← Back to leaderboard  ·  Show all task instances

rvben/rumdl

Fast Markdown linter and formatter written in Rust

1,051 rs hard
3,322
Generated Behavioral Tests
64.7%
Best Score
Claude Opus 5 (xhigh)

Results by Model

Hover for details · Click a model to open it · Drag to zoom the x-axis · Shift-drag to pan · The line marks the Pareto frontier (best score per cost)

Per-Model Results

25 runs
# Model Score Cost Calls
1 Claude Opus 5 (xhigh) 64.7% $119.42 480 trace →
2 GPT-5.6 Sol (xhigh) 61.2% $7.77 40 trace →
3 Muse Spark 1.3 (max) 56.0% $5.12 352 trace →
4 Claude Opus 4.8 (xhigh) 49.5% $21.14 165 trace →
5 Muse Spark 1.3 (xhigh) 49.2% $2.58 335 trace →
6 GPT 5.5 (xhigh) 48.9% $9.73 63 trace →
7 GLM-5.2 48.7% $57.45 290 trace →
8 GPT 5.5 (high) 48.7% $4.29 32 trace →
9 Claude Opus 4.6 40.6% $8.20 206 trace →
10 GPT-5.6 Sol 38.3% $1.18 14 trace →
11 GPT 5.5 37.8% $1.37 16 trace →
12 Muse Spark 1.2 (xhigh) 37.8% $0.94 103 trace →
13 Claude Opus 4.7 (xhigh) 34.1% $3.78 98 trace →
14 Gemini 3.7 Flash 31.5% $1.62 162 trace →
15 Gemini 3 Flash 25.9% $0.34 77 trace →
16 Gemini 3.1 Pro 23.5% $1.22 66 trace →
17 Gemini 3.5 Flash 17.2% $9.57 228 trace →
18 Claude Opus 4.7 14.6% $1.36 41 trace →
19 GPT 5.4 10.8% $0.36 7 trace →
20 GPT 5 mini 4.9% $0.04 21 trace →
21 Muse Spark 1.1 (xhigh) 2.8% $0.52 152 trace →
22 Claude Sonnet 4.6 2.8% $63.87 785 trace →
23 Gemini 3.6 Flash 2.7% $4.14 146 trace →
24 Claude Haiku 4.5 1.4% $0.60 88 trace →
25 GPT 5.4 mini 0.0% $0.06 13 trace →

Click a row to replay how that model rebuilt this program, or the model name to open its full run