Date:
February 28, 2025
OpenAI launches GPT-4.5, its most powerful AI yet, promising better accuracy, emotional intelligence, and smarter interactions—but at a high cost.
OpenAI has officially rolled out GPT-4.5 (code name: Orion), the latest iteration of its AI language model. It is the last version of OpenAI with a “non-chain of thought model.” Yet, it’s bigger, smarter, and—according to CEO Sam Altman—one step closer to feeling like a real conversation.
This is the first time I've been surprised by actually good advice from an AI,
-Altman said during the launch
But there's a catch. The model is massive, and running it isn’t cheap. OpenAI is dealing with GPU shortages, which means only ChatGPT Pro users get access for now, with a broader rollout dependent on hardware availability. However, it seems like GPT 4.5 is not too far along for other plan holders as well. In a post, Sam Altman said:
“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.”
What Makes GPT-4.5 Different?
Compared to its predecessor, GPT-4o, this new model boasts better accuracy, stronger emotional intelligence, and a more nuanced understanding of social cues (unsupervised learning). In simpler terms - it’s designed to "get" you better—whether you’re looking for writing help, coding assistance, or just a solid piece of advice.
One of the biggest improvements comes in the form of reduced AI hallucinations. OpenAI reports that GPT-4.5's tendency to generate incorrect or misleading information has dropped to 37%, a significant improvement over GPT-4o's nearly 60% rate.
But let’s be clear: this isn’t a model built for high-level reasoning just yet. While it’s a leap forward in conversational intelligence, it’s not a dedicated problem-solving AI. "We’re still in the early stages of what AI can achieve," said Mark Chen, OpenAI’s chief research officer.
The Bigger Picture
GPT-4.5 arrives amid a whirlwind of AI advancements from major tech players, all competing to push the boundaries of artificial intelligence. With OpenAI prioritizing scalability, its focus remains on making AI more natural, responsive, and useful in everyday interactions.
For now, early testers seem impressed, and while Altman acknowledges the model’s high costs and infrastructure demands, OpenAI is betting big on its potential.
So, what’s next? More GPUs, wider access, and—if history is any indicator—a GPT-5 on the horizon before we know it.
