OpenAI unveils GPT-5.4, its newest AI model, which the company highlights for its enhanced capabilities in reasoning, coding, and professional applications such as spreadsheets, documents, and presentations. Notably, GPT-5.4 marks OpenAI's first foray into native computer use, empowering it to operate a device and execute tasks across various applications autonomously.
This new model represents a significant stride towards the "agentic future" envisioned by AI developers, where interconnected AI agents work in the background to accomplish intricate tasks both online and within software environments. This builds upon previous innovations, such as OpenAI's introduction of ChatGPT Agent last year, among other agentic tools designed to take control of a user's computer for specific functions, like sourcing and purchasing meal ingredients.
OpenAI is integrating GPT-5.4 into its API and the AI-powered coding platform, Codex, while simultaneously deploying its specialized reasoning model, GPT-5.4 Thinking, within ChatGPT. According to OpenAI, GPT-5.4 possesses the ability to generate code for computer operations and execute keyboard and mouse commands based on visual input from screenshots. Furthermore, the model demonstrates enhanced performance in web browsing and a more precise and efficient capability to leverage various tools and APIs for task completion.
The model also exhibits superior proficiency in addressing inquiries that necessitate information synthesis from diverse sources. OpenAI states that it “can more persistently search across multiple rounds to identify the most relevant sources, particularly for ‘needle-in-a-haystack’ questions, and synthesize them into a clear, well-reasoned answer.” OpenAI asserts that GPT-5.4 stands as its “most factual model yet,” with individual claims demonstrating a 33 percent reduced likelihood of being false when compared to its predecessor, GPT-5.2.
Within ChatGPT, GPT-5.4 Thinking will furnish users with an outline of its processing for more intricate queries, alongside the ability to modify or refine their requests even as the model generates its response. OpenAI elaborates, stating, “This makes it easier to guide the model toward the exact outcome you want without starting over or requiring multiple additional turns.” This functionality is currently accessible via the ChatGPT web application and Android devices, with an anticipated release for the iOS app in the near future.
The rollout of GPT-5.4 is currently underway across ChatGPT, Codex, and the API. Specifically, the GPT-5.4 Thinking model will be made available to Plus, Team, and Pro subscribers. Additionally, a GPT-5.4 Pro model, optimized for “maximum performance on complex tasks,” is being deployed within the API and extended to ChatGPT Enterprise and Edu users.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.