Nonobench
mauricekleine/nonobenchAn open-source benchmark of how well LLMs solve nonogram puzzles, from 5x5 to 20x20. Query every result through a REST API or MCP server.
Score: 100/100
Handshake verified by our own probe.
Tools · 11
Models ranked by accuracy; defaults to all effort levels for compatibility.
Provider ids, names, families and variant counts.
Model families, available efforts and best variants.
Side-by-side core overall and per-size accuracy, cost, latency and token results for model or family names.
Accuracy, cost, latency and token use for one model, broken down by grid size.
The benchmark puzzles with their ids and row/column clues.
One puzzle, including the clue text models were prompted with. The reference solution is only included on request; some puzzles have several valid solutions.
Check a grid against a puzzle's clues, using the same rule as the benchmark grader. Reports which rows and columns do not match.
Per-model outcomes for one puzzle. Answers are omitted unless requested.
Which puzzles one model solved, missed, timed out on, or has not run.
Individual benchmark runs (one model on one puzzle), optionally with the raw prompt and model output.
How to use
Claude Desktop or claude.ai: add a custom connector (Customize → Connectors; on Team and Enterprise plans an owner adds it under Organization settings → Connectors) with this server URL: https://www.nonobench.com/mcp
Cursor or Cline: add to your MCP config:
{
"mcpServers": {
"nonobench": {
"url": "https://www.nonobench.com/mcp",
"transport": "http"
}
}
}