The American artificial intelligence giant, OpenAI, has officially deployed the GPT-5.4 model, introducing a groundbreaking native computer control function. The new architecture allows the AI to independently operate the cursor, keyboard, and applications, marking a milestone towards autonomous agents. The system debuts in Pro and Thinking versions, offering record performance in professional tasks and deep integration with Microsoft Excel and Google Sheets spreadsheets.
AI Computer Control
The GPT-5.4 model can independently operate the mouse cursor and applications in the operating system.
Spreadsheet Integration
New financial plugins allow for the automation of work in Excel and Google Sheets programs.
Pro and Thinking Variants
The system offers two modes of operation: speed for professionals and deep reasoning for complex tasks.
Reduction of Linguistic Mannerisms
Version 5.3 Instant has been modified to make responses more natural and less formulaic.
The American company OpenAI has taken a significant step in the development of artificial intelligence systems by releasing the GPT-5.4 model. The key innovation that distinguishes this solution from its predecessors is the native computer control mode. Thanks to this, ChatGPT ceases to be merely a text generator, transforming into an active autonomous agent that can copy data between windows, manage files, and perform complex operations in a graphical environment. The manufacturer placed particular emphasis on the financial sector by introducing dedicated plugins for Microsoft Excel and Google Sheets programs, which is intended to enable rapid data analysis and reporting automation. Concurrently, the company refreshed the base model to version 5.3 Instant, responding to user complaints about unnatural speech style and a tendency to replicate irritating linguistic mannerisms. Since the revolution initiated by the GPT-3 model in 2020, the development of artificial intelligence has evolved from simple next-word prediction to systems capable of writing code and managing entire digital infrastructure.Internal tests by OpenAI suggest that GPT-5.4 achieves results 83% better than human specialists in tasks requiring expert knowledge. The new offering structure is divided into the Pro variant, intended for daily high-performance work, and the Thinking variant, optimized for multi-step reasoning and real-time error verification. This strategic launch is seen as an attempt to regain market initiative in the face of growing pressure from the competing Claude model by Anthropic. Developers have gained access to a cheaper and faster API, which is expected to accelerate the adoption of the new technology in business. However, the increased autonomy of AI systems raises questions about operational safety. OpenAI assures that implemented security protocols prevent the model from taking unauthorized destructive actions, and every session in computer control mode is fully monitored by the user and can be terminated at any time. The launch of GPT-5.4 coincides with a global debate on legal regulations concerning algorithm transparency and the protection of data processed by intelligent bots. The new model is designed not only to provide better answers but, above all, to genuinely relieve workers of repetitive, tedious digital processes, posing challenges to the traditional office work model.