OpenAI announces latest AI model — GPT-5.2 — and says it's better at professional tasks

On Thursday, OpenAI introduced its latest artificial intelligence model, GPT-5.2, which the company claims significantly enhances its performance for various professional applications. This new iteration surpasses its predecessors in several key areas, including spreadsheet creation, presentation development, image interpretation, code writing, and long-context comprehension. Starting today, users can access GPT-5.2 through OpenAI's ChatGPT chatbot and its application programming interface (API). This launch follows closely on the heels of the GPT-5.1 model, with the competitive landscape heating up as rivals like Anthropic and Google recently rolled out their own models. In light of this intensifying competition, OpenAI initiated a 'code red' strategy aimed at prioritizing resources and efforts towards enhancing ChatGPT. Fidji Simo, OpenAI’s CEO of applications, explained in a media briefing that this approach is vital for defining company priorities and reallocating focus where necessary. "We have had an increase in resources focused on ChatGPT in general, which contributes to the timing of this model's release, but that’s not the sole reason it's available this week,” Simo clarified. The new GPT-5.2 model comes in three variants: Instant, Thinking, and Pro. The Instant version is optimized for speed in writing and information retrieval, while the Thinking variant excels in structured tasks such as coding and planning. The Pro version aims to provide the most accurate answers for complex inquiries. "This has been in the works for many months," Simo noted, emphasizing the thorough preparation behind this model's release. "While we are proud of our ability to quickly launch new models, this particular integration was carefully developed over time."