← Back to leaderboard  ·  Show all task instances

jqlang/jq

Command-line JSON processor

34,541 c medium
6,072
Generated Behavioral Tests
94.7%
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) 94.7% $100.84 341 trace →
2 Muse Spark 1.1 (xhigh) 91.3% $0.35 119 trace →
3 GPT 5.4 90.1% $0.15 140 trace →
4 Muse Spark 1.3 (max) 83.2% $5.87 399 trace →
5 GPT-5.6 Sol (xhigh) 81.3% $6.05 44 trace →
6 GPT 5.5 (xhigh) 75.4% $15.55 148 trace →
7 Muse Spark 1.3 (xhigh) 72.1% $2.89 327 trace →
8 Claude Opus 4.7 63.9% $2.69 80 trace →
9 GPT-5.6 Sol 62.2% $1.27 17 trace →
10 GPT 5.5 (high) 60.0% $2.38 26 trace →
11 Gemini 3.6 Flash 56.1% $7.69 170 trace →
12 Muse Spark 1.2 (xhigh) 39.5% $0.82 172 trace →
13 GPT 5.5 31.5% $0.67 13 trace →
14 Gemini 3.5 Flash 30.0% $4.38 192 trace →
15 Claude Haiku 4.5 21.6% $0.92 111 trace →
16 Gemini 3 Flash 18.7% $0.21 46 trace →
17 Gemini 3.1 Pro 16.1% $0.72 60 trace →
18 GPT 5 mini 7.8% $0.01 10 trace →
19 GLM-5.2 1.1% $59.61 268 trace →
20 Gemini 3.7 Flash 1.1% $2.36 192 trace →
21 Claude Opus 4.7 (xhigh) 1.1% $12.97 153 trace →
22 Claude Opus 4.6 1.1% $18.26 384 trace →
23 Claude Sonnet 4.6 1.0% $4.40 289 trace →
24 Claude Opus 4.8 (xhigh) 1.0% $40.74 179 trace →
25 GPT 5.4 mini 1.0% $0.02 15 trace →

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