OpenAI GPT-4.5 model introduced

OpenAI has introduced GPT-4.5, their latest and most advanced chat model. Among its highlights are its broader knowledge base, emotional intelligence (EQ), and more.

This model has an improved ability to follow a user’s intent, built on advancements in pre-training and post-training. Specifically, it was trained on Microsoft Azure AI supercomputers and leverages architectural innovations, data, and power.

This allows it to offer deeper world knowledge, making it reliable in a handful of topics. In other words, it’s more useful in tasks like writing, programming, and problem-solving.

OpenAI noted two approaches to advance AI capabilities in unsupervised learning and scaling reasoning. The former betters the model’s intuition and understanding of the world. Meanwhile, reasoning works by developing problem-solving skills step-by-step.

The development of this model was furthered by combining these approaches. With that, OpenAI was able to balance creativity and accuracy. Suffice it to say, GPT-4.5 can certainly excel in tasks that require context, collaboration, and empathy.

The model supports searches, file and image uploads, and canvas features. However, users should take note that it doesn’t support multimodal capabilities yet. This includes voice mode, video, and screen sharing.

In addition, OpenAI put an emphasis on safety improvements for GPT-4.5. They leveraged traditional supervised fine-tuning and reinforcement learning from human feedback for this. The model was tested under the company’s Preparedness Framework.

GPT-4.5 is now available for ChatGPT Pro users via web, mobile, and desktop. It will be rolling out to Plus, Team, Enterprise, and Edu users in the coming weeks.