OpenAI Introduces GPT-5.2, Strongest AI Model For Science And Mathematics

GPT-5.2 is officially released (photo: dock. OpenAI)

JAKARTA OpenAI has finally launched its newest Artificial Intelligence (AI) model, namely GPT-5.2 Pro and GPT-5.2 Thinking. Both are referred to as the strongest AI models for mathematical and science work.

According to OpenAI, the ability of GPT-5.2 in terms of mathematical reasoning proves that this AI model is reliable in technical and scientific needs. This ability also allows GPT-5.2 to carry out multi-step logic.

The quantity is also more consistent so that this model is able to avoid fine errors in real analysis, such as in simulations or statistics. This ability directly affects crucial scientific workflows, including encoding and data analysis.

FrontierMath, a mathematical benchmark in the form of expert level evaluations, has proven the ability of GPT-5.2. GPT-5.2 Thinking managed to solve 40.3 percent of the problem, highlighting its deep mathematical reasoning capabilities.

In the postgraduate-level question and answer benchmark, GPQA Diamond, the GPT-5.2 Pro reached an impressive score of 93.2 percent. GPT-5.2 Thinking followed with a score of 92.4 percent on the same benchmark.

Although this model is powerful, OpenAI emphasizes the importance of the role and responsibility of human researchers. Models such as GPT-5.2 can serve as a tool to support reasoning and accelerate early-stage exploration if used properly.

"If used carefully, such a system can help simplify important aspects of theoretical work without replacing the central role of human judgment in scientific investigations," OpenAI wrote in its release, quoted on Friday, December 12.