← Back to leaderboard  ·  Show all task instances

jonas/tig

Text-mode interface for git

13,200 c medium
1,586
Generated Behavioral Tests
86.9%
Best Score
Muse Spark 1.3 (max)

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

24 runs
# Model Score Cost Calls
1 Muse Spark 1.3 (max) 86.9% $4.91 256 trace →
2 Claude Opus 5 (xhigh) 86.8% $52.79 269 trace →
3 Muse Spark 1.3 (xhigh) 85.5% $0.83 195 trace →
4 Claude Opus 4.8 (xhigh) 85.5% $15.90 163 trace →
5 GPT 5.5 (xhigh) 85.2% $4.82 39 trace →
6 GPT 5.5 (high) 85.0% $2.06 32 trace →
7 Muse Spark 1.1 (xhigh) 84.7% $0.22 119 trace →
8 Gemini 3.7 Flash 84.6% $1.36 126 trace →
9 GPT 5.5 84.2% $0.82 16 trace →
10 Gemini 3.1 Pro 83.9% $1.02 72 trace →
11 Claude Opus 4.7 (xhigh) 83.8% $1.57 65 trace →
12 GLM-5.2 83.7% $12.69 137 trace →
13 Claude Opus 4.7 83.6% $0.84 44 trace →
14 Muse Spark 1.2 (xhigh) 83.1% $0.26 93 trace →
15 Claude Haiku 4.5 82.9% $0.22 80 trace →
16 Claude Opus 4.6 79.6% $10.78 281 trace →
17 GPT-5.6 Sol (xhigh) 79.5% $10.23 53 trace →
18 GPT-5.6 Sol 79.3% $0.82 15 trace →
19 GPT 5 mini 76.9% $0.01 12 trace →
20 Gemini 3 Flash 71.2% $0.19 54 trace →
21 GPT 5.4 mini 39.0% $0.03 8 trace →
22 GPT 5.4 38.1% $0.12 7 trace →
23 Gemini 3.5 Flash 3.6% $2.59 99 trace →
24 Claude Sonnet 4.6 0.0% $14.93 382 trace →

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