OpenAI is rolling out a new artificial intelligence model designed to make ChatGPT better at coding, science and a wide range of work tasks, weeks after Alphabet Inc.’s Google put the startup on defense with the well-received launch of Gemini 3.
The new model, GPT-5.2, is faster and more adept at finding information, writing and translating, the company said Thursday. The software, available in three tiers, is also intended to be better at mimicking the human process of reasoning to handle more complicated, lengthier tasks in fields such as math and programming.
Once the undisputed leader in AI development, OpenAI is facing renewed competition from Google and Anthropic, both of which have introduced new models in recent weeks. Google’s Gemini 3, in particular, has earned praise for its reasoning and coding capabilities, and quickly rose to the top of closely watched AI leaderboards on LMArena and Humanity’s Last Exam.
Earlier this month, OpenAI Chief Executive Officer Sam Altman declared a “code red” in a company memo and said internal resources should be redirected to speed up improvements to ChatGPT. The directive echoed Google’s widely reported move to issue a code red shortly after ChatGPT launched three years ago.
With GPT-5.2, OpenAI is competing to draw in more business customers and boost revenue as it commits to spending more than $1 trillion on infrastructure to support AI development. To that end, the company said the model is designed to “unlock even more economic value for people,” including by doing better at creating spreadsheets and presentations, as well as writing code. GPT-5.2 will begin rolling out to paid users on Thursday.
In August, OpenAI released a long-awaited new model called GPT-5 that was met with mixed reviews, with some expressing doubts that it represented a revolutionary change from prior offerings. In November, OpenAI pushed out a newer version of the model, 5.1, but that update was soon overshadowed by the more powerful Gemini 3.
In an interview with CNBC on Thursday, Altman said, “Gemini 3 has had less of an impact on our metrics than maybe we feared.” Altman also said he expects OpenAI to exit its code red mode by January “in a very strong position.”
OpenAI also said it has begun rolling out software that can predict users’ ages — a move it’s making to help control how ChatGPT responds to those under 18. In a briefing with reporters Thursday, OpenAI CEO of applications Fidji Simo said the company wants to introduce that capability before launching an “adult mode” that Altman previously said might allow uses such as “erotica for verified adults.” Simo said “adult mode” will launch in the first quarter of next year.
