← Back to leaderboard  ·  Show all task instances

mikefarah/yq

yq is a portable command-line YAML, JSON, XML, CSV, TOML, HCL and properties processor

15,281 go
2,000
Generated Behavioral Tests
73.7%
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) 73.7% $102.58 372 trace →
2 Muse Spark 1.3 (max) 56.1% $4.15 328 trace →
3 Muse Spark 1.3 (xhigh) 54.1% $1.97 286 trace →
4 GPT-5.6 Sol (xhigh) 49.9% $4.21 29 trace →
5 Gemini 3.7 Flash 48.4% $2.79 136 trace →
6 GLM-5.2 44.0% $25.03 183 trace →
7 Claude Sonnet 4.6 39.5% $35.51 532 trace →
8 Claude Opus 4.8 (xhigh) 38.9% $16.89 113 trace →
9 Muse Spark 1.1 (xhigh) 38.4% $1.21 283 trace →
10 GPT 5.5 (xhigh) 37.5% $9.50 87 trace →
11 GPT 5.5 (high) 29.6% $2.60 32 trace →
12 GPT-5.6 Sol 27.4% $1.63 22 trace →
13 Muse Spark 1.2 (xhigh) 18.1% $10.13 429 trace →
14 Claude Opus 4.7 (xhigh) 16.5% $5.49 92 trace →
15 Claude Haiku 4.5 16.0% $1.55 183 trace →
16 GPT 5.5 14.5% $1.00 13 trace →
17 Gemini 3.5 Flash 14.4% $2.45 76 trace →
18 Claude Opus 4.7 11.5% $1.28 38 trace →
19 Gemini 3.1 Pro 10.3% $1.02 60 trace →
20 GPT 5.4 6.9% $0.76 20 trace →
21 GPT 5.4 mini 5.1% $0.02 6 trace →
22 GPT 5 mini 4.0% $0.02 8 trace →
23 Gemini 3.6 Flash 1.1% $4.68 144 trace →
24 Gemini 3 Flash 1.1% $0.19 65 trace →
25 Claude Opus 4.6 0.0% $2.85 217 trace →

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