OpenAI GPT-4.5: ‘Most Knowledgeable’ But Not Frontier Model Like o1, o3-mini; Sam Altman's ‘Good-Bad’ News

OpenAI has released its largest and most recent AI language model: GPT-4.5. Despite claiming that GPT-4.5 is its “most knowledgeable model yet,” OpenAI has cautioned that it is not a frontier model and may not outperform o1 or o3-mini.

ChatGPT Pro users will be able to begin using GPT-4.5 as a research preview. OpenAI plans to release GPT-4.5 to Plus and Team users the following week.

GPT-4.5 improves the AI’s capacity to comprehend subtleties, have more contextual capabilities, and have more natural conversations, whereas earlier generations such as GPT-4o concentrated on speed and multimodal skills.

According to OpenAI, GPT-4.5 will have better knowledge of the world and what’s happening around it, its writing skills are enhanced, and it possesses a “refined personality over previous models,” as the company put it. GPT 4.5 will feel more “natural” to converse with, according to OpenAI, which also noted that the model is stronger at identifying patterns, forming connections, offering creative insights, which makes it perfect for “solving practical problems.”

Thanks to its “emotional intelligence” and improved steerability, it can better grasp user intent, decipher subtle clues, and sustain interesting conversations that feel intelligent and personalised.

As per OpenAI, GPT-4.5 hallucinates significantly less than GPT-4o and marginally less than o1. During OpenAI’s webcast, researcher Raphael Gontijo Lopes stated, “We aligned GPT-4.5 to be a better collaborator, making conversations feel warmer, more intuitive, and emotionally nuanced,” “We asked human testers to compare it to GPT-4o in order to gauge this, and GPT-4.5 performed better in almost every category.”

OpenAI, however, pointed out that it won’t add enough new features for GPT-4.5 to qualify as a frontier model. “GPT-4.5 is not a frontier model, but it is OpenAI’s largest LLM, improving on GPT-4’s computational efficiency by more than 10x,” OpenAI stated in a document that was leaked before the model was announced. 

“It does not introduce 7 net-new frontier capabilities compared to previous reasoning releases, and its performance is below that of o1, o3-mini, and deep research on most preparedness evaluations,” OpenAI had said. However, the company has since reportedly removed this mention from an updated version of the document.

OpenAI CEO Sam Altman admitted in a post on X that GPT-4.5 is a “giant, expensive model” but “won't crush benchmarks.”

He wrote: “Good news: it is the first model that feels like talking to a thoughtful person to me. i have had several moments where i've sat back in my chair and been astonished at getting actually good advice from an AI.”

“Bad news: it is a giant, expensive model. we really wanted to launch it to plus and pro at the same time, but we've been growing a lot and are out of GPUs. we will add tens of thousands of GPUs next week and roll it out to the plus tier then,” he added.