← Back to leaderboard  ·  Show all task instances

oppiliappan/statix

lints and suggestions for the nix programming language

882 rs medium
815
Generated Behavioral Tests
95.8%
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) 95.8% $98.62 382 trace →
2 Muse Spark 1.3 (max) 92.1% $4.33 309 trace →
3 GPT-5.6 Sol (xhigh) 89.8% $11.43 71 trace →
4 GLM-5.2 85.1% $35.48 205 trace →
5 Gemini 3.7 Flash 83.6% $2.13 123 trace →
6 Muse Spark 1.3 (xhigh) 79.5% $1.65 227 trace →
7 Claude Opus 4.8 (xhigh) 79.4% $14.72 107 trace →
8 GPT 5.5 (xhigh) 69.0% $9.28 83 trace →
9 GPT 5.5 (high) 65.2% $4.29 48 trace →
10 GPT 5.5 62.1% $1.19 18 trace →
11 GPT-5.6 Sol 59.4% $1.03 15 trace →
12 Muse Spark 1.2 (xhigh) 58.9% $1.06 127 trace →
13 Claude Opus 4.7 42.8% $3.28 89 trace →
14 Claude Opus 4.7 (xhigh) 40.7% $6.92 142 trace →
15 Gemini 3 Flash 30.6% $0.20 51 trace →
16 Gemini 3.1 Pro 29.3% $1.18 60 trace →
17 Gemini 3.5 Flash 29.2% $6.32 164 trace →
18 GPT 5.4 25.4% $0.30 11 trace →
19 GPT 5.4 mini 24.0% $0.02 8 trace →
20 GPT 5 mini 21.7% $0.01 18 trace →
21 Gemini 3.6 Flash 11.3% $2.99 140 trace →
22 Muse Spark 1.1 (xhigh) 11.3% $1.26 225 trace →
23 Claude Opus 4.6 11.3% $14.17 282 trace →
24 Claude Sonnet 4.6 11.3% $24.17 391 trace →
25 Claude Haiku 4.5 0.0% $0.94 140 trace →

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