OpenAI officially launches GPT-5.5: A new leap forward in thinking and performance.
OpenAI has officially launched its latest version, GPT-5.5, marking a significant leap forward in performance and processing capabilities compared to its predecessor. According to the announcement, GPT-5.5 boasts significant improvements in reasoning, programming, and scientific research, while also operating faster, more efficiently, and more securely.
Developed on the foundation of GPT-5.4, the new version adds many important upgrades, allowing for efficient handling of complex multi-step tasks with superior planning, tool usage, and self-correction capabilities. The model particularly demonstrates strengths in programming and debugging, achieving an accuracy rate of 82.7% in the Terminal-Bench 2.0 test.
Despite being more intelligent, GPT-5.5 maintains similar speed to its predecessor but uses fewer tokens per task, significantly reducing costs for businesses and developers. This is considered one of the key improvements in the context of the growing need to optimize AI operational costs.
In addition, OpenAI has introduced several variants of GPT-5.5 to meet different needs. The GPT-5.5 Thinking version is optimized for solving complex problems with faster speed and more intelligent reasoning capabilities. Meanwhile, the GPT-5.5 Pro aims for higher output quality and lower latency, suitable for high-performance workflows. These versions give businesses and professionals flexibility in choosing according to their purpose, from research to large-scale automation.
Regarding security, OpenAI states that GPT-5.5 is equipped with the strongest protections to date. The model underwent rigorous testing to assess risks in areas such as cybersecurity and biosecurity, with the participation of nearly 200 early testing partners. The assessment results ranked the model as having a “high” risk but not yet reaching the “severe” threshold, meaning it could exacerbate existing risks but not create entirely new ones.
GPT-5.5 will be deployed to users of Plus, Pro, Business, and Enterprise plans on ChatGPT and Codex, while the GPT-5.5 Pro version will be for Pro, Business, and Enterprise plans. OpenAI stated that API deployment requires additional layers of protection, and the company is working closely with partners to ensure security requirements are met before large-scale expansion in the near future.
https://baocantho.com.vn