← Back to leaderboard  ·  Show all task instances

Isona/dirble

Fast directory scanning and scraping tool

632 rs medium
715
Generated Behavioral Tests
94.8%
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) 94.8% $59.78 338 trace →
2 Muse Spark 1.3 (max) 84.9% $6.36 435 trace →
3 GPT 5.5 (xhigh) 81.5% $7.31 54 trace →
4 Muse Spark 1.3 (xhigh) 79.6% $0.94 178 trace →
5 Claude Opus 4.8 (xhigh) 76.8% $24.57 174 trace →
6 GPT-5.6 Sol (xhigh) 75.5% $4.19 49 trace →
7 GPT 5.5 (high) 73.8% $3.20 38 trace →
8 GLM-5.2 71.5% $32.04 291 trace →
9 Claude Opus 4.7 (xhigh) 71.0% $6.03 98 trace →
10 Muse Spark 1.1 (xhigh) 69.8% $0.34 88 trace →
11 GPT-5.6 Sol 69.7% $1.30 24 trace →
12 Claude Opus 4.6 67.0% $5.89 95 trace →
13 Muse Spark 1.2 (xhigh) 64.3% $0.25 139 trace →
14 GPT 5.5 64.1% $0.91 14 trace →
15 Gemini 3.6 Flash 63.4% $5.90 143 trace →
16 Gemini 3.5 Flash 60.7% $3.88 106 trace →
17 Claude Sonnet 4.6 52.0% $15.98 345 trace →
18 GPT 5.4 51.7% $0.39 11 trace →
19 Gemini 3 Flash 51.7% $0.31 80 trace →
20 Gemini 3.1 Pro 46.9% $0.85 33 trace →
21 Claude Opus 4.7 29.8% $2.28 71 trace →
22 GPT 5 mini 27.3% $0.02 9 trace →
23 Claude Haiku 4.5 25.6% $0.64 111 trace →
24 GPT 5.4 mini 24.8% $0.04 9 trace →
25 Gemini 3.7 Flash 0.0% $0.89 86 trace →

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