MCP serverio.github.mauricekleine/nonobench

An open-source benchmark of how well LLMs solve nonogram puzzles, from 5x5 to 20x20.

UNRATEDActivestreamable-httpwww.nonobench.com

Overview

Score?

UNRATED 0.040

of what a free look can see, on 5 looks

Looks

last 8 hr ago

Tools

URL

www.nonobench.com/mcp

streamable-http

Says it is

nonobench 1.1.0

protocol 2025-06-18

In the record since

4 days ago

0median 0.606 · this server 0.040 · highest on record 0.8561

The tools this server lists, read out of the definition it returned
Tool	Schema
check_solution Check a grid against a puzzle's clues, using the same rule as the benchmark grader. Reports which rows and columns do not match.	input · output
compare_models Side-by-side core overall and per-size accuracy, cost, latency and token results for model or family names.	input · output
get_leaderboard Models ranked by accuracy; defaults to all effort levels for compatibility.	input · output
get_model_puzzles Which puzzles one model solved, missed, timed out on, or has not run.	input · output
get_model_results Accuracy, cost, latency and token use for one model, broken down by grid size.	input · output
get_puzzle One puzzle, including the clue text models were prompted with. The reference solution is only included on request; some puzzles have several valid solutions.	input · output
get_puzzle_results Per-model outcomes for one puzzle. Answers are omitted unless requested.	input · output
list_families Model families, available efforts and best variants.	input · output
list_providers Provider ids, names, families and variant counts.	input · output
list_puzzles The benchmark puzzles with their ids and row/column clues.	input · output
list_runs Individual benchmark runs (one model on one puzzle), optionally with the raw prompt and model output.	input · output

Verify it yourselfnpx teppi-check https://www.nonobench.com/mcpcurl -s https://api.teppi.xyz/v1/trust/mcp/mcs_01M3RGDVP8KP94ZDQ41EHJ8V5B