← Back to leaderboard  ·  Show all task instances

kyoheiu/felix

tui file manager with vim-like key mapping

888 rs medium
501
Generated Behavioral Tests
97.4%
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) 97.4% $85.95 384 trace →
2 Muse Spark 1.3 (max) 97.4% $4.95 278 trace →
3 Claude Opus 4.8 (xhigh) 90.2% $11.03 121 trace →
4 Muse Spark 1.3 (xhigh) 89.4% $1.27 237 trace →
5 GPT 5.5 (high) 88.2% $3.14 51 trace →
6 GPT 5.5 88.2% $1.12 22 trace →
7 Gemini 3.5 Flash 80.4% $14.30 200 trace →
8 Claude Opus 4.7 (xhigh) 76.6% $1.68 52 trace →
9 GPT-5.6 Sol (xhigh) 70.7% $8.68 65 trace →
10 Muse Spark 1.2 (xhigh) 67.3% $0.04 40 trace →
11 Gemini 3.6 Flash 65.1% $4.03 143 trace →
12 GLM-5.2 59.3% $30.22 247 trace →
13 GPT 5.5 (xhigh) 56.9% $6.12 46 trace →
14 Gemini 3.1 Pro 49.5% $1.12 62 trace →
15 GPT 5.4 mini 46.1% $0.06 44 trace →
16 Claude Opus 4.7 45.3% $1.13 52 trace →
17 Muse Spark 1.1 (xhigh) 43.7% $0.26 92 trace →
18 GPT-5.6 Sol 40.3% $0.70 12 trace →
19 GPT 5.4 39.1% $0.42 12 trace →
20 Gemini 3 Flash 33.3% $0.27 76 trace →
21 GPT 5 mini 22.4% $0.05 20 trace →
22 Claude Haiku 4.5 20.2% $1.01 142 trace →
23 Gemini 3.7 Flash 1.6% $1.39 118 trace →
24 Claude Opus 4.6 0.0% $2.48 244 trace →
25 Claude Sonnet 4.6 0.0% $4.16 360 trace →

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