OpenAI launches GPT-5.5, its smartest model that moves towards “more intuitive computing”

OpenAI has launched its new artificial intelligence (AI) model GPT-5.5, the most smart and intuitive with enhanced capabilities for carry out tasks autonomously and understand user intentions more precisely and with greater speed, moving towards “more intuitive computing” for its “super application.”

Just a month after the launch of its latest GPT-5.4 update, the technology company led by Sam Altman has unveiled its new iteration, which is faster and more accurate, while requiring fewer resources.

OpenAI has highlighted the progress made with GPT-5.5which understand user intentions faster and, furthermore, it can do “much of the work” yourselfmoving toward the company’s “super app” goal.

This was detailed by the co-founder and president of OpenAI, Greg Brockman, who stated that the new model shows a advance “towards more intuitive computing with greater capacity for interaction with users”, as reported by TechCrunch.

As OpenAI specified in a statement on its blog, these improvements mean that, instead of having to manage each step, users can assign GPT-5.5 a complex task that requires many steps and “trust” that you will plan the work autonomouslyyou will use the necessary tools, verify your work and continue with the process until the task is completed.

This is due, in part, to the fact that it can now better manage agent coding, he computer use and the scientific research early, areas that depend on contextual reasoning and prolonged decision making, as the technology company has explained.

In addition to all this, the company has assured that GPT-5.5 excels at writing and debugging codethe online researchhe data analysisthe creation of documents and spreadsheets, the management of ‘software’ and the alternation between tools.

Another factor that OpenAI has highlighted is that, despite this leap in intelligence and the ability to execute highly difficult tasks, GPT-5.5 is not sloweras is usually the case in larger and more capable models, but “equals the latency per token” of its previous version GPT-5.4. Added to this is that it uses fewer tokens to complete the same Codex tasks.

In numbers, GPT-5.5 reaches a high performance in tests like GDPval, that evaluates the ability of agents to produce well-specified knowledge work in 44 occupations, with a score of 84.9 percent. Likewise, in OSWorld-Verifiedwhich measures whether you can use real computing environments autonomously, achieves a score of 78.8 percent.

GPT-5.5 THINKING Y PRO EN CHATGPT

Inside the wizard ChatGPTthis model has the version GPT-5.5 Thinking that offers “faster help for complex problems. That is, it offers smarter and more concise responses, designed to help users advance faster and more efficiently in complex tasks.

Specifically, it excels in professional tasks of programming, research, synthesis and analysis of information, where a large number of documents are usually required.

Likewise, for the ChatGPT assistant there is also the version GPT-5.5 Prowhich guarantees a “significant improvement” in both difficulty and quality of work. Compared to the GPT-5.4 Pro version, the company has clarified that the answers are more complete, well structured and more precise, standing out in the business, legal, educational and data science fields.

On the other hand, OpenAI has indicated that it has released GPT-5.5 with its set of “most robust security measures to date“adding specific tests for advanced capabilities of cybersecurity and biology. Thus, it is designed to reduce misuse of its capabilities and, at the same time, “preserve access for beneficial purposes.”

The new GPT-5.5 model is available for users subscribed to the Plus, Pro, Business and Enterprise plans on ChatGPT and Codex. Additionally, GPT-5.5 Pro version is rolling out to Pro, Business, and Enterprise users on ChatGPT. They will soon be available for the API.