On December 12, 2025 (Beijing Time), as it marked its 10th anniversary, OpenAI unveiled its latest GPT - 5.2 series models. This series stands as the first significant update following the "Red Alert" incident. It comprises three versions: Instant, Thinking, and Pro, which represent the most powerful models so far for professional knowledge - based tasks.

GPT - 5.2 demonstrates a substantial leap over its predecessors in various aspects. It shows remarkable improvements in creating spreadsheets, presentations, image perception, code writing, and long - context comprehension. What's particularly noteworthy is that in the Thinking version, a staggering 70.9% of the tasks achieved performance levels equal to or superior to those of top industry experts in the GDPval test, which encompasses 44 different professions.

Moreover, GPT - 5.2 shines brightly in the financial AI sector. Compared to its previous iteration, it has slashed error rates by 38% and has set new benchmarks across multiple areas. These include programming, mathematics, scientific reasoning, visual understanding, and long - text reasoning.

Currently, the model is being gradually rolled out to all ChatGPT paid subscribers and API users. This move further intensifies OpenAI's competitive edge in the ever - evolving AI landscape.