GPT-4.5 officially released: OpenAI's largest and strongest chat model is the first to be opened to ChatGPT Pro users

On February 28, OpenAI announced through an official blog post that its latest development GPT-4.5 AI model has been officially unveiled. This model is hailed by OpenAI as its largest and best-performing chat model and will be the first to open the experience to ChatGPT Pro users as a research preview.

GPT-4.5 has significantly expanded on pre-training and post-training scales, and this change directly enhances its unsupervised learning capabilities. It is reported that the model has made breakthrough progress in understanding patterns, building connections, and generating creativity. Early test feedback showed that the interactive experience with GPT-4.5 was more natural and smooth, its knowledge base covers a wider range, and a deeper understanding of user intentions. In addition, GPT-4.5's performance in "emotional intelligence" is also quite impressive, which helps improve writing and programming skills and ability to solve practical problems, while effectively reducing the occurrence of "illusion" phenomena and improving overall reliability.

GPT-4.5 has further improved its ability to non-supervised learning by expanding the scale of computing and data, optimizing architecture, and introducing innovative technologies. Its broader knowledge reserves and deep understanding of the world provide users with a more accurate and reliable interactive experience. At the same time, the superb skills shown by GPT-4.5 in understanding user intentions and its effectiveness in reducing "illusion" phenomena have brought this model to a new level in reliability.

It is worth mentioning that GPT-4.5 adopts new scalable technology, which can use the data of smaller models to train larger and more powerful models. The introduction of this technology has improved GPT-4.5 in terms of controllability, understanding of nuances, and natural dialogue capabilities. In the fields of writing and design, GPT-4.5 is also outstanding, showing strong creativity and aesthetic intuition.

GPT-4.5 has a wide range of application scenarios, including but not limited to areas such as writing assistance, communication, learning, guidance and brainstorming that require higher emotional intelligence and creativity. In addition, the model also demonstrates strong capabilities in agent planning and execution, supporting multi-step coding workflows and automated processing of complex tasks.

Although GPT-4.5 is a huge and compute-intensive model that costs more than GPT-4, OpenAI said they are evaluating whether they will provide GPT-4.5 services in the API for a long time. It is reported that GPT-4.5 will be launched first for Pro users, and will then be promoted to Plus and Team users next week, and then gradually expanded to Enterprise and Edu users. It is worth noting that users are now trying to use GPT-4.5 on Microsoft's Azure AI Foundry platform.