← Back to leaderboard  ·  Show all task instances

hpjansson/chafa

📺🗿 Terminal graphics for the 21st century.

4,648 c medium
1,931
Generated Behavioral Tests
58.4%
Best Score
Claude Sonnet 4.6

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 Sonnet 4.6 58.4% $50.80 532 trace →
2 Claude Opus 5 (xhigh) 55.9% $24.34 132 trace →
3 Claude Haiku 4.5 50.2% $0.82 136 trace →
4 Gemini 3 Flash 47.1% $0.17 65 trace →
5 Muse Spark 1.3 (xhigh) 33.9% $1.05 159 trace →
6 Claude Opus 4.6 33.5% $14.64 252 trace →
7 Muse Spark 1.1 (xhigh) 33.0% $2.56 297 trace →
8 GPT 5.5 (xhigh) 28.0% $8.49 77 trace →
9 GPT-5.6 Sol (xhigh) 27.3% $7.07 47 trace →
10 Muse Spark 1.3 (max) 25.9% $9.58 356 trace →
11 Claude Opus 4.8 (xhigh) 24.0% $11.77 96 trace →
12 Muse Spark 1.2 (xhigh) 22.4% $0.20 201 trace →
13 GPT 5.5 (high) 21.4% $3.78 44 trace →
14 GLM-5.2 19.4% $17.55 203 trace →
15 Claude Opus 4.7 (xhigh) 18.7% $7.73 146 trace →
16 GPT 5.5 16.6% $1.74 22 trace →
17 Claude Opus 4.7 16.1% $1.45 42 trace →
18 GPT-5.6 Sol 11.6% $0.53 8 trace →
19 Gemini 3.5 Flash 11.4% $4.78 160 trace →
20 GPT 5.4 8.7% $0.16 7 trace →
21 Gemini 3.7 Flash 6.1% $1.58 138 trace →
22 GPT 5 mini 4.6% $0.01 6 trace →
23 GPT 5.4 mini 3.5% $0.04 9 trace →
24 Gemini 3.6 Flash 3.4% $6.45 210 trace →
25 Gemini 3.1 Pro 0.0% $1.65 92 trace →

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