GPT-5.4 is here: AI can now beat humans at tasks
Technology
OpenAI just released GPT-5.4, its newest AI model, and it's a big leap—this version can match or even beat humans at 83% of tasks.
It's more accurate, makes fewer mistakes than the last version (GPT-5.2), and is rolling out on ChatGPT and in Codex, with API access forthcoming.
The model focuses on helping people with complex projects
GPT-5.4 shines in areas like finance, healthcare, and tech after passing tough industry tests.
With better workflow tools, sharper computer vision for images, and smoother software automation via Codex integration, this update is built to help people handle complex projects faster and smarter—especially if you're into coding or tech innovation.