An Artificial Analysis evaluation found that OpenAI's GPT-6.2 Astra and Sol produce answers 34 to 40 percent shorter than their predecessors. Their tendency to admit uncertainty dropped sharply, with Astra falling from 96 to 24. Both models score lower on structure and transparency than Anthropic's Claude Fable 5.2 and Opus 5.5, though Opus 5.5 posts the highest informativeness score among all four models.
No score is assigned. Sources and their independence are shown in the citation chain below.