OpenAI introduces the GPT-4.5 AI model, featuring enhanced understanding and emotional depth.

      OpenAI has recently unveiled its newest AI model, named GPT-4.5, which the company asserts is its largest and most advanced model to date. This model does not possess reasoning abilities, a feature exclusive to the O-series models. However, GPT-4.5 is described as a more natural communicator with an enhanced emotional understanding and better problem-solving skills.

      In terms of key features, it has access to the most up-to-date information available online, allows for the uploading of files and multimedia, and supports a canvas platform for programming tasks. Nonetheless, it currently does not support voice recognition, video analysis, or screen-sharing functionalities.

      At present, this model is in the research preview stage, meaning it is not yet broadly available to ChatGPT Plus subscribers. It appears that infrastructure limitations are preventing the wider release of GPT-4.5 to the public.

      OpenAI CEO Sam Altman explained, “It's a large, costly model. We really aimed to launch it for both Plus and Pro at the same time, but due to our significant growth, we currently lack the necessary GPUs. We will be adding tens of thousands of GPUs next week and will then roll it out to the Plus tier."

      In contrast to models that provide answers through a thought process, GPT-4.5 utilizes a method known as unsupervised learning, which focuses on expanding the training data and computing requirements.

      Sam Altman shared the excitement about the model, stating, “Good news: it is the first model that feels like conversing with a thoughtful person to me. There have been multiple times I've leaned back in my chair, amazed at receiving genuinely good advice from an AI.”

      The advantages of this model include a reduction in hallucination, which is a significant issue for many generative AI chatbots, a larger knowledge base, and an enhanced understanding. According to benchmarks released by the company, GPT-4.5 demonstrates much greater result accuracy compared to the GPT-4o, o3-mini, and o1 AI models, with a hallucination rate less than half that of the o3-mini and lower than its other counterparts.

      OpenAI noted, “For GPT-4.5, we developed new, scalable methods that allow for the training of larger and more powerful models using data from smaller models.” This model is better equipped to handle creative queries, professional inquiries, and nuanced discussions in a more natural manner than any previous model launched by OpenAI.

      Overall, the AI powerhouse claims GPT-4.5 excels in creativity, aesthetic intelligence, writing, and design-related tasks. Access to GPT-4.5 is now being rolled out to ChatGPT Pro subscribers on various platforms including web, mobile, and desktop. Subscribers of ChatGPT Plus and Team tiers will receive access next week, followed by Enterprise and Education bundles.