OpenAI Unveils GPT-5.2 Amid AI Race with Google
278OpenAI has launched a new version of its AI model, GPT-5.2, claiming it delivers major improvements in handling real-world business tasks. The update comes less than a year after GPT-5.1 was released in November. The company says GPT-5.2 performs at an “expert level” across a wide range of benchmarks, making it a significant step forward in AI capabilities.
Boosts in Business and Problem-Solving Skills
OpenAI highlights that GPT-5.2 performs much better than its predecessor, especially in tasks that require reasoning, coding, and understanding complex contexts. Using its GDPval benchmark, which tests the model’s ability to complete 44 different business tasks to a human expert standard, GPT-5.2 matched or surpassed humans in over 70% of tests. In comparison, GPT-5.1 only reached about 39%.
This leap in performance means GPT-5.2 can do things like fully format spreadsheets, create presentations, and write code more effectively. The company emphasizes that the new model is better at understanding long conversations, perceiving images, and using tools to handle complex projects. These enhancements are designed to unlock more economic value for users in various professional settings.
Improvements Across Benchmarks and Practical Use
Besides GDPval, GPT-5.2 has shown gains on other important benchmarks, such as ARC-AGI-1 and ARC-AGI-2, which test general problem-solving skills. It also outperforms previous versions on SWE-Bench Pro and SWE-Bench Verified, which focus on real-world software development tasks. This means the model is now more reliable at debugging code, implementing new features, refactoring large projects, and shipping fixes with less manual effort.
OpenAI says these improvements translate into more dependable tools for professionals, reducing the need for extensive manual work. The company also points out that GPT-5.2 can better understand and process long documents and multiple-step projects, making it a more versatile assistant for business and technical tasks.
The new model is rolling out to ChatGPT users on paid plans, with no change in subscription prices. API access to GPT-5.2 costs $1.75 per million input tokens and $14 per million output tokens, with discounts available for cached inputs. Although more expensive than GPT-5.1, OpenAI claims GPT-5.2’s increased efficiency actually lowers the overall cost to achieve a given quality level, thanks to better token utilization.
Context in the AI Competition and Future Outlook
The release of GPT-5.2 so soon after GPT-5.1 signals a rapid pace of development for OpenAI. Around December, CEO Sam Altman issued a “code red” memo warning that if GPT-5 wasn’t developed quickly, OpenAI risked falling behind Google’s advanced Gemini 3 model. Since then, the company has claimed that Gemini’s progress has been less dramatic than initially feared, and that the urgent tone may ease by January.
However, the company’s public announcement did not include a direct comparison between GPT-5.2 and Gemini 3. Instead, a separate briefing offered only limited insights. Industry analysts have questioned the benchmarks used by OpenAI, noting that performance metrics like GDPval may not fully capture the models’ real-world abilities. As the AI race heats up, many are watching closely to see how GPT-5.2 stacks up against emerging competitors.
Overall, GPT-5.2 marks a major step for OpenAI, emphasizing improved business skills and problem-solving. Whether it will maintain its edge or face new challenges from rivals remains to be seen, but the rapid development shows how fierce the competition in AI has become.
What do you think?
It is nice to know your opinion. Leave a comment.