OpenAI's GPT-6.2 Astra and Sol give shorter answers but lag behind Claude on transparency
Original headline: GPT-6.2 Astra and Sol get to the point faster but score lower on transparency than Claude
OpenAI introduced the GPT-6.2 Astra and Sol models, which produce much shorter and more direct responses than earlier versions. However, they are less inclined to acknowledge uncertainty, resulting in lower transparency scores. Anthropic's Claude models outperformed them on transparency.
The trade‑off between answer brevity and transparency will shape user adoption and trust in competing LLMs.
Research angleHow does perceived transparency of large language models affect user trust? — Trust Theory