Models: 10 · Tiers: 26 · Samples: 80 · Invalid: 2

Day highlights

GPT-6 Astra

8.79 · max (best)
Model page

GPT-6.1 Sol

8.78 · max (best)
Model page

GPT-6 Sol

7.89 · max (best)
Model page

GPT-5.6 Sol

7.55 · max (best)
Model page

low

6.79 ± 0.68
Judge scores vary between passes

GPT-6 Luna

7.41 · max (best)
Model page

low

5.28 ± 1.80
Judge scores vary between passes

Grok 4.6

6.87 · high (best)
Model page

Grok 4.5

6.42 · high (best)
Model page

GPT-5.6 Terra

6.35 · medium (best)
Model page

medium

6.35 ± 1.40
Judge scores vary between passes

low

5.72 ± 0.73
Judge scores vary between passes

Grok 4.7

6.25 · medium (best)
Model page

GPT-5.6 Luna

6.19 · max (best)
Model page

medium

4.73 ± 1.08