The American company OpenAI announced this Thursday GPT-5.5, its new artificial intelligence model designed to perform tasks on computers and automate workflows like a human.
In , the company reveals that the model excels at writing and debugging code, online research, data analysis, creating documents and spreadsheets, managing software, and switching between tools until a task is completed.
Despite being more complex, the new model operates with the same efficiency as the previous one, GPT-5.4. The improvement lies in the number of tokens needed to perform the same tasks, optimizing time.
“The model is faster and more accurate, and consumes less computational resources than 5.4”, says the president of OpenAI, Greg Brockman.
Introducing GPT-5.5
A new class of intelligence for real work and powering agents, built to understand complex goals, use tools, check its work, and carry more tasks through to completion. It marks a new way of getting computer work done.
Now available in ChatGPT and Codex.
— OpenAI (@OpenAI)
OpenAI states that the model includes its “strongest protective measures to date” and which has been tested by almost 200 partnersincluding companies and researchers working in software, finance, communications, pharmacy and scientific research.
According to , users can therefore assign GPT-5.5 a complex, multi-step task. The model will then plan, use tools, review its work, deal with ambiguous situations and continue working.
“Improvements are notable in coding, computer use, knowledge work, and early scientific research — areas where progress depends on contextual reasoning and taking action over time,” the company explains.
Its availability has already started and will be available to users of different plans, including Plus, Pro, Business and Enterprise of ChatGPT and Codex. In turn, the GPT-5.5 Pro is being rolled out to Pro, Business and Enterprise users on ChatGPT.
The model will also be accessible through a application programming interface (API), a software that allows developers and companies to connect the model directly to their applications and services. The company did not specify when or where it will be available.
