OpenAI Launches GPT-4.5, Its Most Advanced AI Model Yet

By siliconindia  |  Friday, 28 February 2025, 09:39 IST

OpenAI Launches GPT-4.5, Its Most Advanced AI Model Yet
OpenAI has launched GPT-4.5, its biggest and most powerful language model to date, in a global research preview. OpenAI says that GPT-4.5 showcases better pattern recognition, deeper links, and more innovative insights than earlier models while still retaining a more sophisticated and emotionally rich conversation style.
In the live unveiling, OpenAI emphasized the ways in which GPT-4.5 feels more natural and intuitive to communicate with. "Its larger knowledge base, enhanced capacity for adhering to user intent, and higher emotional intelligence make it better suited to such tasks as writing improvement, code support, and problem-solving", the firm reported. The model is also said to be less susceptible to hallucinations, which are a hallmark problem of large language models.
Raphael Gontijo Lopes, a researcher from OpenAI, pointed to the development of the model throughout the livestream: "We trained GPT-4.5 to be a more collaborative partner, making interactions warmer and more intuitive. Human evaluators rated GPT-4.5 higher than GPT-4o in nearly every category". OpenAI explained that while it has advanced, GPT-4.5 is not considered a 'frontier model' but increases the computational efficiency of GPT-4 by more than ten times.
Following the launch, OpenAI CEO Sam Altman took to X (formerly Twitter) to share both excitement and concerns. He praised the model's capabilities, describing interactions as akin to speaking with a thoughtful individual. However, Altman revealed the challenges of scaling access to GPT-4.5, citing a significant GPU shortage.
"We've been growing very quickly and are GPU-out," Altman wrote. "We'll add tens of thousands of GPUs next week and roll it out to the Plus tier then. This isn't how we'd like to run, but anticipating growth spikes is difficult".
GPT-4.5 incorporates real-time search functionality, file and image upload functionality, and canvas functionality for writing and coding assignments. The model still does not have multimodal features such as Voice Mode, video, and screen share. OpenAI will try to fill these holes in upcoming releases and build a more fluid AI experience for users.
With its greater capabilities and human-like responsiveness, GPT-4.5 creates a new standard for AI-facilitated communication and productivity, even with the logistical challenges of scaling so powerful a model. As OpenAI continues to scale its infrastructure, wider availability of GPT-4.5 holds the promise to transform the way users engage with AI across a range of domains.

ON THE DECK