OpenAI releases GPT-5.4, AI can now operate a computer

OpenAI has unveiled its latest artificial intelligence model, GPT-5.4, also referred to as GPT-5.4 Thinking, alongside the API and Codex. In an announcement on the company’s official site, OpenAI said the model is designed to handle a wide range of professional tasks more efficiently. ‘Today we are releasing GPT-5.4 (as GPT-5.4 Thinking), the API, and Codex. This is our most capable and efficient model for professional work,’ OpenAI wrote in the announcement. GPT-5.4 is described as combining enhanced reasoning, coding, and a variety of professional tasks involving spreadsheets, documents, and presentations. The AI model also introduces one of OpenAI’s latest feature updates: the ability to operate a computer directly. With this capability, GPT-4.1 can operate a computer on behalf of the user, automatically completing cross‑application tasks across software. GPT-5.4 can also write code to execute commands on the keyboard and mouse based on information drawn from screenshots. The model is claimed to be more proficient at using a web browser, and can call various tools and APIs more accurately to complete tasks. OpenAI also introduced GPT-5.4 into the API and Codex-based AI coding tools, as previously mentioned. The ability to give and execute commands marks a significant improvement for the AI model. GPT-5.4 can now utilise tools and API access more accurately and efficiently when completing a range of tasks. In other words, AI is no longer just handling simple tasks, but can also perform more complex work automatically in the background of a computer across various software applications. GPT-5.4 is also claimed to provide better answers by incorporating more information-gathering processes from multiple sources on the internet.