Hover a point for details · The line marks the Pareto frontier (best score per cost) · Click a point to see model details

21 runs
# Model Score Cost Calls
1 Claude Opus 5 (xhigh) 86.5% $39.23 300 trace →
2 GPT-5.6 Sol (xhigh) 80.9% $2.75 24 trace →
3 GPT 5.5 (xhigh) 80.7% $6.08 69 trace →
4 GLM-5.2 78.9% $28.56 243 trace →
5 GPT 5.5 (high) 77.1% $2.24 31 trace →
6 Gemini 3.5 Flash 76.3% $7.15 268 trace →
7 Gemini 3.7 Flash 76.1% $2.28 189 trace →
8 Claude Opus 4.7 (xhigh) 76.1% $17.05 263 trace →
9 Gemini 3.6 Flash 75.1% $3.47 146 trace →
10 Claude Opus 4.8 (xhigh) 74.3% $22.43 146 trace →
11 Claude Opus 4.6 74.3% $13.45 345 trace →
12 Gemini 3 Flash 73.3% $0.30 112 trace →
13 Claude Opus 4.7 72.1% $8.64 213 trace →
14 GPT 5.5 70.1% $0.91 14 trace →
15 GPT-5.6 Sol 69.3% $0.99 14 trace →
16 Claude Sonnet 4.6 69.1% $13.76 462 trace →
17 Gemini 3.1 Pro 44.8% $1.08 71 trace →
18 GPT 5.4 39.8% $0.20 9 trace →
19 Claude Haiku 4.5 35.7% $0.67 88 trace →
20 GPT 5 mini 27.5% $0.03 21 trace →
21 GPT 5.4 mini 3.0% $0.05 12 trace →

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