Check out our newest benchmark: SWE-sweep (<10% solved!)

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)

30 runs
# Model Score Cost Calls
1 Claude Opus 5.5 (xhigh) 98.9% $51.58 359 trace →
2 Claude Fable 5.1 (xhigh) 94.1% $111.75 389 trace →
3 GPT 6 Astra (max) 93.8% $21.05 74 trace →
4 Muse Wonder Full 89.8% TBA 416 trace →
5 Muse Wonder (xhigh) 55.2% $34.31 296 trace →
6 Muse Spark 1.3 (max) 39.6% $2.84 179 trace →
7 GPT 5.5 (xhigh) 38.4% $10.67 99 trace →
8 Muse Spark 1.3 (xhigh) 38.3% $1.04 199 trace →
9 Claude Opus 5 (xhigh) 36.8% $66.82 278 trace →
10 Muse Spark 1.2 (xhigh) 31.7% $0.63 133 trace →
11 Gemini 3.7 Flash 31.4% $0.93 88 trace →
12 GPT 5.5 (high) 29.4% $2.75 27 trace →
13 GPT-5.6 Sol 28.9% $0.90 13 trace →
14 Claude Opus 4.7 (xhigh) 24.0% $5.13 110 trace →
15 Gemini 3.6 Flash 22.8% $3.68 139 trace →
16 Gemini 3 Flash 20.2% $0.19 53 trace →
17 GPT 5.5 19.2% $1.27 16 trace →
18 Claude Opus 4.6 18.4% $14.43 260 trace →
19 Gemini 3.5 Flash 10.3% $3.77 91 trace →
20 GPT 5.4 mini 1.5% $0.04 7 trace →
21 GPT-5.6 Sol (xhigh) 0.8% $13.78 85 trace →
22 Claude Opus 4.8 (xhigh) 0.8% $17.60 134 trace →
23 GLM-5.2 0.8% $32.60 165 trace →
24 Claude Haiku 4.5 0.8% $0.38 90 trace →
25 Claude Opus 4.7 0.7% $2.54 76 trace →
26 Claude Sonnet 4.6 0.7% $38.31 461 trace →
27 GPT 5.4 0.7% $0.46 12 trace →
28 Gemini 3.1 Pro 0.7% $0.89 44 trace →
29 GPT 5 mini 0.7% $0.01 10 trace →
30 Muse Spark 1.1 (xhigh) 0.0% $1.06 113 trace →

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