OpenAI just introduced GPT-5.5internally nicknamed “Spud,” just six weeks after releasing GPT-5.4. The pace is beginning to be difficult to follow: the company updates its models so quickly that it is almost dizzying to try to keep track. The new model excels at writing, code debugging, online research, data analysis, document creation, software management, and chaining complex tasks without constant supervision.
But the most interesting thing about GPT-5.5 is not so much the list of capabilities as the way it operates. It can manage multi-step workflows more autonomously, with fewer instructions from the user, and maintains the same response speed for token that GPT-5.4which is non-trivial when you are increasing the capacity of the model.
More autonomy, less micromanagement
Greg Brockman, president of OpenAI, summed it up to reporters: “What's special about this model is how much more it can do with less guidance. “He can look at a vague problem and figure out for himself what needs to be done.” The idea is that it is no longer necessary to go step by step: you give him a task without much structure and he plans, uses tools, check your work and move forward alone.
Early access teams reported saving up to 10 hours of work per week, although these figures come from OpenAI itself and should be read with that context in mind. More significant, perhaps, is the testimony of the Bank of New York, which has been testing it for weeks and highlighted its resistance to hallucinations, something critical for a highly regulated financial institution.
Availability and what remains pending
GPT-5.5 is now available for users of the Plus, Pro, Business and Enterprise plans, both in ChatGPT as in Codex. API access is coming “very soon,” although they are still finalizing certain additional security safeguards.
In that sense, the model went through extensive third-party testing and red team exercises focused on cyber and biological risks, especially after the launch of Mythos, Anthropic's model, which raised some alarms about the offensive potential of these systems. Brockman also mentioned that GPT-5.5 brings OpenAI closer to what the company calls its «superapp»although that concept remains more of a declared intention than something concrete.
With more than 900 million weekly active users and 50 million subscribersOpenAI has a lot at stake if the narrative that it is losing ground to Anthropic ends up taking hold. GPT-5.5 is, in part, a response to that pressure.
The question that is beginning to float in the air is whether this pace of releases is sustainable from the attention side: Will each new model continue to matter, or will there come a point where users simply stop looking?
