OpenAI’s GPT-5.2 has launched, touted as its most capable model yet for professional knowledge work. However, initial tests by ZDNET show it only slightly outperforms its predecessor, GPT-5.1, raising concerns for users considering the $20/month Plus subscription. Although GPT-5.2 excels in areas like summarizing, explaining academic concepts, and literary analysis, it disappoints in coding tasks, revealing significant errors.
Brevity in responses, aimed at enhancing user experience, may frustrate professionals accustomed to detailed answers. For instance, while it received high marks for creative writing, it struggled with a travel itinerary that lacked comprehensive recommendations.
Image generation results were mixed, achieving a total score of 17 out of 20, but also showing limitations, particularly in detailed prompts. Overall, GPT-5.2’s incremental improvements prompt users to weigh its value against the subscription cost.
Source link
