OpenAI's new GPT-5.4 clobbers humans on pro-level work in tests - by 83% (www.zdnet.com)

OpenAI has officially launched GPT-5.4, a significant upgrade that reportedly matches or surpasses human professionals in complex tasks 83% of the time. This model, branded as "GPT-5.4 Thinking," integrates advanced reasoning with the frontier coding capabilities of its predecessor, GPT-5.3-Codex. Improvements include an 18% reduction in error rates and a 33% decrease in false claims compared to GPT-5.2, making it a game-changer for productivity across various industries, from finance to healthcare. The implications of GPT-5.4's performance are profound: it raises questions about the future of skilled jobs as AI demonstrates the potential to not just assist but effectively replace human workers in high-value occupations. OpenAI's tests, which evaluated performance in real-world tasks, have shown that GPT-5.4 not only excels in coding but also enhances its ability to utilize external tools and process complex visual information. The rapid advancement of such models may lead to increased reliance on AI in professional settings, sparking discussions about the balance between augmentation and automation.