OpenAI’s new GPT-5.4 model is a big step toward autonomous agents

OpenAI has released GPT-5.4, its newest AI model that integrates advancements in reasoning, coding, and professional document handling. A key feature is its native computer use capabilities, allowing it to operate a computer on a user's behalf across various applications, marking a significant move toward autonomous AI agents. The model, which can write code to control computers and issue keyboard/mouse commands based on screenshots, is rolling out across the API, Codex, and ChatGPT. Inside ChatGPT, the GPT-5.4 Thinking variant offers users an outline of its complex query process, allowing for mid-response adjustments. OpenAI claims GPT-5.4 is its "most factual model yet," with a 33 percent reduction in false claims compared to GPT-5.2, and shows improved ability to synthesize information from multiple sources for better answers.

(Source：The Verge)