OpenAI launched 'GPT-5.4' on the 6th, introducing a groundbreaking feature that allows artificial intelligence to manipulate computers and software, enabling automation of tasks such as spreadsheet management and presentation creation. This model marks a significant advancement in integrating reasoning, coding, and work automation into a single system, specifically designed for enhancing the efficiency of office workers. By unifying previous separate functionalities of 'GPT-5.2 Thinking' and 'GPT-5.3 Codex', it aims to streamline office tasks considerably.

The new capabilities of GPT-5.4 include notable improvements in spreadsheet modeling and document creation. Internal benchmarks simulated Excel-based data analysis tasks at a level similar to entry-level investment banking analysts, achieving an average score of 87.5%, significantly surpassing the previous model's score of 68.4%. Additionally, the ability to create visually appealing and accurate presentations has also been enhanced, with improvements noted in design completeness, visual variety, image usage, and factual accuracy. With stronger analytical skills and design capabilities, Office tasks can be handled more effectively.

Importantly, GPT-5.4 is the first universal model to include a 'computer-use' capability, allowing the AI to operate within real computer environments and switch between various applications seamlessly. OpenAI emphasizes that this model is the most cost-effective reasoning model to date, significantly reducing the number of tokens needed to solve problems compared to its predecessor, GPT-5.2, making it an essential tool for those looking to improve productivity in complex work environments.