 7 | $55 | $65 | $85 | $105 | $125 | |
| 8 | $70 | $85 | $110 | $135 | $160 | |
| 10 | $110 | $135 | $170 | $215 | $255 | |
| 12 | $160 | $195 | $250 | $310 | $370 | |
| 15 | $250 | $305 | $385 | $485 | $575 | |
The base price covers standard mutually exclusive puzzle generation. These options are available at no extra cost—specify them when you submit a request:
Your cost per level depends on your difficulty distribution. Audit is always 20% of the generation price.
| Level type | Dimensions | Per level |
|---|---|---|
| Easy | 6 × 4 | $40 (audit: $8) |
| Medium | 8–10 × 4 | $70–110 (audit: $14–22) |
| Hard | 12–15 × 4 | $160–250 (audit: $32–50) |
A studio buying 200 levels across a typical easy/medium/hard distribution averages roughly $90–110/level for generation, or $18–22/level for audit only. Auditing is iterative—you get 20 free re-audits of the same puzzle as you refine it (as long as 80%+ of the words are unchanged).
Enterprise and volume discounts are negotiable—contact us for annual contracts and custom arrangements.
Two options, same pricing:
Authenticated dashboard for your content team. Configure puzzle dimensions, generate levels, review results, download JSON or TSV. See your account balance and spend history. No engineering required.
POST-based endpoint for programmatic access. Submit puzzle dimensions or an existing level as JSON, get back generated puzzles or audit reports. Same pricing, same results. Integrate into your content pipeline.
Submit a single puzzle—via the web portal or API—and get back a per-puzzle report:
Overlap detection—words that plausibly belong to more than one category in the same level.
Familiarity flags—words your target audience may not recognize, with suggested replacements.
Expansion suggestions—additional words and categories that fit your existing level without conflicts.
This is useful whether you build levels in-house, with LLMs, or with freelancers. The audit catches what row-by-row construction misses—the cross-references that accumulate as puzzles get larger.
Note: Auditing evaluates one puzzle at a time. Cross-level tracking—making sure words and topics don’t repeat across your full level set—is built into our generation service automatically.
Audit pricing is 20% of the generation price for the same puzzle dimensions. The same matrix applies—because the audit complexity scales with the puzzle complexity. More categories means more cross-references to check.
| Your puzzle | Generation | Audit (20%) |
|---|---|---|
| 6 categories × 4 words | $40 | $8 |
| 8 categories × 5 words | $85 | $17 |
| 10 categories × 5 words | $135 | $27 |
| 15 categories × 10 words | $485 | $97 |
Submit levels as a spreadsheet, JSON, or via API. Audit your most-complained-about levels first—if the system catches what your players are flagging, the data is real. Volume discounts apply. Contact us to discuss integration.
Every level is generated from Linguabase’s semantic network—400K words, 100M+ connections, organized by sense. The system clears the wordspace around each category as a puzzle is assembled, so nothing overlaps. It tracks what’s been used across your full level set, so level 847 doesn’t echo level 212.
We support three puzzle architectures:
Delivered as TSV, JSON, or whatever format your pipeline needs. See puzzle formats and examples →
For studios building their own content pipeline, the underlying data is available as a separate license. Over 2 million terms ranked by familiarity, 100M+ weighted associations, definitions, short clues, content filters, word families, and usage examples.
400K words ranked by familiarity. Larger sets available.
100M+ weighted connections organized by meaning. ~50K curated categories.
~55-word readable paragraphs covering all senses.
Short hints for gameplay—1 to 5 words, multiple angles per term.
Blocklists at two severity levels—offensive and suggestive.
Conjugations, variations, and words from the same root.
Up to 3 per word from literature and journalism.
Data licensing is priced based on scope and usage—contact us for details. Browse the data →
Michael Douma — About Linguabase
linguabase@idea.org