On Thursday, OpenAI unveiled its most recent AI models, dismissing concerns about how it will profit from significant investments in its technological competition with Google.
GPT-5.2 Pro and GPT-5.2 Thinking were hailed by the San Francisco-based AI giant as its greatest models to date for managing math or science tasks.
In a blog post, OpenAI stated, “Reliability in scientific and technical work is based on strong mathematical reasoning.”
“These skills are also directly related to the advancement of general intelligence.”
In the computer industry, artificial general intelligence has emerged as a kind of holy grail, a point at which robots can think just as well as or even more like humans.
The announcement follows OpenAI CEO Sam Altman’s call for his team to work hard to stay ahead of Google, the search engine behemoth that has been steadily advancing AI.
OpenAI has been investing tens of billions of dollars in computing infrastructure without making a profit, whereas Google can use its enormous online ad revenue to invest in AI.
In an interview with CNBC on Thursday, Altman stated, “We are confident we can continue to drive the revenue growth to meet” the investments in processing capacity.
“Obviously, we can’t drive the revenue growth without the infrastructure investments, but we see way more reasons to be optimistic than reasons to be pessimistic.”
During a briefing on the new models, Fidji Simo, chief of applications at OpenAI, told reporters that she anticipates a ChatGPT “adult mode” to launch early next year. She added that the business wants to enhance user age detection before releasing it.
Earlier this year, Altman revealed plans to loosen limitations so that adult customers might use ChatGPT to have sensual discussions.
Families have filed numerous lawsuits against OpenAI, alleging that the business permitted teenagers to engage in risky conversations with its AI chatbots, which in some cases resulted in suicide.
Simo denied that it has accelerated the distribution of new GPT models, but he did acknowledge that OpenAI had sent a “red alert” regarding Google racing ahead.
After being taken aback by ChatGPT’s introduction three years ago and ridiculed for its early mistakes in its pursuit of OpenAI, Google this month unveiled its most recent Gemini AI model, concluding a spectacular turnaround.
The race for AI innovation accelerates, which following Google’s Gemini 3, GPT-5.2 after it was released. OpenAI’s earlier models were outperformed by Gemini 3 on a number of tests. Sam Altman, the CEO of OpenAI, decided to speed up work as a result.
The intense competitiveness is highlighted by the benchmark comparisons;
- Gemini 3 scored 76.2% on the SWE-bench for coding, whereas GPT-5.2 earned 80%.
- In the AIME 2025 maths competition, GPT-5.2 scored 100% without the use of any tools.
The two AI businesses are still in rivalry though, as seen by the quick release cycle. Enterprise applications are the main focus, where dependability and efficiency are crucial.
GPT-5.2 is now available in paid ChatGPT membership levels thanks to OpenAI. Through its API, it also made the model accessible to developers.