This benchmark gives the same intelligence score for GPT-6 Astra (max), GPT-5.6 Sol (max), and Grok 4.6 (high)? That seems very wrong to me, unless I'm misinterpreting the visualizations.