OpenAI has launched GPT-5.2, its most advanced AI model to date, showing improvements in writing, coding, and reasoning benchmarks. This release follows a recent “code red” declaration by CEO Sam Altman, aimed at enhancing ChatGPT amidst growing competition, particularly from Google and its Gemini model.
Fidji Simo, OpenAI’s CEO of applications, stated that the code red initiative is intended to focus the company’s resources on ChatGPT, clarifying that the launch of GPT-5.2 was not expedited, despite additional resources being allocated to the project.
While OpenAI’s models were leading the market upon the debut of ChatGPT in 2022, competition has intensified. Google has shown significant advancements with its Gemini app, which boasts over 650 million monthly active users, nearly matching OpenAI’s 800 million weekly active users. Consequently, OpenAI has scaled back some ambitious plans and redirected efforts to enhance its core AI technologies.
GPT-5.2 consists of three variations: Instant, designed for faster responses and information retrieval; Thinking, which excels in tasks such as coding and mathematical reasoning; and Pro, the highest tier that offers top accuracy for complex queries. This release targets everyday professional applications, with GPT-5.2 Thinking reportedly outperforming human professionals in over 70% of various tasks, completing them significantly faster.
OpenAI asserts that GPT-5.2 will markedly reduce the incidence of inaccuracies, or "hallucinations," common in AI responses. Compared to its predecessor, GPT-5.2 Thinking displayed a 38% reduction in such inaccuracies.
The new model will be made available to ChatGPT users and developers via OpenAI’s API product, promising enhanced performance for both routine and more advanced needs.
However, past experiences with user feedback, such as the backlash against GPT-5’s perceived colder responses, remind OpenAI that user experience goes beyond benchmark scores. There have been concerns around mental health issues related to ChatGPT’s use, prompting the company to implement better safeguards for sensitive topics while planning to introduce an age-prediction feature to protect younger users.
In light of increasing competitive pressures, OpenAI aims to expand its user base. Recent internal communications indicated that the company’s leadership feels it faces unprecedented competition, leading to a set goal of growing daily active users by 5% before 2026.
Looking ahead, OpenAI plans to introduce an "adult mode" in early 2026, allowing for discussions with ChatGPT on mature topics, a move that aligns with its strategy to enhance user engagement while managing its complex user interactions.
