(Image source from: Ndtv.com)

OpenAI has launched GPT-5.4, its newest artificial intelligence model, which represents a major advancement in creating independent AI agents capable of completing tasks for users. The organization stated that this updated model has built-in abilities for computer usage, enabling it to manage devices and software directly. This signifies that the system can undertake various activities across different programs, like managing spreadsheets, documents, and presentations. As per OpenAI, the model can generate code that allows it to command computers and execute actions like sending keyboard and mouse instructions based on screenshots. It has also demonstrated better performance while using web browsers and engaging with tools and application programming interfaces (APIs). This release signifies a step forward in the industry's progress towards a future where numerous AI-driven agents function behind the scenes to tackle complicated online tasks.

Previously, OpenAI unveiled ChatGPT Agent, which is a tool intended to operate a computer and carry out tasks like searching for details or making online purchases. OpenAI is providing GPT-5.4 through its API and the AI coding aid, Codex. At the same time, a version that focuses on reasoning, called GPT-5.4 Thinking, is being made available in ChatGPT. The organization mentioned that this new model is more effective at addressing intricate questions that need information from various sources. It can search more thoroughly and merge outcomes into clearer answers. OpenAI further asserts that GPT-5.4 is its most accurate model to date, with specific statements being about one-third less likely to be incorrect compared to the previous GPT-5.2 model.